Search is not available for this dataset
article stringlengths 4.36k 149k | summary stringlengths 32 3.35k | section_headings listlengths 1 91 | keywords listlengths 0 141 | year stringclasses 13
values | title stringlengths 20 281 |
|---|---|---|---|---|---|
Denitrifying bacteria accumulate NO2− , NO , and N2O , the amounts depending on transcriptional regulation of core denitrification genes in response to O2-limiting conditions . The genes include nar , nir , nor and nosZ , encoding NO3− - , NO2− - , NO- and N2O reductase , respectively . We previously constructed a dynamic model to simulate growth and respiration in batch cultures of Paracoccus denitrificans . The observed denitrification kinetics were adequately simulated by assuming a stochastic initiation of nir-transcription in each cell with an extremely low probability ( 0 . 5% h-1 ) , leading to product- and substrate-induced transcription of nir and nor , respectively , via NO . Thus , the model predicted cell diversification: after O2 depletion , only a small fraction was able to grow by reducing NO2− . Here we have extended the model to simulate batch cultivation with NO3− , i . e . , NO2− , NO , N2O , and N2 kinetics , measured in a novel experiment including frequent measurements of NO2− . Pa . denitrificans reduced practically all NO3− to NO2− before initiating gas production . The NO2− production is adequately simulated by assuming stochastic nar-transcription , as that for nirS , but with a higher probability ( 0 . 035 h-1 ) and initiating at a higher O2 concentration . Our model assumes that all cells express nosZ , thus predicting that a majority of cells have only N2O-reductase ( A ) , while a minority ( B ) has NO2− - , NO- and N2O-reductase . Population B has a higher cell-specific respiration rate than A because the latter can only use N2O produced by B . Thus , the ratio BA is low immediately after O2 depletion , but increases throughout the anoxic phase because B grows faster than A . As a result , the model predicts initially low but gradually increasing N2O concentration throughout the anoxic phase , as observed . The modelled cell diversification neatly explains the observed denitrification kinetics and transient intermediate accumulations . The result has major implications for understanding the relationship between genotype and phenotype in denitrification research .
To devise robust strategies for mitigating global N2O emissions , a good understanding of its primary source is imperative , i . e . , genetics , physiology , and regulatory biology of denitrifiers . Any knowledge of the environmental controllers of N2O is incomplete without understanding the causal relationships of such controllers at the physiological level [6] . The biogeochemical models developed for understanding the ecosystem controls of denitrification and N2O emissions treat the denitrifying community of soils and sediments as a single homogenous unit with certain characteristic responses to O2 and NO3− concentrations [6 , 7] . Natural denitrifying communities , however , are mixtures of organisms with widely different denitrification regulatory phenotypes [8] . The regulatory response of such mixtures is not necessarily equal to the ‘sum of its components’ because there will be interactions , not the least , via the intermediates NO and NO2− . Hence , it is probably a mission impossible to predict the regulatory responses of complex communities based on their phenotypic composition . Nevertheless , investigations of the regulation in model organisms like Pa . denitrificans provide us with essential concepts , enhancing our ability to understand the regulatory responses of mixed communities and to generate meaningful hypotheses . Thus , future biogeochemical models of N2O and NO emissions are expected to have more explicit simulations of the regulatory networks involved , and a first attempt has recently been published [9] . Dynamic modelling has been used to a limited extent to analyse various denitrification phenotypes; for example , to analyse NO3− and NO2− reduction and gas-kinetic data for individual strains [10] and mixtures of selected phenotypes [11]; to model the consequence of competition for electrons between denitrification reductases [12 , 13]; to investigate the control of O2 on denitrification enzymes and inhibition of cytochrome c oxidase by NO in Agrobacterium tumefaciens [14]; and to examine the effect of copper availability on N2O reduction in Paracoccus denitrificans [15] . In our previous model [16] , we simulated O2 and N2 kinetics from batch incubations of Pa . denitrificans [8 , 17] to test if a postulated cell diversification , driven by stochastic initiation of nirS , could explain the N2 production kinetics in NO2− -supplemented media . The available data also contained NO3− -supplemented treatments but NO3− and NO2− were not monitored , and the experiment provided no information about the N2O kinetics , except that the concentrations were extremely low ( below the detection limit of the thermal conductivity detector used ) . Recently , a neat dataset was generated from batch incubations supplemented with NO3− , with frequent measurements of NO2− and a more sensitive detection of N2O by an electron capture detector [18] . That encouraged us to extend our previous model and simulate the cell diversification during transition from oxic to anoxic conditions , targeting the regulation of Nar and cNor/NosZ ( N2O emissions ) in Pa . denitrificans . Pa . denitrificans is a facultative anaerobe capable of reducing NO3− all the way to N2: NO3−→NarNO2−→NirSNO→cNorN2O→NosZN2 In response to impending anoxic conditions , the organism sustains respiratory metabolism by producing the membrane-bound cytoplasmic nitrate reductase ( Nar ) , cytochrome cd1 nitrite reductase ( NirS ) , cytochrome c dependent nitric oxide reductase ( cNor ) , and nitrous oxide reductase ( NosZ ) . Transcription of the genes encoding these reductases ( narG , nirS , norBC , and nosZ , respectively ) are regulated by the FNR-type proteins FnrP , NarR , and NNR . FnrP contains a 4Fe-4S cluster for sensing O2 , and NNR harbours a NO-sensing haem; NarR , however , is poorly characterised and is most likely a NO2− -sensor [19–21] . All these sensors remain inactive during aerobic growth conditions [19] . Denitrification proteome , once produced in response to an anoxic spell , is likely to linger within the cells under subsequent oxic conditions , ready to be used if anoxia recurs . But the proteome will be diluted by aerobic growth because the transcription of denitrification genes is inactivated under oxic conditions [20] . Hence , a population growing through many generations under fully oxic conditions is expected to undertake de novo synthesis of denitrification enzymes when confronted with anoxia . Batch cultivations of such aerobically raised Pa . denitrificans provided indirect evidence for a novel claim that , in response to anoxia , only a small fraction of the incubated population is able to produce denitrification proteome [8 , 17 , 27 , 28] . Our dynamic modelling of Bergaust et al . ’s [17] NO2− -supplemented incubations corroborated this , suggesting that a probabilistic function ( specific probability = 0 . 005 h-1 ) resulting in the recruitment of 3 . 8–16 . 1% of all cells to denitrification is adequate to explain the measured N2 kinetics [16] . Our model was based on the hypothesis that the entrapment of a large fraction in anoxia is due to a low probability of initiating nirS transcription , which in response to O2 depletion is possibly mediated through a minute pool of intact NNR , crosstalk with other factors ( such as FnrP ) , unspecific reduction of NO2− to NO by Nar , and/or through non-biologically formed traces of NO found in a NO2− -supplemented medium . Regardless of the exact mechanism ( s ) , once nirS transcription is initiated , the positive feedback via NO/NNR ( Fig 1 , see P2 ) would allow the product of a single transcript of nirS to induce a subsequent burst of nirS transcription . The activated positive feedback will also help induce nor and nosZ transcription via NNR , rapidly transforming a cell into a full-fledged denitrifier . We further hypothesised that recruitment to denitrification will only be possible as long as a minimum of O2 is available because , since Pa . denitrificans is non-fermentative , the synthesis of first molecules of NirS will depend on energy from aerobic respiration . The above hypothesis was modelled by segregating the culture into two pools ( subpopulations ) : one for the cells without ( ND− ) and the other with denitrification enzymes ( ND+ ) . Initially , all cells were ND− , growing by consuming O2 . As [O2] fell below a certain threshold , ND− recruited to ND+ with a constant probability ( h-1 ) , assumed to be that of the nirS transcriptional activation , and the recruitment halted as O2 was completely exhausted , assuming lack of energy ( ATP ) for enzyme synthesis . The present model is an extension of that developed in Hassan et al . [16] . Here we have divided the respiring culture into four pools ( Fig 2A ) : All these subpopulations are assumed to scavenge O2 ( if present ) and produce NosZ in response to impending anoxia . The latter because the nosZ genes are readily induced by the O2-sensor FnrP [24] . The Z− pool ( Fig 2A ) contains the inoculum that grows by aerobic respiration . As [O2] falls below a critical threshold [empirically determined , 18] , the cells within Z− are assumed to start synthesising Nar with a certain probability and populate the ZNa pool . The aim here is to investigate whether , like for nirS , the initiation of nar transcription ( by some combined activity of FnrP and NarR ) can also be explained as a probabilistic phenomenon , quickly differentiating a cell into a full-fledge NO3− scavenger through product ( NO2− ) induced transcription via NarR ( Fig 1 , see P1 ) . If so , we were interested to estimate what fraction of the cells is required to adequately simulate the measured data ( NO2− production ) , aiming at scrutinising the general assumption that all cells in batch cultures produce Nar in response to impending anoxia . Next , when [O2] is further depleted to another critical threshold [18] , the Z− and ZNa cells are assumed to initiate nirS transcription with a low per hour probability and , thereby , populate the ZNi and ZNaNi pools , respectively . As explained above for our previous model , NirS + cNor production is assumed to be a ) coordinated because the transcription of both nirS and nor is induced by NO via the NO-sensor NNR ( Fig 1 ) , and b ) stochastic because the initial transcription of nirS ( paving the way for the autocatalytic expression of NirS and substrate-induced nor transcription ) happens in the absence of NO or at too low [NO] to be sensed by NNR . Synthesis of denitrification enzymes requires energy , which all the subpopulations can obtain by respiration only . Hence , the initiation of the autocatalytic expression of nar and nirS ( i . e . , recruitment to ZNa and ZNaNi/ZNi , respectively , Fig 2A ) depends on the availability of the relevant terminal e--acceptor ( s ) above a critical concentration to sustain a minimum of respiration . For Z− , the only relevant e--acceptors are O2 and the traces of N2O produced by ZNi and ZNaNi . The same applies For ZNa , but in addition , this subpopulation can also obtain energy by reducing NO3− , if present . In our previous model [16] , we assumed that recruitment to denitrification was sustained by energy from O2-respiration only; not NO3− because we simulated NO2− -supplemented treatments , and not by N2O because we naively assumed that the pool of this e--acceptor was insignificant ( N2O concentrations were below the detection limit of the system used for those experiments ) . However , the present model assumes that the recruitment from Z− to ZNa and Z− to ZNi is sustained by both O2- and N2O-reduction , and the recruitment from ZNa to ZNaNi is sustained by O2- , N2O- and NO3− -reduction , when above a critical minimum ( vemin− ) . The default value for vemin− was set to an arbitrary low value ( = 0 . 44% of the maximum e--flow rate to O2 ) , and we have investigated the consequences of increasing , decreasing , and setting vemin− = 0 . The expressions of nar and nirS + nor ( recruitments to ZNa and ZNaNi/ZNi , respectively , Fig 2A ) are modelled as instantaneous discrete-events in each cell , thus ignoring the time-lag from the initiation of gene transcription till the cell is fully equipped with the reductase ( s ) in question . That is because the lag observed between the emergence of denitrification gene transcripts and the subsequent gas products suggests that the synthesis of denitrification enzymes takes less than half an hour [17 , 18] , which is negligible for our purposes here . The main purpose of the present modelling is to investigate if a full-fledged model , including all four functional denitrification reductases , could adequately simulate the observed kinetics and stoichiometry of denitrification products [18] . These cultures reduced all available NO3− to NO2− prior to the onset of gas production and accumulated traces of N2O throughout the anoxic phase , as illustrated in S1 Fig In particular , we were interested to investigate the NO2− kinetics , controlled by nar- and nirS transcription , and to test if the peculiar N2O kinetics ( low , but increasing concentrations throughout the anoxic phase ) could be explained by our modelled cell diversification .
The model is constructed in Vensim DSS 6 . 2 Double Precision ( Ventana Systems , inc . http://vensim . com/ ) using techniques from the field of system dynamics [31] . Most of the parameter values used in the model are well established in the literature ( see Table 2 ) ; however , uncertain parameters include KmO2 , KmN2O , vemaxO2− , and vemin− . KmO2 ( Eq 17 ) . Pa . denitrificans has three haem-copper terminal oxidoreductases [39] with KmO2 ranging from nM to µM [40 , 41] , so we decided to estimate the parameter value by optimising KmO2 for the low [O2] treatments data . Vensim was used for the optimisation , where KmO2 = 2 . 25×10−7 neatly simulated the O2 depletion for both the succinate- and butyrate-supplemented treatments . KmN2O . In vitro studies of NosZ from Pa . denitrificans estimate the values for KmN2O = 5 μM at 22°C and pH 7 . 1 [42] and 6 . 7 μM at 25°C and pH 7 . 1 [43] . When our model was simulated with KmN2O in this range , given our empirically estimated vemaxN2O− [24] , the simulated N2O reached concentrations much higher than that measured ( see Results/Discussion ) . A more adequate parameter value ( = 0 . 6 μM ) was found by optimising KmN2O in Vensim . The value is within the range determined for soil bacterial communities [44] . vemaxO2− ( Eq 17 ) could be estimated using the empirically determined cell yield per mole of electrons to O2 ( YeO2− , cells per mol e- ) and the maximum specific growth rate ( μ , h-1 ) : vemaxO2−=μYeO2− . We are confident about the yields for the two C-substrates used , but the empirically determined μ for the butyrate treatments is suspiciously low ( = 0 . 067 h-1 ) , providing vemaxO2− = 2 . 45×10−15 mol e- cell-1 h-1 . Simulations with this value grossly underestimated the rate of O2 depletion compared to measured , which forced us to estimate the parameter value by optimisation: vemaxO2− = 4 . 42×10−15 and 4 . 22×10−15 mol e- cell-1 h-1 for the succinate- and butyrate treatments , respectively . These values give μ = 0 . 22 and 0 . 12 h-1 , respectively: for the succinate treatments , the value is very close to that empirically determined ( = 0 . 2 h-1 ) ; for the butyrate treatments , the value seems more realistic than 0 . 067 h-1 . vemin− ( Eqs 2 , 5 and 7 ) is the per cell velocity of e--flow to O2 ( veO2− ) assumed to generate minimum ATP required for synthesising the initial molecules of denitrification enzymes . Since we lack any empirical or other estimations for this parameter , it is arbitrarily assumed to be the veO2− when [O2]aq reaches 1 nM . At this concentration , vemin− is determined by the Michaelis-Menten equation ( vemin−=vemaxO2−×[O2]aq ( KmO2+[O2]aq ) ) , using vemaxO2− and KmO2 given above . The values obtained for the succinate- and butyrate-supplemented treatments = 1 . 96×10−17 and 1 . 87×10−17 mol e- cell-1 h-1 , respectively , which for both the cases is 0 . 44% of vemaxO2− . To investigate the impact of vemin− on the model behaviour ( rNa and rNi , Eqs 1 , 2 , 4 , 5 , 6 and 7 ) , sensitivity analyses were performed by simulating the model with vemin− corresponding to [O2]aq = 5×10−9 , 5×10−10 , and 0 mol L- 1 ( see Results/Discussion ) .
To test the assumption of a single homogeneous population with all cells producing Nar in response to O2 depletion , we simulated the model with the specific probability for a Z− cell to initiate nar transcription ( rNa ) = 4 h-1 . This resulted in 98% of the cells possessing Nar within an hour ( see Eqs 1–3 ) . Evidence suggests that less than half an hour is required to synthesise denitrification enzymes [17 , 18] , but an hour’s time is assumed here to allow margin for error . The results show that , for all the treatments , the simulated NO2− production ( mol vial-1 ) grossly overestimates that measured ( Fig 3 ) . To find a reasonable parameter value , we optimised rNa for the 0% O2 treatments , so that the simulated NO2− production matches that measured . The results ( Table 3 ) suggest that a low probabilistic initiation of nar transcription ( average rNa = 0 . 035 h-1 ) is adequate to simulate the measured NO2− kinetics ( Fig 3 ) . In the Butyrate , 7% O2 treatment ( Fig 3B ) , the simulated NO2− starts earlier , but the rate of accumulation is similar to that measured . Once O2 falls below a certain threshold , the production of Nar is assumed to trigger with rNa = 0 . 035 h-1 and last until a minimum of respiration is sustained by the e--flow to O2 and N2O ( veO2− and veN2O− ) , assumed to fulfil the ATP needs for Nar production ( Eqs 1 and 2 ) . But the production of Nar sustained by veN2O− was inconsequential for simulating the measured NO2− production , since NO3− was already exhausted when N2O started accumulating ( i . e . , when veN2O− > 0 ) . For this reason , the fraction that produced Nar ( FNa , Eq 3 and Table 4 ) is calculated as functional ( = 0 . 23–0 . 43 ) and theoretical ( = 0 . 56–0 . 81 ) , where the first is the fraction actually responsible for NO2− production ( sustained by veO2− ) , but the latter also incorporates the fraction that produced Nar after the exhaustion of NO3− ( sustained by veO2− as well as veN2O− ) . The rationale behind calculating the theoretical FNa is the empirical data indicating that Nar transcription is not turned off in response to NO3− depletion [18] . Although our model cannot test the theoretical FNa , but the functional FNa suggests that , contrary to the common assumption , the measured NO2− kinetics can be neatly explained by only 23–43 . 3% of the population producing Nar in response to O2 depletion . When we optimised the specific probability of nirS transcriptional activation ( rNi , see Eqs 4 , 5 , 6 and 7 ) to fit the measured data , the average rNi = 0 . 004 h-1 ( Table 3 ) adequately simulated the measured NO2− depletion and N2 accumulation ( Fig 4 ) . The recruitment to denitrification lasted for 19 . 5–47 . 3 h , i . e . , the time when [O2] was below a critical concentration and the velocity of e--flow to O2 and the relevant NOx− /NOx remained above a critical minimum ( Eqs 4 , 5 , 6 and 7 ) . The resulting fraction recruited to denitrification ( FNi , see Eq 8 and Table 4 ) was 0 . 08–0 . 18 , the bulk of which depended on the e--flow to NO3− and N2O ( instead of aerobic respiration ) . To test whether the measured data could be explained without the recruitment sustained by NO3− and N2O respiration , we also simulated the model with the recruitment as a function of O2 alone and re-optimised rNi , which on average increased to 0 . 012 h-1 ( providing FNi = 0 . 083–0 . 35 ) . This was expected since O2 is exhausted rather quickly , shrinking the time-window available for the recruitment . Comparatively , these simulations were less satisfactory: using the average rNi = 0 . 012 h-1 generally resulted in larger deviations than for the default simulations ( S2 Fig ) , and the optimal rNi for individual treatments varied grossly ( 50% higher values for the ~0% O2 treatments than for the 7% O2 treatments ) . This contrasts the default simulations , where the optimal rNi values for individual treatments were quite similar . Recruitment to denitrification ( both nar and nirS transcription ) is assumed to continue only as long as the combined e--flow to O2 , NO3− and N2O is greater than vemin− ( Eqs 1 , 2 , 4 , 5 , 6 and 7 ) . To test the model’s sensitivity to this parameter , we estimated rNa and rNi by optimisation for different values of vemin− , relative to the default value = 1 . 95×10−17 mol e- cell-1 h-1 . For all cases , the model was able to adequately simulate the measured N2 kinetics by moderate adjustments of rNa and rNi . Table 5 shows the average optimal values of rNa and rNi , obtained by fitting the simulated N2 kinetics to the data , for different values of vemin− . S3 Fig shows adequate simulations of the measured N2 kinetics assuming vemin− = 0 , with optimised rNa = 0 . 033 h-1 and rNi = 0 . 0033 h-1 . Thus , although assuming vemin− > 0 appears logical , it is not necessary to explain the measured data . To simulate the N2O kinetics , we used vemaxN2O− = 5 . 5×10−15 mol e- cell-1 h-1 , empirically determined under similar experimental conditions as simulated here [24] , and adopted the literature values for KmN2O [= 5 and 7 μM 42 , 43 , respectively] . But with KmN2O = 5 μM , the model predicted N2O accumulation ~10–20 times higher than measured for the ~0% and ~2–3 times higher for the 7% O2 treatments ( Fig 5 ) . This forced us to simulate the model with the parameter value estimated by optimisation , providing the average KmN2O = 0 . 6 μM . The measured N2O shows a conspicuous increase throughout the entire active denitrification period , and this phenomenon is neatly captured by the model . The reason for this model prediction is that the number of N2O producing cells ( ZNaNi + ZNi , Fig 2A ) is low to begin with compared to the number of N2O consuming cells ( Z− + ZNa+ ZNaNi + ZNi ) , but the fraction of N2O producers will increase during the anoxic phase for two reasons: one is the recruitment to ZNaNi & ZNi , another is the fact that the model predicts approximately three times faster cell-specific growth rate for ZNaNi & ZNi than for Z− & ZNa ( veN2O− is identical for all groups , while veNO2−− and veNO− are both zero for Z− & ZNa but for ZNaNi & ZNi , it holds that veNO2−− ≈ veNO− > veN2O− . To illustrate this phenomenon , we ran the model , assuming that the Z− & ZNa cells had no N2O reductase , resulting in a ) constant N2O concentration throughout the entire anoxic phase and b ) much higher N2O concentrations than measured ( Fig 5 ) . The overestimation is a trivial result , easily avoidable by increasing vemaxN2O− or decreasing KmN2O moderately . However , the prediction of a constant N2O concentration is clearly in conflict with the experimental data , and no parameterisation could force the model to reproduce this phenomenon other than the differential expression of denitrification genes . Hence , although there is room for further refinements , our default assumption regarding differential expression of NirS and NosZ explains the observed N2O kinetics: 1 ) abrupt initial accumulation to very low levels due to recruitment of relatively small numbers to the N2O producing pools ( ZNaNi & ZNi ) , and 2 ) increasing N2O concentration due to recruitment and faster cell-specific growth of ZNaNi & ZNi than that of the cells only consuming N2O ( Z− + ZNa ) . This modelling exercise sheds some light on the possible role of regulatory biology of denitrification in controlling N2O emissions from soils . If all cells in soils had the same regulatory phenotype as Pa . denitrificans , their emission of N2O would probably be miniscule , and soils could easily become strong net sinks for N2O because the majority of cells would be ‘truncated denitrifiers’ with only N2O reductase expressed . It remains to be tested , however , if the regulatory phenotype of Pa . denitrificans is a rare or a common phenomenon among full-fledged denitrifiers . We foresee that further exploration of denitrification phenotypes will unravel a plethora of response patterns . Using dynamic modelling , we have demonstrated that the denitrification kinetics in Pa . denitrificans can be adequately explained by assuming low probabilistic transcriptional activation of the nar and nirS genes and a subsequent autocatalytic expression of the enzymes . Such autocatalytic gene expressions are common in prokaryotes , rendering a population heterogeneous because of the stochastic initiation of gene transcription , with a low probability [45] . For N2O kinetics , our hypothesis was that a ) the gas is produced by a fraction of the incubated population that is able to initiate nirS transcription with a certain probability , leading to a coordinated expression of nirS + nor via NO [16] , and b ) N2O is consumed by the entire population because , in response to anoxia , nosZ is readily induced by FnrP [24] . Our model corroborated this hypothesis by reasonably simulating the N2O kinetics with the specific-probability of nirS transcriptional activation = 0 . 004 h-1 , resulting in 7 . 7–22 . 1% of the population producing NirS + cNor ( hence N2O ) , but all cells producing NosZ ( hence equally consuming N2O ) . | Denitrifiers generally respire O2 , but if O2 becomes limiting , they may switch to anaerobic respiration ( denitrification ) by producing NO3− - , NO2− - , NO- and/or N2O reductase , encoded by nar , nir , nor , and nosZ genes , respectively . Denitrification causes transient accumulation of NO2− and NO/N2O emissions , depending on the activity of the four reductases . Denitrifiers lacking nosZ produce ~100% N2O , whereas organisms with only nosZ are net consumers of N2O . Full-fledged denitrifiers are equipped with all four reductases , genetic regulation of which determines NO2− accumulation and NO/N2O emissions . Paracoccus denitrificans is a full-fledged denitrifying bacterium , and here we present a modelling approach to understand its gene regulation . We found that the observed transient accumulation of NO2− and N2O can be neatly explained by assuming cell diversification: all cells expressing nosZ , while a minority expressing nar and nir+nor . Thus , the model predicts that in a batch culture of this organism , only a minor sub-population is full-fledged denitrifier . The cell diversification is a plausible outcome of stochastic initiation of nar- and nir transcription , which then becomes autocatalytic by NO2− and NO , respectively . The findings are important for understanding the regulation of denitrification in bacteria: product-induced transcription of denitrification genes is common , and we surmise that diversification in response to anoxia is widespread . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results/Discussion"
] | [] | 2016 | Transient Accumulation of NO2- and N2O during Denitrification Explained by Assuming Cell Diversification by Stochastic Transcription of Denitrification Genes |
Our ability to detect target sounds in complex acoustic backgrounds is often limited not by the ear's resolution , but by the brain's information-processing capacity . The neural mechanisms and loci of this “informational masking” are unknown . We combined magnetoencephalography with simultaneous behavioral measures in humans to investigate neural correlates of informational masking and auditory perceptual awareness in the auditory cortex . Cortical responses were sorted according to whether or not target sounds were detected by the listener in a complex , randomly varying multi-tone background known to produce informational masking . Detected target sounds elicited a prominent , long-latency response ( 50–250 ms ) , whereas undetected targets did not . In contrast , both detected and undetected targets produced equally robust auditory middle-latency , steady-state responses , presumably from the primary auditory cortex . These findings indicate that neural correlates of auditory awareness in informational masking emerge between early and late stages of processing within the auditory cortex .
On a busy street corner , in a crowded restaurant , or in a rainforest at twilight , the sounds emitted from multiple sources mix together to form a highly convoluted and complex acoustic environment . Ecologically relevant warning or mating calls , or the speech from your neighbor at a restaurant table , must be heard out of this background cacophony . When a certain sound is not heard out of a background mixture , it is said to be masked . Many examples of masking can be explained in terms of the way sounds are processed in the inner ear , or cochlea [1] . The background or masking sound produces a pattern of excitation in the cochlea that either swamps or suppresses the activity due to the target sound , so that the target is no longer accurately represented in the auditory nerve [2] . This form of masking , traditionally known as “energetic masking , ” has been the subject of most formal psychophysical studies of masking dating back nearly 100 years [3] . In general such masking , measured behaviorally , corresponds well to predictions based on physiological measurements from the cochlea or auditory nerve [4 , 5] . The maskers and targets used in such experiments are typically predictable ( i . e . , the same sounds are presented over many repetitions ) , and are easily distinguished from one another . More recently it has become clear that the principles and predictions of energetic masking may not hold in many natural situations , where competing sounds are neither predictable nor readily distinguishable . Masking under conditions of uncertainty and timbral similarity has been referred to as “informational masking . ” The term informational masking , which was initially applied to the perception of elemental sounds , such as pure tones [6 , 7] , has more recently been applied to a wide range of contexts , including the masking of speech by other speech sounds [8] . Although it is unlikely that the same mechanisms underlie all forms of informational masking , they all have in common that the effects cannot be explained in terms of interactions in the auditory periphery ( the cochlea and auditory nerve ) [9] . In this study , we investigated the neural correlates of informational masking as it applies to the detection of a target tone sequence embedded in a random multi-tone background . Where and how informational masking occurs in the auditory system remains unknown . In fact , with our current state of knowledge , informational masking may originate at any processing stage along the auditory pathways , from the cochlear nucleus in the brainstem , up to ( and possibly beyond ) the auditory cortex ( AC ) . We combined a behavioral informational masking paradigm with simultaneous magnetoencephalography ( MEG ) recordings in humans to investigate the role of the AC in informational masking in particular , and auditory awareness in general . Our listeners' task was to detect a stream of regularly repeating target tones against a background of masking tones that were randomly placed in time and frequency ( Figure 1A ) . The stimuli are similar to those used in earlier studies of informational masking using random multi-tone backgrounds [10 , 11] , with the exception that our masking tones were not synchronized with the target tones . This desynchronization allowed us to separate the time-locked MEG responses evoked by the target tones from those evoked by the masker tones . To limit the contribution of energetic masking and peripheral interactions between the targets and maskers , the target tones were separated from the masking tones by a fixed minimum frequency gap or “protected region . ” A frequency gap also promotes the perceptual segregation of target and masker tones into distinct sound “streams , ” making it easier for listeners to identify the regularly repeating , constant-frequency target tones amid the randomly varying masker tones [12–14] . Although the presence of the target is obvious in the visual representation of Figure 1A , the targets in this configuration were not clearly audible; in fact , listeners reported hearing them on only about half the presentations . On some trials , the target tones “popped out” from the background and became clearly audible well before the end of the stimulus sequence; on other trials they were not heard at all . Such dramatic changes in perception from one trial to the next are typical in informational masking experiments . Because detection in this task is not associated with systematic changes in the physical stimuli ( the exact same stimulus can elicit detection on one occasion and not on another ) , this paradigm provides ideal conditions for identifying neural correlates of auditory awareness , independent of both physical stimulus manipulations and peripheral auditory interactions . We compared MEG signals that were time-locked to either detected or undetected target tones in the AC . We identified robust early AC responses ( the middle-latency steady-state responses—SSR ) to the target , which remained the same whether the target was detected or not . Changes in later AC responses , starting approximately 70 ms after target onset , were found to depend critically on whether listeners were aware of the target tones . This longer-latency MEG response was strong when listeners reported hearing the target , but was not measurable when listeners failed to detect the target , or when their attention was directed elsewhere . The finding of robust early neural responses in the AC to sounds , regardless of whether they are detected , in conjunction with later AC responses that are highly correlated with detection , suggests that auditory awareness in a classical informational masking paradigm emerges from within the AC , rather than in lower-level brainstem or higher-level supra-modal cortical structures .
In the first experiment , listeners were presented with 10 . 4-s stochastic tone sequences generated by adding multiple tone bursts with pseudo-random frequencies and onset times . In two-thirds of these random-onset multi-tone sequences , a tone repeating regularly at a constant frequency throughout the sequence was added ( Figure 1A ) . To indicate when they were aware of these targets , listeners were instructed to press a key as soon as they began to hear the regularly repeating target tones against the randomly varying background tones . The probability that listeners detected the target stream increased over the duration of each sequence , reaching on average about 0 . 6 by the end of the sequence . The rate of false-alarms ( i . e . , target-detection responses on trials in which only the masker tones were presented ) also increased slightly over the course of the stimulus sequence , reflecting listeners' increasing expectation to hear out the target tones , but remained low overall ( Figure 1B ) . The listeners' unbiased detection performance , d' , computed as the difference between the z-transformed hit and false-alarm rates , increased over the duration of the sequence , reaching an average value close to 2 at the end ( Figure S1 ) . The experiment was repeated with two different informational maskers: in one ( Experiment 1A ) , the average stimulus-onset asynchrony ( SOA ) , defined as the time interval between the onsets of two consecutive tones within each of the masker frequency bands , was 200 ms; in the other ( Experiment 1B ) , the average SOA was 800 ms , producing a more sparsely populated masking stimulus ( compare left and right panels in Figure 1A ) . The behavioral results obtained with these two variants of the experiment were very similar overall ( Figure 1B ) . Although hit and false-alarm rates were slightly higher in the 200-ms SOA condition than in the 800-ms SOA condition , the average values of d' ( 1 . 81 and 1 . 76 , respectively ) did not differ significantly from each other ( F ( 1 , 11 ) = 0 . 09; p = 0 . 7674; Figure S1 ) , indicating that the amount of informational masking was essentially the same in both conditions . Therefore , the data from Experiments 1A and 1B were pooled in most instances for the analyses presented below . To determine whether time-locked brain activity in response to the targets depended on them being consciously detected by the listeners , MEG responses to detected target tones were averaged separately from MEG responses to undetected target tones . Detected targets evoked a prominent bilateral wave with maximal amplitudes on gradiometers positioned over the temporal lobes in the time range from 50 to 250 ms after stimulus onset . The topography of this wave was similar to that of the well-known N1m , evoked by single tones in silence . A source analysis with two dipoles , one for each auditory cortex , consistently resulted in dipole locations in Heschl's gyrus or planum temporale , or very close to it , with respect to the listeners' individual magnetic resonance imaging ( MRI ) anatomy ( Figure 1C ) . Averaged across listeners , Talairach coordinates ( Table 1 ) were located in the central AC , at the border between Heschl's gyrus and planum temporale , as determined in representative populations [15 , 16] . The variance was similar to that found for other components generated in the auditory cortex [17 , 18] . The location of fitted dipoles in the presence of the masking tones was not significantly different from the location of the dipoles fitted to the N1m measured in the target-alone condition , in the absence of any masking tones ( F ( 2 , 22 ) = 0 . 028; p = 0 . 7603 ) . The fitted dipoles were then used as a spatial filter to generate source waveforms [19] , estimating the time course of MEG activity in the auditory cortex . The source waveforms were qualitatively very similar when the detected-target or the target-only conditions were used to fit the dipoles , and the following summary is based on the detected-target conditions . The source waveforms associated with these dipoles are shown in Figure 1D ( confidence intervals represent bootstrap based t-intervals , p < 0 . 05 , two-tailed ) . The averaged response to detected target tones showed a prominent negativity ( detected versus undetected targets: F ( 1 , 11 ) = 32 . 15; p = 0 . 0001 ) , peaking around 120–200 ms after tone onset [mean peak latency: 183 ± 14 ms s . e . m . ( 200-ms-SOA masker ) ; 141 ± 9 ms s . e . m . ( 800-ms-SOA masker ) ] . The wave was broad-based , and the deviation of the trace from 0 was statistically significant ( p < 0 . 05 ) everywhere in a 71–283-ms range around the negative peak . There was no significant difference in amplitude ( F ( 1 , 11 ) = 0 . 63; p = 0 . 4425 ) or latency ( F ( 1 , 11 ) = 1 . 16; p = 0 . 3044 ) between right and left hemispheres . We refer to the negative wave evoked by the detected tones as the “awareness related negativity” or ARN . This functional label was chosen for convenience and to avoid premature assignment to another response component; it does not imply that the ARN is necessarily a completely separate component of the auditory evoked fields . The ARN peak was somewhat smaller in magnitude and longer in latency than the typical N1m evoked by the target tones presented without the masker ( lower right panel in Figure 1C ) , which peaked at 108 ms ( ±10 ms s . e . m ) , and was significant from 50 to 276 ms post stimulus onset ( p < 0 . 05 ) . In contrast , the average MEG response to undetected target tones was essentially flat , similar to the average response to sequences containing only the masker and no target ( undetected targets versus masker-only trials: F ( 1 , 11 ) = 4 . 12; p = 0 . 0672 ) . The fact that the no-target trace is flat confirms our expectation that the systematic randomization of masking-tone onset effectively cancelled out the responses to the masking tones . In addition , this trace provides a baseline against which the responses to undetected targets can be compared . The results suggest that the responses to undetected targets are very similar to those found for no targets . In experiment 1A and 1B , recording of the ARN was coupled to an active task that involved motor responses . Here we evaluated if the ARN could also be recorded when no active task was performed while listening . Subjects listened passively to a set of shorter-duration sequences ( 4 . 8 s ) , consisting of six target tones and an 800-ms SOA random-onset multitone masker , as well as control conditions comprising only target or only masker tones . Because the listeners' perception remained unknown in the passive setup , an additional manipulation was introduced: the identical masker and target sequences were presented twice ( at random positions within the presentation ) , once in isolation , and once preceded by a cue before the target , in an effort to make the subsequent target tones more detectable [20] . The cue consisted of three tones of identical frequency , and presented with the same 800-ms SOA , as the subsequent target tones , which were presented in silence prior to the multitone masker . It was assumed that the ARN would be larger in the cued condition , because of reduced uncertainty and informational masking , allowing for a larger number of consciously perceived target tones in the cued trials . The results of experiment 1C confirmed this prediction ( Figure 3 ) . A significantly larger ARN was evoked by the six target tones following the cue compared to the six non-cued target tones ( F ( 1 , 11 ) = 15 . 67; p = 0 . 0022; difference significant from 57 to 307 ms , p < 0 . 05 ) . This result indicates that the ARN does not depend on motor preparation or other task-related processes . Conscious detection of a target tone will generally be expected to involve the allocation of attention toward the target . In the context of a multitone masker ( without a cue ) , the perceptual salience of the target is comparatively low , and the contribution of bottom-up ( exogenous ) mechanisms that could attract attention toward the targets is unlikely to be very strong . Therefore , the direction of attention toward the targets is likely to facilitate detection , while directing attention away from the targets is likely to impair detection . If so , directing listeners' attention away from the targets should lead to a reduced or absent ARN in response to the targets . In experiment 2 , the target and masker tone sequences were presented to the left ear only , while an unrelated stimulus sequence , containing occasional “deviant” tones interspersed among standard tones , was presented to the right ear ( see Materials and Methods for details ) . In the first phase of this experiment , the listeners were instructed to detect the deviant tones in the right ear . They were not informed that regularly repeating tones would sometimes be presented to the left ear and , when later interviewed , ten listeners reported that they had heard only irregular bleeps in their left ear; only two listeners reported occasionally noticing regularly repeating tones . In the second phase , listeners were instructed to attend to stimuli in the left ear , ignoring tones in the right ear , and to indicate when they detected the regularly repeating target tones . The stimuli used in the two phases of the experiment were identical . The average percentage of correct responses in the right-ear deviant-detection task ( first phase of the experiment ) was 86 . 8% ( ±11 . 7% , S . D . ) and the percentage of false alarms was 2 . 1% ( ±2 . 7% , S . D . ) , yielding a d' of 3 . 4 ( ±0 . 7 S . D . ) . This high level of performance confirms that listeners were attending to the right-ear sequence , as intended . The MEG responses to the target tones were averaged into two groups , depending on whether those same ( identical ) physical stimuli were detected or undetected in the second phase of the experiment ( see below ) . The MEG responses collected during this first phase ( top panel in Figure 4A ) reveal that no ARN was evoked by target tones in the left ear when attention was directed away from them ( all targets versus masker epochs: F ( 2 , 22 ) = 0 . 23; p = 0 . 7970 ) . The traces were similar to those obtained during epochs where listeners had not detected the targets in the second phase of the experiment , where they were attending to the targets , comprising a P1m and a hint of an N1m . In the second phase of the experiment , where the task was to detect the repeating target tones in the left ear ( as in the original experiment ) , the average percentage of correct detections ( Figure 4C ) was 40 . 2% ( ±24 . 5 S . D . ) , corresponding to a d' of 1 . 05 ( Figure S1B ) . The MEG responses collected during this second phase ( middle panel of Figure 4A ) confirm the findings of the first experiment . They show a clear ARN in response to detected targets . No ARN was observed in the average MEG response to undetected targets ( detected versus undetected targets: F ( 1 , 11 ) = 25 . 93; p = 0 . 0003; no significant hemisphere effects; undetected targets versus masker-only epochs: F ( 1 , 11 ) = 1 . 24; p = 0 . 2900 ) . In a third and final phase of this experiment , the regularly repeating target tones were presented in the left ear without the multitone masker , while listeners again performed the right-ear distraction task . Although performance was similar to that measured during the first phase [correct responses: 82 . 4% ± 11 . 5%; false alarms: 1 . 6% ± 2 . 0%; d' = 3 . 3 ± 0 . 8 ( mean ± S . D . ) ] , listeners now reported being aware of the presence of regularly repeating tones in their left ear . Likewise , the unattended left-ear targets evoked a prominent N1m ( bottom panel in Figure 4A ) , and a short sustained field because of the longer tone duration ( 250 ms versus 100 ms; note that the ARN was also more sustained ) . Dipole locations for the N1m were not significantly different from those for the ARN measured in the second phase of the experiment ( F ( 2 , 22 ) = 0 . 64; p = 0 . 5264 ) , pointing to a generator of the ARN and N1m in the auditory cortex ( Table 1 ) . If informational masking were completely pre-cortical , neural correlates of target detection should be readily observed in the earliest cortical responses . To address this prediction in experiment 2 , we added sinusoidal amplitude modulation ( AM ) to the target tones in the left ear at a rate of 40 Hz . This allowed us to selectively record the middle-latency SSR evoked by the target tones , which has been identified in earlier studies as an index of early processing in the auditory core region of Heschl's gyrus [21–23] . Compared to the ARN results reported above , fundamentally different findings were obtained for the SSR evoked by the 40-Hz AM of the target tones ( Figure 5 ) . First , the SSR was present in both phases of the experiment , regardless of which ear the listener was attending . Second , when the listeners were attending to the target tones ( second phase of the experiment ) , the SSR was observed regardless of whether or not the target tones were detected . The lack of significant SSR in the masker-only condition ( gray waves in Figure 5B ) confirms the specificity of this measure for the target tones . The masker-evoked SSR was successfully canceled out by the averaging procedure , because the AM frequencies and onset phases of the masker tones were randomized . Overall , the SSR in response to the target tones was not significantly affected by either target detection ( F ( 1 , 11 ) = 0 . 14; p = 0 . 7125 ) or attention ( F ( 1 , 11 ) = 0 . 02; p = 0 . 9008 ) . There were no significant hemisphere effects in the presence of multitone masking , but the SSR was larger in the contralateral ( right ) hemisphere for AM tones presented in silence ( F ( 1 , 11 ) = 22 . 57; p = 0 . 0006 ) . Using the SSR , we detected no differences in early processing of detected and undetected target tones in the AC . This negative finding does not exclude the possibility of differential early processing of detected and undetected targets by mechanisms in or before the AC that are not reflected in this particular analysis . Nevertheless , the SSR data do show that the target tones are represented in the AC , even when they are not consciously perceived by the listener .
The present results demonstrate a clear co-variation between late neural responses from the human AC and listeners' awareness of sounds presented well above their detection threshold in quiet , and not masked in the sensory periphery . At the same time , the results demonstrate earlier neural responses in the AC to tones that remain undetected by the listener . The two MEG components studied here , the SSR and the long latency ARN , are both generated in the AC but reflect different processing stages . Conventional averaging of the SSR was used to maximize early phase-locked activity , and suppress later and non–phase-locked gamma-band activity [24 , 25] , to ensure that the SSR was specifically evoked by the target tones . The phase-locked SSR is tonotopically organized [21 , 26] , and is related to the middle-latency ( 20–50 ms ) response [22 , 23 , 27] , which , like the SSR , is mainly generated in the auditory core area [21 , 23 , 24 , 28 , 29] . Thus , the presence of the SSR during undetected tones provides a dissociation between the early activity in the AC and perceptual awareness , suggesting that although early activity in the auditory core may be necessary for perceptual awareness [30] , it is not sufficient [31] . In contrast to the SSR , the ARN appears to be closely related to the listeners' perceptual awareness of the target tones , as it was not observed for undetected or unattended targets . Source analyses performed on these data clearly indicate that the ARN is generated in the AC , around Heschl's gyrus . However , the dipole source analysis does not permit us to estimate the extent of the ARN source . Based on its latency and polarity , the ARN might be related to the auditory evoked N1m and Nd components . In contrast to the SSR , these components have been shown to be generated across multiple fields of the AC , including lateral Heschl's gyrus , planum temporale , and the superior temporal gyrus [17–19 , 32–34] , comprising the secondary or “belt” regions of the AC [35 , 36] . In summary , the present data indicate that the neural correlates of auditory perceptual awareness , as measured in the context of a relatively simple informational masking paradigm , can be found between early and late processing stages in the AC . In a finer anatomical view , these processes might be situated in core and belt areas of the AC [35 , 36] , respectively , although there is only indirect evidence for the latter hypothesis at present . In comparing the present findings to those of earlier studies , it is important to distinguish between the two forms of masking—“energetic masking” and “informational masking”—outlined in the introduction . Earlier studies have shown that auditory evoked electroencephalography ( EEG ) and MEG responses , including subcortical as well as cortical responses , can be strongly attenuated or abolished by the addition of masking noise [37–39] . The type of masking used in these studies corresponds to energetic masking , involving noise that overlaps in frequency and time with the target , which is commonly thought to originate at a peripheral level , reflecting direct physical interactions between the signal and the masking noise within the cochlea [2] . Using energetic masking and selective averaging based on listeners' responses , previous EEG studies have shown that waves P3 and N1 were observed over the vertex for detected targets only [40 , 41] . The P3 is currently thought to reflect activity in frontal and parietal cortex [42] , usually related to active task performance and novelty detection [43] . The AC might have additionally contributed to the N1 observed in one study [41] , but this was not investigated . In contrast to these earlier findings , the present results cannot be explained in terms of peripheral interactions between signal and masker , or in terms of novelty-detection or task-performance effects . First , the use of a protected spectral region around the target tones greatly reduced the influence of peripheral interactions between signal and masker . Second , the use of stimulus sequences containing multiple tone bursts , combined with a task that required listeners to report only the first detected target-tone repetition in an ongoing stream , dissociated perceptual detection from task-performance , and novelty effects . Finally , our finding that the ARN can be modulated by cueing listeners to the target tones , even when they were not actively performing the detection task , rules out an explanation in terms of task-performance effects . The finding of early cortical activity that is independent of detection on the one hand , and of a strong relationship between the longer-latency ARN and listeners' detection on the other hand , strongly suggests a neural correlate of detection within the AC for the multi-tone informational masking paradigm used here . A number of processes within the AC may determine whether a target is subject to informational multi-tone masking or not . One factor likely to play an important role is selective attention . The ARN had a similar source location to that of the N1m , which is evoked by target tones in the absence of the masker , and the two responses largely overlapped in time . The N1m has traditionally been considered an “automatic” component , which does not critically depend on overt attention [44] . However , this view is based mostly on results obtained under very low attentional loads , where the sounds evoking the N1m were not accompanied by other , competing sounds ( as in the target-only control in experiment 1 ) . In experiments with higher processing loads , where multiple sound streams are present , selective attention has been found to modulate responses in the AC [33 , 34 , 45–47] . However , a salient N1m is still observed in such settings ( as in the target-only control in experiment 2 ) , and listeners are usually aware of the presence of the unattended sound stream . It seems that only at very high processing loads , such as under the informational masking paradigm used here , is this response suppressed to the point where it is not measurable if the target is unattended or remains otherwise undetected . Taking our results together with those of earlier studies , we suggest that the degree to which selective attention affects later AC activity ( like the N1m ) may be explained by attentional load , with higher load leading to greater attentional modulation of the evoked responses . This explanation seems consistent with findings in the visual system , where selective attention has been shown to influence the competition for neural representation in cortex [48 , 49] . In our experiments , listeners were not able to attend selectively to the target tones from the beginning of each sequence , because the frequency at which the target tones were presented differed across presentations . Recent work has shown that under such circumstances , the detection of the target tones nonetheless occurs more rapidly than predicted by a serial search model , indicating additional bottom-up processes , such as an auditory “pop-out” effect [14] . This pop-out effect is expected to be closely related to automatic auditory-scene-analysis mechanisms , which are thought to parse acoustic stimuli based on low-level features ( such as frequency distance , temporal proximity , or spectral continuity over time ) and contribute to the formation of auditory streams [11 , 50] . Recent studies have identified neural phenomena that might subserve the formation of auditory streams in the AC [51–54] , and these streaming mechanisms may then again interact with mechanisms of selective attention via bottom-up activation of the ventral fronto-parietal attention system [55] . Based on these considerations , we suggest that , subsequent to early activation of the auditory core , limited processing resources in the AC [56] are a cause of informational masking , once a certain processing load is exceeded . Bottom-up mechanisms subserving stream segregation [11 , 50] , on the one hand , and top-down mechanisms of selective attention [55] , on the other hand , may bias the competition between auditory streams . This in turn may help determine the processing resources allocated to different streams within the AC , starting after 50–70 ms , in a manner that appears to be critical for auditory perceptual awareness .
Thirty-three listeners without history of hearing disorders participated in the study . Three groups of 12 listeners each ( six male , six female ) participated in experiments 1A and 1B ( one group ) , 1C , and 2 . One listener participated in all three experiments , and another one in all parts of experiment 1; the other listeners were different in each experiment . The study protocol was approved by the institutional review board of the University of Heidelberg Medical School; all participants provided written informed consent . Experiment 1: All stimuli were generated using a set of 18 frequency bands , whose center frequencies were spaced equally on a logarithmic scale between 239 and 5 , 000 Hz ( 239 , 286 , 342 , 409 , 489 , 585 , 699 , 836 , 1 , 000 , 1 , 196 , 1 , 430 , 1 , 710 , 2 , 045 , 2 , 445 , 2 , 924 , 3 , 497 , 4 , 181 , and 5 , 000 Hz ) . The target tones were selected from the six frequencies shown in bold , and remained constant throughout a 10 . 4-s sequence . Target tones were 100 ms in duration , including 10-ms on and off cosine-shaped ramps , and were repeated 12 times with a constant SOA of 800 ms . Two frequency bands on either side of the target frequency were excluded , as a “protected region , ” such that the masker comprised the remaining 13 frequency bands . Within each frequency band , the masker-tone frequency was chosen randomly around the center frequency ( fc ) within the width of one estimated equivalent rectangular bandwidths [ERB = 24 . 7 × ( 4 . 37 × f c + 1 ) ] , where fc is in kHz [1] . The masker started 800 ms before the target , resulting in a 10 . 4-s total duration for the sequence . The SOA between tones was randomized in the range of 100–300 ms or 100–1 , 500 ms , yielding average SOAs of 200 ms ( experiment 1A ) or 800 ms ( experiments 1B and 1C; the tone density and overall masker energy was accordingly lower in this case ) . Each of the six target frequencies was presented together with ten differently randomized masker sequences . Five of the ten masker sequences were also presented without the target tones . The resulting 90 different sequences were presented in random order , separated by silent intervals of 1 . 6 s . Five repetitions of the targets alone ( without the masker ) were presented as a control condition at the end of the session . All tone sequences were presented diotically ( to both ears ) . The level of the target tones was 40 dB sensation level ( SL ) per tone , and the level per tone of the masker was set 18 dB higher . In experiments 1A and 1B , listeners were familiarized with the stimuli before MEG recordings , and they were informed that the regularly repeating tones would not always be present ( although they were not told on what proportion of trials , or whether they would start and end at the same or different times ) . They were instructed to press the left button of the computer mouse whenever , and as soon as , they detected the repeating target tones . Listeners were encouraged to respond as quickly as possible after the onset of a new sequence , and they were told to press the right mouse button if the sequence ended before the masker , or if they had pressed the left button in error . In experiment 1C , listeners were instructed to listen passively to four types of stimulus sequences presented in random order . The first three were similar to the conditions of experiment 1B , but comprised only six consecutive target tones ( yielding a total duration of 4 . 8 s ) . The fourth type of stimulus sequence was obtained by adding three target tones in front of the original target sequence . The unmasked tones at the beginning of the target sequence provided listeners with a cue to the frequency of the target and decreased informational masking . Ten different maskers were generated for each of the six target frequencies . These same 60 masker sequences were used in the masker-only , uncued-target-plus-masker , and cued-target-plus-masker conditions . Experiment 2: Listeners were presented with target-plus-masker and masker-alone sequences similar to those used in the previous experiment . However , in this experiment , the target and masker tones were presented to the left ear only . Also , all tones were sinusoidally modulated in amplitude ( AM depth = 100% ) . For the target tones , an AM rate of 40 Hz was used to allow recording of the auditory 40-Hz SSR . For the masker tones , the AM rate was randomized between 20 and 50 Hz to maintain perceptual similarity , while avoiding interference between the target and masker evoked SSR [26] . Target tone duration was 250 ms , yielding ten modulation cycles per tone . The target tones were repeated at a constant SOA of 800 ms . The six target-tone frequencies were restricted to the range of 699 to 1 , 710 Hz ( frequency bands 7–12 , see also experiment 1 ) . The level of the masker tones was set 6 dB higher than that of the target tones . The masker-tone SOA varied between 250 and 1 , 350 ms ( average SOA = 800 ms ) . In their right ear , listeners were presented with a sequence of 100-ms AM tones ( 10-ms on and off ramps , AM rate = 100 Hz ) , the frequency of which was randomized between 700 and 1 , 700 Hz , and the SOA between 250 and 1 , 350 ms ( average SOA = 800 ms ) . The AM depth was 100% for the standards and 18 dB less for the 10% deviants , which were randomly interspersed among standards . The tone sequence in the right ear continued through the 1 . 6-s silent gaps separating consecutive stimulus sequences in the left ear . In a first phase , listeners were instructed to ignore the sounds in their left ear , attend to the sounds in their right ear , and press the right mouse button whenever ( and as precisely as possible after ) they detected a deviant in that ear . They were not informed that the stimuli presented in their left ear would sometimes contain repeating tones . In a second phase , listeners were instructed to ignore the sounds in their right ear , and to indicate the presence of the target sequence in their left ear , as in experiment 1 . In a final phase , the repeating target tones were presented to the listeners' left ear without any masker , while the listeners received the same instructions as in phase 1 ( attend right , ignore left ) . The MEG was recorded with a Neuromag-122 whole-head system . Data were recorded continuously with a 1 , 000-Hz sampling rate , in direct coupled mode . Stimuli were presented via foam ear pieces connected to ER-3 earphones ( Etymotic research ) by 1-m-long plastic tubes . Structural MRI scans ( T1 weighted , MPRAGE , voxel size 1 × 1 × 1 . 3 mm ) were obtained from each listener with a 3-T ( Siemens Trio ) whole-body scanner . MEG activity was averaged relative to the target tones . For sequences containing only maskers , MEG activity was averaged relative to times at which the target tones occurred in the combined sequences . Because the onset times of the masker tones were randomized independently from those of masker tones at other frequencies and from those of the target tones , the activity evoked by the masker tones canceled out in the averaging . In experiment 2 , the right-ear task ( phase 1 and 3 ) caused a low-frequency baseline fluctuation that was apparent during the masker-alone condition . Therefore , the average response to masker-alone sequences was subtracted from the average response to the other conditions in experiment 2 . In experiments and conditions where listeners indicated when they heard the targets , MEG activity recorded during epochs where listeners had detected the target tones were averaged separately from epochs where the listeners had not detected the target tones . Assuming that the target tones could only be identified after at least two such tones had been heard , the two tones prior to the response were also considered detected . Spatio-temporal dipole source analysis [19] was performed on the averaged data using brain electrical source analysis ( BESA ) . The data were low-pass filtered at 20 Hz ( 6 dB , zero-phase shift Butterworth filter ) . A baseline was set 25 ms prior to sound onset , and drifts and slow activity in the subsequent baseline epoch were removed by PCA-based spatial filtering . Dipole analysis of the detected condition was performed in a 100-ms analysis window encompassing the ARN peak . Two dipoles , one in each AC , were fitted to the data . This dipole model was then used as a spatial filter to explore the activation of the AC in the other conditions . For analysis of the 40-Hz SSR in experiment 2 , the data were band-pass filtered between 28 and 48 Hz ( 6 and 12 dB/octave , zero-phase Butterworth filter ) . The dipole model was fitted to the SSR of the unmasked targets from the control run . These dipoles were used as a fixed spatial filter to generate source waveforms of the SSR for conditions where the masker was present . The combined data of experiment 1 A and B were also averaged selectively for ( a ) each of the 12 target presentations , and ( b ) for each of the six target frequency bands used . Source waveforms were derived with the same dipole model as used for the above analysis ( additional 1-Hz high pass filter ) . Amplitudes and latencies were measured in the individual source waveforms . ARN and N1m amplitudes were measured as the average in the time window 75–175 ms , unless mentioned otherwise . Latencies were measured at the maximum in the time interval 75–275 ms . The peak-to-peak amplitude of the 40-Hz steady-state response was measured after averaging over the ten 25-ms cycles of modulation contained in the 250-ms target-tone duration . Confidence intervals for source waveforms were estimated by calculating t-intervals based on standard errors derived with the bootstrap technique based on 1 , 000 resamples [57] . Dipole positions were co-registered to the individual MRI morphology , and transformed into Talairach space using Brain Voyager . | Sounds that are well above the sensory threshold may sometimes fail to be perceived when they occur amid competing sounds , as often happens in everyday life . This phenomenon is generally referred to as “informational masking . ” We took advantage of this effect to isolate brain responses that correlate with conscious auditory perception . Human listeners performed an auditory detection task in which they had to indicate when they heard a stream of repeating tones ( targets ) embedded in a stochastic tone background ( masker ) . At the same time , brain responses were recorded using magnetoencephalography . By comparing the responses to perceptually detected and undetected target tones in the auditory cortex , we isolated a neural response component in the latency range of 50–250 ms , which was only present for detected sounds . We propose that this component , the “awareness related negativity , ” specifically reflects conscious sound perception . In contrast , earlier responses in the auditory cortex were evoked by both detected and undetected target tones . These results suggest that conscious sound perception emerges from within the auditory cortex . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"neuroscience"
] | 2008 | Neural Correlates of Auditory Perceptual Awareness under Informational Masking |
Following attachment to primary receptor heparan sulfate proteoglycans ( HSPG ) , human papillomavirus type 16 ( HPV16 ) particles undergo conformational changes affecting the major and minor capsid proteins , L1 and L2 , respectively . This results in exposure of the L2 N-terminus , transfer to uptake receptors , and infectious internalization . Here , we report that target cell cyclophilins , peptidyl-prolyl cis/trans isomerases , are required for efficient HPV16 infection . Cell surface cyclophilin B ( CyPB ) facilitates conformational changes in capsid proteins , resulting in exposure of the L2 N-terminus . Inhibition of CyPB blocked HPV16 infection by inducing noninfectious internalization . Mutation of a putative CyP binding site present in HPV16 L2 yielded exposed L2 N-terminus in the absence of active CyP and bypassed the need for cell surface CyPB . However , this mutant was still sensitive to CyP inhibition and required CyP for completion of infection , probably after internalization . Taken together , these data suggest that CyP is required during two distinct steps of HPV16 infection . Identification of cell surface CyPB will facilitate the study of the complex events preceding internalization and adds a putative drug target for prevention of HPV–induced diseases .
Cyclophilins ( CyP ) comprise a family of peptidyl-prolyl cis/trans isomerases , which are evolutionarily conserved and ubiquitously expressed [1] , [2] . CyP facilitate folding of nascent proteins and through this have been implicated in RNA splicing , stress responses , gene expression , cell signaling , mitochondrial function , and regulation of kinase activity [3] . The 16 human family members differ mainly by terminal extensions , which are probably responsible for subcellular localization and protein-protein interactions , and by tissue specific expression . CyP were initially identified as high affinity binding proteins for cyclosporin A ( CsA ) , an immunosuppressive agent [4] . CsA blocks the enzymatic acitivity of CyP . Cyclophilin A and B ( CyPA and CyPB ) are the most abundant among the family , where CyPA mainly localizes to the cytoplasm and CyPB , which encodes a signal peptide , is associated with the endoplasmic reticulum ( ER ) . CyPB can be secreted and is detected on the cell surface , where it colocalizes with heparan sulfate proteoglycans ( HSPGs ) like syndecan-1 [5] . Recent reports suggest that CyPB preferentially bind HSPG molecules that carry a 3-O-sulfated N-unsubstituted glucosaminoglycan residue in the heparan chain [6] . 3-O-sulfation is the least abundant modification of heparan sulfate and thus only few HSPG molecules on the cell surface are associated with CyPB . The core protein required for triggering biological function of cell surface CyPB is most likely syndecan-1 [5] . Several viruses exploit CyP for life cycle completion . The capsid protein of human immunodeficiency virus type 1 ( HIV-1 ) harbors a CyPA binding site resulting in the incorporation of this chaperone into the virion [7] . In addition , target cell CyPA is required for efficient infection of human cells [8] , [9] . Inhibition of CyPA prevents the transport of reverse transcribed viral genome to the nucleus without interfering with reverse transcription [10] . A number of observations were interpreted as CyPA preventing the interaction of the viral capsid protein with restriction factors rather than it promoting viral uncoating . In some nonpermissive cells , CyPA activity is required for binding of the restriction factor TRIM5 to the capsid protein ( for review see [7] ) . Hepatitis C virus ( HCV ) is another example requiring CyPB activity for efficient replication . It interacts with the viral polymerase NS5B thus promoting RNA binding [11] . Furthermore , mouse cytomegalovirus ( MCMV ) infection of neural stem/progenitor cells is facilitated by CyPA by an unknown mechanism [12] . Here we demonstrate that CyPB activity facilitates infection of human papillomavirus type 16 ( HPV16 ) and HPV18 . HPV are non-enveloped epitheliotropic DNA viruses with a circular , chromatinized , double stranded DNA genome of approximately 8000 bp . They induce benign lesions of the skin and mucosa that in some instances progress to malignancies . HPV induced malignancies , including cervical carcinoma , contribute to more than 7% of all cancers in women worldwide . The viral capsid is composed of 360 copies of the major capsid protein , L1 , and up to 72 copies of the minor capsid protein , L2 [13]–[15] . L1 protein , which is organized in 72 pentamers , called capsomeres , mediates the primary attachment of viral particles to the cell surface [16]–[18] and/or extracellular matrix ( ECM ) of susceptible cells [19] , most probably via HSPG [20] . The need for HS can be bypassed by treatment of immature HPV16 pseudovirions with furin convertase [21] . The primary attachment is mediated by surface-exposed lysine residues located at the rim of capsomeres [22] . HPV33 binding to the cell surface requires O-sulfation of HS , whereas both N- and O-sulfation are needed for HSPG to function as an initiator of the infectious entry pathway [23] . These data suggested that secondary HSPG interactions may play a role in infection , which was recently supported by the use of the HS binding drug DSTP-27 [18] . Virus attachment triggers conformational changes in both capsid proteins [23]–[25] , which seem to be required for transfer to putative secondary receptors and infectious internalization [18] . Conformational changes result in the exposure of the N-terminus of L2 protein , which contains a highly cross-reactive neutralizing epitope , and subsequent cleavage of 12 N-terminal amino acids catalyzed by furin convertase [24] , [26] . Data presented below suggest that cell surface CyPB facilitates exposure of the L2 N-terminus , which is required for infectious internalization .
We used a well established pseudovirus system for our studies , which relies on the expression of codon-modified forms of L1 and L2 in human embryonic kidney 293TT cells harboring a high copy number packaging plasmid [27] . We packaged a green fluorescent protein ( GFP ) –based marker plasmid that has been successfully used before to study early events of HPV infection [28]–[31] . We observed that CsA efficiently blocked HPV16 infection of 293TT cells with an inhibitory concentration 50 ( IC50 ) of approximately 2 µM ( Figure 1A ) . Similar results were obtained for HaCaT , which is currently the most commonly used keratinocytes-derived cell line for analysis of HPV infection , and HPV-harboring HeLa ( Figure 1B ) . CsA has been shown to block activity of calcineurin and CyP as well as P-glycoproteins , also known as ABC transporters . We used more specific inhibitors to narrow down the cellular target responsible for the observed inhibition . Neither INCA-6 nor the cell permeable R-VIVIT peptide and FK506 , inhibitors of the interaction between calcineurin and nuclear factor of activated T cells ( NFAT ) , blocked infection ( Figure 1C ) . Similarly , Virapamil and nifedipine , specific inhibitors of P-glycoproteins , had no effect . In contrast , NIM811 , which blocks both P-glycoproteins and CyP , inhibited HPV16 infection as efficiently as CsA . Identical results were obtained for key inhibitor NIM811 in HaCaT cells ( data not shown ) . All inhibitors reduced cell growth of 293TT ( Figure 1D ) , HaCaT and HeLa ( not shown ) cells to a similar extent . Cell growth inhibition of these inhibitors is well established . Taken together , these results strongly suggest that CyP facilitate HPV16 and HPV18 infection . In order to determine , which CyP family member may be involved and to confirm our findings , we used an siRNA approach to knock down individual CyP . First , we used an siRNA , si-CyP[broad] , which has been shown to target several members of the CyP family including CyPA , CyPB , CyPE , and CyPH [11] . 293TT cells were transfected with si-CyP[broad] 48 prior to infection with HPV16 . Western blot confirmed the significant reduction of steady state CyPA and CyPB protein levels ( Figure 2B ) and infection was reduced to 11% ( p<0 . 01 ) compared to cells transfected with a control siRNA ( Figure 2A ) . Individual knock down of CyPA and CyPB with specific validated siRNAs [11] also reduced infection to 59% ( p<0 . 05 ) and 35% ( p<0 . 01 ) , respectively . Specificity of the siRNA knock down for their target was confirmed by Western blot ( Figure 2B ) . The data indicated that both CyP may play a role in HPV16 infection . Compared to CyPA , knockdown of CyPB consistently resulted in stronger inhibition ( p<0 . 05 ) . Similar results were obtained for HaCaT cells . However , due to reduced transfection efficiency of HaCaT cells ( 70% vs . 95% for 293TT ) the inhibitory effect was not as pronounced ( Figure 2C and 2D ) . To identify the stage at which infection is blocked by CyP inhibitors , we first measured internalization using immunofluorescence ( IF ) . It was shown by several groups that most surface-exposed conformational epitopes that are recognized by neutralizing monoclonal antibodies ( NmAb ) are destroyed following entry and that L1 protein segregates from the L2/DNA complex in acidic endocytic compartments [18] , [29] , [30] . During this process reactivity of antibodies specific for hidden linear L1 epitopes is gained [32] . We used NmAb H16 . 56E to determine if conformational epitopes are lost in the presence of NIM811 . H16 . 56E binding site includes but is not restricted to the N-terminal portion of the FG loop ( HPV16 L1 residues 260–270 ) [33] . We also used mAb 33L1-7 , which binds a linear epitope ( residues 303–313 ) that is neither accessible in capsomeres nor in intact particles [34] , [35] and recognizes L1 protein late in HPV entry [32] . In untreated cells at 18 h post infection ( hpi ) with HPV16 pseudovirus , H16 . 56E reactivity was hardly detectable but perinuclear 33L1-7 staining was obvious indicative of particle internalization and accessibility of the 33L1-7 epitope ( Figure 3A ) . In contrast , we observed a strongly increased perinuclear signal with H16 . 56E when infection was performed in the presence of 10 µM NIM811 . The signal for 33L1-7 was greatly diminished under these conditions ( Figure 3A ) . Similar results were obtained when NIM811 was replaced by CsA ( data not shown ) . We will use the term ‘stabilized capsid phenotype’ to describe the increased reactivity of internalized pseudovirions with H16 . 56E . These data demonstrate that , first , viral particles are indeed internalized in the presence of CyP inhibitor and , second , the conformational L1 epitope recognized by H16 . 56E is stabilized . We again used siRNA knock down to identify the CyP family member responsible for the stabilized capsid phenotype . For this , HaCaT cells were transfected with unspecific control siRNA , si-CyP[broad] , si-CyPA or si-CyPB 48 h prior to infection with HPV16 pseudovirus . Successful down regulation of CyPB was confirmed by IF ( Figure 3B ) . Down regulation of CyPA could not be determined by IF because of lack of CyPA-specific antibody reactivity in this assay . However , successful transfection was monitored using FITC-labeled siRNA ( not shown ) and knock down of CyPA was confirmed by Western blot . Cells with reduced levels of CyPB following transfection with si-CyPB or si-CyP[broad] displayed a stabilized capsid phenotype at 18 hpi , whereas adjacent cells , which were not transfected as indicated by strong staining for CyPB , showed much less reactivity with H16 . 56E ( Figure 3B ) . A stabilized capsid phenotype was not detected in cells transfected with si-CyPA , even though basal level of reactivity with H16 . 56E is evident . Taken together these data suggest that blockage of CyPB activity may be responsible for the stabilized capsid phenotype . Previously we observed a stabilized capsid phenotype when transfer to secondary receptors on the cell surface was blocked by antibodies or drugs [18] . Furthermore , CyPB is found on the cell surface where it is associated with HSPG [6] , [36] . Therefore , we hypothesized that CyPB may facilitate the conformational shifts reported for both capsid proteins upon interaction with cell surface HSPG [23] , [25] . Currently , the only reliable test for these changes measures the exposure of the L2 N-terminus using the L2-specific NmAb RG-1 . RG-1 binds to a peptide encompassing HPV16 L2 residues 17 to 36 [37] . RG-1 reactivity with L2 protein incorporated into virions requires cell attachment-induced exposure of the L2 N-terminus and furin cleavage [24] . To test the role of CyP in conformational shifts , HPV16 pseudovirus was bound to HaCaT cells for 2 h at 4°C and was chased for 4 h at 37°C prior to cell surface staining with RG-1 ( a kind gift of R . B . Roden , John Hopkins University ) and K75 polyclonal VLP antisera . In control infection we found strong RG-1 signal , which perfectly overlapped with cell-associated L1-specific K75 binding ( Figure 4A ) . RG-1 reactivity was greatly diminished albeit not completely abolished when HaCaT cells were infected in the presence of NIM811 ( Figure 4A ) , whereas reactivity with K75 was not decreased . Similarly , CsA treatment decreased RG-1 signal albeit not as pronounced ( data not shown ) . We quantified the RG-1- and K75-specific signals using software provided by Zeiss and found statistically significant reductions of over 70% and 59% of relative RG-1 signal strength in presence of NIM811 and CsA , respectively ( p<0 . 01 ) ( Figure 4B ) . Taken together , these data strongly suggest that CyPB activity is required for exposure of the RG-1 epitope on the viral capsid and lend support for a function of cell surface CyPB in HPV16 infection . Recently , it has been shown that the presence of RG-1 antibody during infection of HaCaT cells with HPV16 pseudovirions prevents infection and virus internalization and relocates viral particles from the cell surface to ECM [24] . We took advantage of this observation to strengthen our findings . We reasoned that , irrespective of presence of RG-1 , viral particles should still be internalized and display a stabilized capsid phenotype in presence of NIM811 , if the RG-1 epitope is indeed not accessible to antibody binding after drug treatment . To test this , HPV16 pseudovirus was bound to HaCaT cells for 2 h at 4°C in the presence or absence of this drug . After washout of unbound virus , cells were incubated overnight in presence of NIM811 and RG-1 . Confirming previous findings [24] , RG-1 treatment alone induced deposition of the majority of viral particles to ECM in the absence of NIM811 , as evidenced by colocalization of capsid-specific H16 . 56E signal with the ECM marker Laminin 5 ( Figure 5A ) . We also confirmed the neutralizing capacity of RG-1 using 293TT cells ( Figure 5C ) to ascertain that this antibody is functional in our hands . Inhibition of HPV16 pseudovirus infection by this antibody using HaCaT cells was previously demonstrated by others [24] , [37] . However , the presence of RG-1 antibody in addition to drugs did not prevent internalization of viral capsids , as evidenced by a stabilized capsid phenotype ( Figure 5B ) and did not result in increased deposition of viral particles on ECM ( not shown ) . It should be noted that RG-1 treatment in the absence of NIM811 displayed a weak but reproducible stabilized capsid phenotype ( Figure 5B ) suggesting that not all particles are displaced from the cell surface and are instead internalized in a noninfectious manner . These data further support our notion that CyPB action on the cell surface is required for the conformational change resulting in exposure of the RG-1 epitope , which is a prerequisite for infectious internalization . Not much information is available regarding CyPB substrate binding sites . However , CyPA binding to the HIV capsid protein has been mapped to 85-PXXXGPXXP-93 , which is located between Helix 4 and 5 [7] . We found similar sequence elements at the N-terminus of L2 conserved among many but not all members of the Papillomaviridae family ( Figure 6A ) . We exchanged glycine and proline residues of L2 at positions 99 and 100 within the putative CyP binding site for alanine to test their importance for HPV16 infection . We hypothesized that this mutant is either defective for infection due to loss of CyP binding or does not require active CyP for exposure of the L2 N-terminus due to higher flexibility in this L2 region induced by amino acid exchanges . We found that 16L2-G99A-P100A ( 16L2-GP-N ) is incorporated into particles similar to wt L2 ( not shown ) . Mutant pseudovirus retains full infectivity in 293TT ( Figure 6B ) and HaCaT cells ( data not shown ) , which is consistently and statistically significantly increased compared to wt ( p<0 . 01 ) . When we bound 16L2-GP-N mutant pseudovirus to HaCaT cells and surface-stained with RG-1 and K75 after a 4 h chase at 37°C , we observed similar reactivity of RG-1 with cell-bound pseudovirions in absence or presence of NIM811 ( Figure 6C ) . Quantitative analysis of signal strength confirmed that reactivity of RG-1 with mutant pseudovirus is not significantly reduced by this drug ( Figure 6D ) in contrast to wt pseudovirus ( Figure 4 ) . These data suggested that 16L2-GP-N mutant pseudovirus does not require CyP activity for exposure of the RG-1 epitope . Nevertheless , infection was still sensitive to CsA ( Figure 7A ) and siRNA knock down of CyP ( Figure 7B ) . However , unlike wt pseudovirus mutant pseudovirus did not produce the stabilized capsid phenotype after treatment with drugs ( Figure 7C ) or siRNA knock down of CyP ( not shown ) , although H16 . 56E was still able to detect mutant viral particles on the cell surface and on ECM ( data not shown ) . Taken together , these data indicate not only that 16L2-GP-N mutant pseudovirus bypasses the requirement for cell surface CyPB but also that HPV16 infection requires CyP at a second , possibly intracellular , stage of entry and transport . Furthermore , they strongly support our previous notion that , in presence of CyP inhibitors , wt virus is shunted into a noninfectious entry pathway . To determine whether the requirement for CyP is a conserved feature among papillomaviruses we tested a number of low and highrisk HPV types as well as BPV-1 for sensitivity to CsA . We found that HPV6 , HPV45 and HPV58 were inhibited by CsA similar to HPV16 and HPV18 , whereas BPV1 , HPV5 , HPV31 , and HPV52 were relatively resistant to CsA ( Table 1 ) . These data suggest that different papillomavirus types have different requirements for CyP , which may be reflective of the entry strategies these viruses evolved .
Here , we report that CyP facilitate infection of the oncogenic HPV16 and 18 among other HPV types . Focusing on HPV16 and using specific drugs , siRNA knock down and mutant pseudovirus we provide evidence that CyP are required at two different stages following primary attachment to host cells . In addition , siRNA knock down data point to the involvement of two members of the CyP family in the infection process: CyPA and CyPB . Combined knock down using siCyP[broad] affected infection more severely than individual knock downs suggesting they may facilitate different steps of HPV16 infection . Our data indicate that CyPB is functioning on the cell surface . However , we were not yet able to identify the step requiring CyPA . Also , at the moment we cannot completely rule out the involvement of additional CyP family members , like CyPE and CyPH , whose expression should also be affected by siCyP[broad] . We provide evidence that cell surface CyPB is essential for triggering events that lead to infectious internalization of viral particles probably by catalyzing conformational changes of viral capsid proteins . It is well established that both HPV16 capsid proteins undergo conformational changes on the cell surface prior to internalization . Conformational changes induced in L1 are not well defined but seem to involve the BC loop [23] . Conformational changes induced in L2 protein result in exposure of some forty N-terminal amino acids , which allows furin convertase-mediated cleavage of L2 and binding of the L2-specific NmAb RG-1 [24] , [25] , [38] . CyP inhibition greatly reduced exposure of the RG-1 epitope following cell attachment as measured directly by IF and indirectly by determining the fate of cell bound pseudovirus in the presence of CyP inhibitors and RG-1 . This strongly indicates that CyP activity is required to make the RG-1 epitope accessible to antibody binding . However , the block was not complete and residual reactivity with RG-1 was observed in presence of inhibitors , which could possibly be attributed to the presence of activated particles in the pseudovirus preparation [18] , [28] and/or to the baseline spontaneous conformational change in absence of CyP activity due to receptor engagement . Nevertheless , the reduction in RG-1 reactivity by CyP-specific drugs was found to be correlated with reduction of infectivity by drugs . We also provide evidence that L2 protein may be the substrate for CyP . First , we were able to bypass the requirement for cell surface CyP by introducing amino acid changes in a putative CyP binding site within L2 , which is accessible in mature virions [39] . Mutant pseudovirus did not require CyP activity for exposure of L2 as demonstrated by IF . However , at this point we cannot completely rule out that CyP functions rather indirectly by ( i ) modifying cell surface receptors , since CyP have been shown to isomerize prolyl peptide bonds of cell surface markers , thus modifying their biological function [40] , [41] , and by ( ii ) regulating cell trafficking and cell surface expression of proteins [42] . However , this is rather unlikely since BPV1 , which uses the same route of internalization as HPV16 [43] , [44] , is not blocked by CyP inhibitors . Second , specific blockage of CyPB induced noninfectious internalization with the hallmark of a stabilized capsid phenotype . In this respect CyPB inhibition is similar to post-attachment treatment with the BC loop-specific antibody H33 . J3 , heparinase , or the HS binding drug DSTP-27 , which also induce noninfectious internalization and stabilization of viral capsids [18] . It was suggested that these treatments all block secondary receptor interactions , which seems to require an exposed L2 N-terminus [18] . It is unlikely that L1 rather than L2 protein is the substrate of CyP . This is based on our unpublished observations that CsA , NIM811 or CyP-specific siRNAs do not block L1 conformational changes occurring on the cell surface . Interestingly , mutant 16L2-GP-N pseudovirus remained sensitive to CyP inhibitory drugs and siRNA knock down . However , bypassing the need for cell surface CyPB using mutant pseudovirus yielded an inhibition phenotype distinct from wt particles . We no longer observed capsid stabilization . This suggests that CyP activity is required at a subsequent step during internalization and/or intracellular transport . So far , we were not able to identify the exact step ( s ) that require CyP activity and therefore cannot predict which specific CyP family member may be involved . However , a second putative CyP binding site is located near the C-terminus of L2 ( 409-PLVSGPDIP-417 ) . This sequence is close to a region that has been shown to mediate interaction of L2 with L1 capsomeres in HPV11 [45] . It is therefore tempting to speculate that endocytic CyP mediates segregation of L2 from L1 . As the C-terminal section of L2 is required for membrane destabilization and passage of membranes [30] as well as for interaction with dynein [46] , this could free the C-terminus allowing association and penetration of surrounding membranes and consequently the L2/DNA complex to egress from endosomes and retrograde transport towards the nucleus . CyPB encodes a signal peptide and is therefore found in the luminal compartment of intracellular membranes making it a likely candidate . However , CyPA is also secreted into the extracellular space , even though it lacks a signal peptide , suggesting it finds its way into the luminal compartment of at least the secretory pathway . Host cell CyP do not facilitate infection of all papillomaviruses . Support for this notion came from our finding that HPV5 , HPV31 , HPV52 , and BPV1 are rather resistant to CyP-specific drugs . This may reflect the evolution of different internalization strategies . For example , HPV31 is internalized via caveolae-dependent endocytosis [31] , [47] whereas HPV16 uses a caveolae- and clathrin-independent pathway [32] . BPV1 L2 does not harbor putative CyP binding sites and replacing key proline residues by more flexible amino acids may make a catalytic activity dispensable for L2 exposure , as we have shown with HPV16L2-GP-N . The entry pathways of HPV5 and HPV52 have not been investigated yet . However , the L2 protein of both HPV types has a putative N-terminal CyP binding site . Attachment-induced conformational changes are a common theme in virus infection . They are usually triggered by interaction with specific receptors , which allows interaction with secondary receptors or , more often , trigger cell fusion events . Although chaperones present in endocytic vesicles or the endoplasmic reticulum have been shown to facilitate virus uncoating and translocation across membranes [48] , [49] , this is the first report to implicate chaperones in mediating conformational changes of capsid proteins on the surface of target cells . With this report we are adding another virus family to the list of viruses dependent on CyP activity for completion of their life cycle . Despite 15 years of study , the role of CyPA in HIV-1 infection is not yet fully defined ( for review see [7] , [50] ) . Similarly , its involvement in MCMV infection of neural progenitor cells has not been characterized in molecular detail [12] , whereas it was convincingly demonstrated for HCV that ER-resident CyPB enhances the RNA binding activity of the NS5B RNA polymerase and consequently genome amplification [11] . With the identification of CyPB as modifier of oncogenic HPV capsid protein conformation , which activates the virus for entry via an infectious pathway , for the first time we have characterized its role at the molecular level during cell surface events of viral infections . This should allow characterizing the complex events preceding internalization in more detail and adds a putative drug target for prevention of HPV-induced diseases , especially since CsA has been approved for and is already being used in clinical settings .
293TT cells and expression plasmids for codon-optimized structural genes coding for HPV5 , HPV6 , HPV18 , HPV31 , HPV45 , HPV52 , HPV58 as well as BPV1 were kindly provided by John Schiller and Chris Buck , Bethesda [27] , [51] . Codon-optimized HPV16 L1 and L2 expression plasmids were a kind gift from Martin Müller , Heidelberg [52] . HPV16L1-specific rabbit polyclonal antisera K75 , mouse monoclonal antibody H16 . 56E and 33L1-7 have been described previously [34] , [35] . Anti-CyPA polyconal rabbit antibody was obtained from Dharmacon ( cat #: 07-313 ) . CyPB polyclonal rabbit antibody was purchased from Affinity BioReagents Inc ( Golden , Colorado; cat #: PA1-027 ) . However , we noticed that only lot number 328-120 and prior lots were reactive in IF . All subsequent lots tested were not reactive in IF analyses . Laminin 5 rabbit polyclonal antibody was from Abcam ( cat #: ab14509 ) . AF488-labeled GFP-specific rabbit polyclonal antibody was obtained from Invitrogen . Mouse monoclonal L2-specific RG-1 antibody was kindly provided by Richard Roden , John Hopkins University , Baltimore . AlexaFluor ( AF ) –labeled secondary antibodies and phalloidin were purchased from Invitrogen . Pseudovirions were generated and purified using Optiprep gradient centrifugation following published procedures [27] . Pseudovirus yield was determined by green fluorescent protein ( GFP ) –specific quantitative real time polymerase chain reaction ( qRT–PCR ) . Cyclosporin A was obtained from Toronto Research Chemicals ( cat #: C988900 ) . NIM811 was a kind gift from Novartis . Verapamil ( cat #: 676777 ) , Nifedipine ( cat #: 481981 ) , 11R-VIVIT ( cat #: 480401 ) and INCA-6 ( cat #: 480403 ) were obtained from Calbiochem . FK 506 ( cat #: F1030 ) was purchased from A . G . Scientific ( San Diego ) . The cell viability and proliferation assay ‘CellTiter96 Aqueous One Solution’ was purchased from Promega ( Madison , WI ) . This assay measures the quantity of formazan product , which is directly proportional to the number of living cells . 293TT cells were seeded a day before and allowed to attach . Next day , drugs were serially diluted in complete DMEM in 24 well-plates and adequate amounts of pseudoviruses were added to achieve infection levels of 10 to 30% . Infectivity was scored by counting GFP expressing cells at 72 hpi using flow cytometry . Similar protocol was followed for infection assay using HaCaT and HeLa cells except that cells were fixed with 2% paraformaldehyde , permeabilized with 0 . 2% Triton X-100 in phosphate buffered saline ( PBS ) , stained with AF488-labeled GFP-specific antibody and counted using a Leica DMBI 6000 fluorescence microscope . Unless otherwise stated standard deviation was based on at least five replicates from at least two independent experiments . RNA interference was carried out using synthetic siRNA duplexes with symmetric 3′-deoxythymidine overhangs . siRNA duplexes si-CyPA , 5′-AAGCATA CGGGTCCTGGCATC-3′; si-CyPB , 5′-AAGGTGGAGAGCACCAAGACA-3′; and si-CyP ( broad ) , 5′-AAGCATGTGGTGTTTGGCAAA-3′ ) , which have been described and validated before [11] , were purchased from Integrated DNA Technologies Inc . Non-specific siRNA , si-NS , 5′-AAGTCCGTGCCGTCAGTTCTCAGAA-3′ was obtained from Invitrogen . Cells were transfected with 3 µg of siRNA duplexes in serum-free medium using MATra reagent ( IBA biotagnology , Goettingen; cat . #: 7-2001-100 ) according to manufacturer's protocol . Typical siRNA transfection efficiency was found to be 70% for HaCaT and 95% for 293TT cells as monitored by fluorescein-labeled control siRNA duplex . CyP knockdown was confirmed 48 h post siRNA transfection ( hpTx ) by Western blot . HaCaT and 293TT cells were transfected with siRNA as mentioned above . 48 hpTx , HaCaT were harvested with trypsin and reseeded onto cover slip for immunofluorescence study . Few hours later , when cells had attached , they were infected . At 18 hpi samples were fixed with 4% paraformaldehyde and stained . Alternatively , cells were incubated for 72 h and subsequently stained for GFP as described above to score infection [18] . For infection assay using 293TT , cells were harvested , reseeded into 96 well plates and allowed to attach . Few hours later , they were infected and scored at 72 hpi by counting GFP positive cells . HaCaT cells were grown on cover slips till ∼50% confluency and infected with HPV16 pseudovirus in presence of NIM811 , antibody , or DMSO . At the indicated times post infection , cells were washed with PBS and fixed with 4% paraformaldehyde for 15 min at room temperature , washed , permeabilized with 0 . 2% Triton X-100 in PBS for 2 min , washed , and blocked with 5% goat serum in PBS for 30 min , followed by a 1 h incubation with primary antibodies at 37°C . After extensive washing , cells were incubated with AlexaFluor-tagged secondary antibodies and fluorescently labeled phalloidin for 1 h . After extensive washing with PBS , cells were mounted in ‘Gold Antifade’ containing Dapi ( Invitrogen ) . Images were captured by confocal microscopy ( Zeiss 510 Laser Scanning Confocal Microscope operated by LaserSharp2000 software ) or by standard fluorescence microscopy ( Leica DMBI 6000 microscope ) . Within individual experiments the same microscope settings and exposure times were used . For quantification of fluorescent signal intensity , the LSM server software provided with the confocal microscope was used . Signal strength was acquired from randomly selected single cells ( n>15 for each group ) . The average region of interest was not significantly different among all groups . Background was determined using mock infected cells and subtracted prior to calculations . RG-1 staining was performed as described [24] . In brief , infected HaCaT cells were shifted to 4°C and incubated with RG-1 and K75 for 1 h in presence of 2% normal goat serum . After extensive washing and incubation with fluorescently labeled secondary antisera , cells were fixed for 20 min in 2% paraformaldehyde . After washing , cells were incubated for 5 min with phalloidin-AF647 conjugate and mounted . CyPA: NM_021130; CyPB: NM_000942; codon optimized HPV16 L1: AJ313179; codon optimized HPV16 L2: AJ313180 | Human papillomaviruses ( HPV ) , especially HPV types 16 and 18 , are a major cause of cancer in women worldwide . HPV16 , like most genital HPV types , relies on heparan sulfate proteoglycans ( HSPGs ) to attach to host cells and to the extracellular matrix . Attachment is mediated by surface-exposed basic residues of the major capsid protein , L1 . This triggers conformational changes affecting L1 and the minor capsid protein , L2 . However , it is not known what interaction triggers these structural changes and if any host cell protein is involved . Now we have identified a host cell chaperone , Cyclophilin B ( CyPB ) , as essential for efficient HPV16 and HPV18 infection . CyPB , which is present on the cell surface in association with specific forms of O-sulfated HSPG as well as in the lumen of intracellular membrane structures , is an energy-independent enzyme , which catalyzes cis/trans isomerization of peptidyl-prolyl bonds . We demonstrate that CyPB facilitates conformational changes resulting in exposure of the L2 N-terminus , which is required for infectious entry . In addition , we present some evidence suggesting that members of the cyclophilin family are required for a second , probably intracellular , step of HPV16 infection . This is the first report implicating cell surface chaperones as essential host factors for viral infection . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"virology/host",
"invasion",
"and",
"cell",
"entry",
"virology"
] | 2009 | Target Cell Cyclophilins Facilitate Human Papillomavirus Type 16 Infection |
Toxicity is an important factor in failed drug development , and its efficient identification and prediction is a major challenge in drug discovery . We have explored the potential of microscopy images of fluorescently labeled nuclei for the prediction of toxicity based on nucleus pattern recognition . Deep learning algorithms obtain abstract representations of images through an automated process , allowing them to efficiently classify complex patterns , and have become the state-of-the art in machine learning for computer vision . Here , deep convolutional neural networks ( CNN ) were trained to predict toxicity from images of DAPI-stained cells pre-treated with a set of drugs with differing toxicity mechanisms . Different cropping strategies were used for training CNN models , the nuclei-cropping-based Tox_CNN model outperformed other models classifying cells according to health status . Tox_CNN allowed automated extraction of feature maps that clustered compounds according to mechanism of action . Moreover , fully automated region-based CNNs ( RCNN ) were implemented to detect and classify nuclei , providing per-cell toxicity prediction from raw screening images . We validated both Tox_ ( R ) CNN models for detection of pre-lethal toxicity from nuclei images , which proved to be more sensitive and have broader specificity than established toxicity readouts . These models predicted toxicity of drugs with mechanisms of action other than those they had been trained for and were successfully transferred to other cell assays . The Tox_ ( R ) CNN models thus provide robust , sensitive , and cost-effective tools for in vitro screening of drug-induced toxicity . These models can be adopted for compound prioritization in drug screening campaigns , and could thereby increase the efficiency of drug discovery .
Toxicity is a major cause of failure in drug development and causes costly withdrawals of drugs from the market . Drug development productivity would be greatly improved if cytotoxic compounds were identified during early in vitro screening [1–4] . Drug-induced cytotoxic effects lead to changes in cell and nuclear morphology which are characteristic of the specific cell-death pathway involved , the best characterized being apoptosis , necrosis , and autophagy [5–7] . The field has advanced with the establishment of high content screening ( HCS ) techniques and the emergence of toxicity reporters revealing specific biochemical pathways triggered during cell death programs or measuring metabolic cell function [8–10] . However , toxicity reporters are often limited to assess the specific biochemical pathways for which they were designed [11–13] , and they are thus unlikely to capture the wide variety of toxic effects that can be triggered by different drugs in screening campaigns . Toxicity-screening approaches have combined multi-parametric image analysis of fluorescently labeled nuclei with the use of toxicity reporters in advanced machine learning pipelines [14–18] . However , toxicity reporters increase experimental complexity , thus reducing throughput and increasing screening costs . There is therefore an urgent need to develop broad specificity , cost-effective in vitro toxicity assays for incorporation in the primary screening phases of drug development . Cytotoxic effects have classically been visually identified from cell and nuclear morphology [5–7] . However , the complexity and variability of toxicity-associated morphological patterns has so far hindered their systematic and quantitative analysis and thus prevented their use as standalone toxicity screening endpoints . Although nuclear fluorescence staining forms the basis of most high content cell-based assays , its use is normally limited to image segmentation and nuclei counting to score cell-loss due to lethal toxicity [19 , 20] , thus disregarding the wealth of information contained in images of fluorescently labeled nuclei . In an effort to exploit this information for the quantification of pre-lethal toxicity , we have explored state-of the-art machine learning tools for automated pattern recognition . The success of classical learning-based computer vision methods relies heavily on extraction and selection of a reasonable set of relevant features that are highly discriminative of the phenotypes being studied . Feature extraction and selection requires in-depth knowledge of the phenotype under study , which is hindered in the current application by the complexity and great variety of drug-induced toxicity-associated nuclei organization patterns . The most recent major advance in machine learning is deep convolutional neural networks ( CNN ) which , similar to a brain , have multiple layers of interconnected artificial neurons [21 , 22] . Through an automated process , deep neural networks learn abstract representations of raw images from pixel information as a progressive hierarchy of sub-images , from which they extract features that can be used to classify complex patterns in a supervised manner . CNNs can thus automate the critical steps of feature extraction and selection by learning to extract high-level features based on spatial relationships , which has enabled them to outperform other machine learning methods in computer vision tasks , as demonstrated for several challenging biomedical applications [23–28] . Thus , deep technology seemed well suited to the analysis and prediction of drug toxicity in images of fluorescently labeled nuclei . Here , we present novel deep-learning approaches for in vitro cell-based toxicity assessment . The Tox_ ( R ) CNN approaches proved to efficiently predict a broad spectrum of toxicity mechanisms from different drugs , nuclear stains and cell lines . The main strength of these tools is their unique ability to predict toxicity based exclusively on nuclei staining; this offers the advantage of improved affordability and applicability of toxicity prediction . The attraction of the Tox_ ( R ) CNN tools relies on their high potential to enable sensitive and efficient compound prioritization based on detection of pre-lethal toxicity in primary screening campaigns .
To test if CNNs can predict cell toxicity based exclusively on nuclear staining , we designed an experimental assay in which HL1 cells were treated with reference compounds at different concentrations . The included reference compounds cover a range of cytotoxic effects: the DNA targeting genotoxic drugs cyclophosphamide ( Ciclo ) , 5-fluorouracil ( 5Fluo ) , and doxorubicin ( Doxo ) ; the apoptosis-triggering drug staurosporine ( Staur ) ; the enzyme inhibitors acetaminophen ( Aceta ) and sunitinib ( Sunit ) , which are not associated with a specific toxicity mechanism; the uncoupler of mitochondrial oxidative metabolism FCCP; and the microtubule stabilizer Taxol , which inhibits mitosis . To guarantee varying degrees of toxicity outcome we used previously empirically established dose-curves by selecting concentrations of drugs ranging between having no effect up to significantly reducing cell number due to moderate cell death . Cells were labeled with the DNA-specific fluorescent probe DAPI and imaged with an automated confocal microscope , revealing a great variety of nuclear patterns induced by the different compounds ( Fig 1A ) . The standard toxicity readout of nucleus count ( Num Nuc ) revealed cell loss due to drug-induced lethal effects , but did not reveal the great variety of toxic effects associated with the reference compounds assayed ( Fig 1B ) . As reference toxicity readouts , we analyzed Caspase 3/7 nuclear translocation and Mitotracker cytoplasmic intensity ( Fig 1C and 1D ) , both of which evidenced dose-dependent toxic effects promoted by the compounds tested . To assess toxicity independently of cell density and to enable detection of pre-lethal toxicity , we implemented a deep CNN architecture for the estimation of cell health status from microscopy images of DAPI-stained nuclei ( Fig 1E ) . Since we are aiming at a cell-based toxicity assessment , we used standard image analysis procedures to segment nuclei and cytoplasm according to the DAPI signal and used the segmentation to crop images ( see Materials and methods ) . Different image cropping strategies were designed based on the regions of interest included in the resulting image crop; nuclei ( Nuc ) , nuclei and cytoplasm ( Cell ) , nuclei and 3 adjacent pixels ( Nuc_Ring ) , and nuclei , cytoplasm , and background ( CNN All ) ( Fig 1F ) . Cropped images , each one containing a mass-centered cell , were used to train independent CNN models for making toxicity prediction ( CNN Nuc , CNN Cell , CNN Nuc_Ring and CNN All ) . An additional model was trained using the combination of images obtained using the different cropping strategies ( CNN 4crops ) . We repeated the training of the models ( CNN Nuc , Nuc_Ring , Cell , All , and 4crops ) 5 times each , in order to increase confidence in model comparison results and evaluate reproducibility . CNN models deliver a “health status” score as output for each cell , which determines a binary classification: healthy or toxicity affected . As a supervised learning technique , the CNN required images labeled according to the expected output ( healthy or toxicity affected ) as ground-truth images for training . However , the reference standards analyzed for this purpose , Caspase 3/7 and Mitotracker , did not qualify as general toxicity labels because neither efficiently captured all the drug-induced cytotoxic effects included in our experimental assay: both yielded poor resolution of Taxol-induced cytotoxicity and of the kinetic effects produced by Doxo and Aceta . As an alternative strategy , we produced a training dataset by labeling image crops according to the treatment exposure; cells from untreated wells were labeled healthy , whereas those from wells treated with the highest drug concentrations were labeled toxicity affected ( S1A Fig ) . The percentage of cells classified as healthy by the different CNN models served as the per-well measurement of “general” toxicity . In spite of the high degree of uncertainty introduced into the impure training set , the CNN healthy predictions efficiently revealed dose-response toxicity curves for all the drugs tested in the assay ( Fig 1G and S1B–S1E Fig ) , providing better resolution than those obtained from Num Nuc or the Caspase or Mitotracker readouts ( Fig 1B–1D ) . At high doses of several toxicants , CNN models trained with information from nuclei only ( CNN Nuc ) underwent an unexpected drop in toxicity prediction . This was not observed in CNN models trained with cytoplasmic information ( CNN Cell/Nuc_Ring/All/4crops ) , which displayed a steady increase in toxicity prediction due to their enhanced ability to reveal DNA release into the cytoplasm during necrosis induced by high drug concentrations . As screening readout , Z-scores were obtained from the percentage of cells classified as healthy by the CNN model for each compound-treated test well by normalizing to the DMSO-treated control cells ( DMSO ) . Untreated cells were not used as negative controls since these are not commonly included as controls in screenings . To allow comparative evaluation of the different toxicity readouts and benchmarking the different cropping strategies , readouts are displayed so that Z-score positive values depict toxic effects . CNN readouts were more efficient at predicting toxicity than the classical Num Nuc readout ( Fig 1H ) , with CNN Nuc displaying the highest Z-scores . Model performance was assessed on cells treated with the apoptosis-inducer staurosporine , which allowed the use of Caspase readout for ground truth labelling ( i . e . , Caspase-negative cells from untreated wells were labelled healthy whereas Caspase-positive cells from staurosporine-treated wells were labeled as toxicity affected ) . This test set was used for computing Receiver Operating Characteristic ( ROC ) curves to evaluate model prediction accuracies using the Area Under the ROC Curve ( AUC ) measurement . Results show high performance ( above 0 . 9AUC ) for all cropping strategies ( Fig 1I ) , being the CNN Cell the most accurate . Importantly , CNN Nuc predictions were less correlated with the number of cells in the field than the other cropping strategies ( Fig 1J ) , which displayed non-negligible correlations , suggesting that predictions of the health status of a single cell are influenced by surrounding information , including the presence of neighboring cells . As consequence , CNN Nuc reveals to be more independent of experimental errors due to cell density inconsistencies and also more indicated to detect pre-lethal toxicity . Accordingly , the CNN Nuc model proved to be the best readout for early prediction of toxicity , since it outperformed other cropping strategies at sub-lethal drug concentrations , where there is no reduction in Num Nuc indicating significant cell loss . Consistent with this pattern , plotting treatments by Z-scores to reveal above-threshold “toxic hits” confirmed CNN Nuc to be the most sensitive method for detecting early toxicity yielding 100 toxic hits ( rounded mean of results from the 5 CNN Nuc models ) out of 184 treated wells ( Fig 2A ) , thus outperforming other CNN models . CNN Cell , Nuc_Ring , All and 4crops readouts yielded 82 , 74 , 77 and 83 hits , respectively ( Fig 2B and 2C and S1A and S1B Fig ) . The standard toxicity readouts Num Nuc , Caspase and Mitotracker detected 27 , 81 , and 6 toxic hits , respectively ( Fig 2D–2F ) . All CNN models yielded significant Z-scores for Taxol at 0 . 1μM , demonstrating broader application than established readouts; Caspase detected toxicity effects only at 2μM Taxol , whereas Mitotracker and Num Nuc did not detect significant Taxol toxicity . The higher sensitivity of CNN Nuc compared with the other CNN models and established readouts was further evidenced by the lower half-maximal toxic concentration values ( EC50 ) obtained for the mild toxicant 5Fluo ( Fig 2 and S2 Fig ) . Based on these results , nuclear crops were used as inputs for CNN models in all subsequent studies , and from here on CNN refers to these CNN Nuc models . To avoid relying on external image segmentation and cropping procedures while providing a per-cell toxicity prediction , we undertook an alternative cell-based deep-learning approach incorporating automated object detection . An RCNN was implemented for the automated localization and classification of individual nuclei using raw images as input , instead of crops . The framework , based on the Faster RCNN algorithm [29] , includes a region proposal network ( RPN ) that uses features extracted from the last convolutional layer of a CNN to detect bounding boxes around individual candidate cell , which are then classified as healthy , toxicity-affected , or background ( Fig 3A and 3B ) . We trained the RCNN model with cell bounding box coordinates obtained with the standard segmentation procedure used in the CNN approach ( see Materials and methods ) . A set of 7 independent experiments in HL1 cells treated with the eight reference drugs , including two experiments in which untreated cells were cultured at different confluencies , was used for in-parallel training of the RCNN and CNN mixed models for toxicity prediction using balanced datasets of healthy and toxicity affected labeled nuclei ( Tox_CNN and Tox_RCNN_balanced ) . Resulting CNN predictions showed that the Tox_RCNN_balanced model , unlike Tox_CNN , erroneously predicted toxicity of untreated cells grown at low densities ( Fig 3C ) . To prevent the models from learning to predict toxicity as a reduction in cell number due to drug-induced lethal effects , we trained an additional mixed model that included extra images of untreated cells cultured at different densities ( 120 extra healthy training wells ) , hereafter referred to as Tox_RCNN . The unbalanced and balanced Tox_RCNN models ( Num Nuc RCNN and Num Nuc RCNN_balanced ) detected a similar number of objects ( nuclei ) ; moreover , the number was consistent with the number obtained by the standard segmentation procedure ( Num Nuc ) at regular cell densities ( Fig 3C–3E ) , demonstrating the efficiency of automated detection by RCNN . Tox_RCNN slightly overestimated cell number at low confluencies , probably due to the recognition of cellular debris that are discarded by the image processing procedure . Both Tox_CNN and Tox_RCNN mixed models successfully classified untreated cells at very low densities as healthy and efficiently predicted the toxic effects of drugs and high DMSO concentrations ( Fig 3C ) , further demonstrating their independence from cell-density fluctuations . Tox_ ( R ) CNN models performed efficiently in the test wells from an experiment used for training ( Fig 3D ) and in one independent experiment with HL1 cells at higher confluency ( S3A Fig ) . Overall , Tox_ ( R ) CNN models were consistently more sensitive than Num Nuc at predicting drug toxicity , with Tox_CNN outperforming Tox_RCNN classification in most cases . Even though these models were trained in HL1 cells , they suitably predicted toxicity from these drugs in two other cell lines , EAHY926 ( Fig 3E ) and MEVEC ( S3B Fig ) , thus confirming the applicability of Tox_ ( R ) CNN models for the prediction of toxicity in different cell types . In addition , these models successfully predicted toxicity of HL1 cells labelled with Hoechst 3342 ( S4 Fig ) . Together , these results demonstrate the robustness of deep-learning-based toxicity prediction with regard to inter-experimental and intra-experimental variability , thus confirming Tox_ ( R ) CNN as powerful screening tools . To demonstrate the value of these Tox_ ( R ) CNN models as tools for broad toxicity prediction , we performed a new screening in which HL1 cells were treated with a panel of 24 drugs , including those used to train the CNN model and additional drugs acting through several mechanisms . Tox_ ( R ) CNN mixed models sensitively predicted the outcome of toxic compound-treatments , thus proving their ability to reveal the toxicity of compounds for which they have not been trained ( Fig 4A ) . Conveniently , Tox_CNN enabled the automated extraction of features from pixel intensity values , which were used for unsupervised hierarchical clustering of compounds ( see Materials and methods ) . Tox_CNN features clustered compounds with known mechanisms of toxicity associated in a biologically meaningful manner , even though the models were not trained for this purpose ( Fig 4B ) . The ionophores FCCP and monensin , which produce ROS and mitochondrial toxicity clustered together . The DNA synthesis inhibitors 5Fluo , gemcitabine , and mitomicine also group in the same cluster . Apoptotic death inducers ( staurosporine , thapsigargin , bortezomib , and imatinib ) were clustered closely together with other drugs of unknown mechanism . The DNA intercalating anthracyclines epirubicin and doxorubicin and the topoisomerase II inhibitor Etoposide , all of which promote double strand breaks also clustered together . Other drugs included , such as microtubule modulators Taxol and Vinblastine did not cluster together . Statins ( lovastatin and simvastatin ) with yet unknown mechanism of toxicity , but expected to be similar because belonging to the same family of proteins , were clustered together . These findings confirm not only that deep CNN are able to perceive general toxic effects , but also that their ability to learn feature representations provides useful knowledge for comparing drug-induced mechanisms of toxicity . To further evaluate the Tox_ ( R ) CNN deep-learning models as screening tools for prioritizing compounds based on their toxicity potential , we re-analyzed a pre-accomplished HCS of primary pancreatic cancer associated fibroblasts ( pan-CAFs ) . Among several assay-specific labels , most of which are irrelevant here , this HCS included DAPI staining in the assay for both image segmentation and nuclei counting as a toxicity endpoint . A transfer learning strategy was applied to the Tox_ ( R ) CNN models delivering Tr_Tox_ ( R ) CNN models . Training strategy was designed to allow the use of pre-run screens lacking reference toxicity-inducing treatments ( see Materials and methods ) . In brief , the training dataset was produced from images from drug-treated wells with a significantly reduced cell number , which were labeled toxicity affected , while cells from DMSO treated wells were labeled healthy , since no untreated cells were available for training . Interestingly , compounds #19 and #33 ( anagrelide and quercetin ) were negligibly lethal according to nuclei counting , but were predicted by Tox_CNN to be toxic at high concentrations ( Fig 5A and 5B ) , demonstrating the greater sensitivity of deep-learning-based prediction compared with Num Nuc . Toxicity of these compounds was confirmed by re-testing in primary cardiac fibroblasts ( S5 Fig ) . Even though Tr_Tox_CNN performed better that the Tox_CNN model , the latter was still a more sensitive predictor of toxic effects than nuclei counting , demonstrating the value of these tools . Nuclei counting by both the transferred and original Tox_RCNN models ( Tr_Tox_RCNN and Tox RCNN ) was consistent with the standard procedure ( Num Nuc ) , revealing that object detection was performing adequately . However , the original Tox_RCNN model was a poor toxicity predictor in pan-CAFs compared with the transferred RCNN model ( Tr_Tox_RCNN ) , further evidencing the need for a transfer-learning approach for RCNN toxicity predictions in cell lines different from those used for training . The screening comprised 60 drugs at 8 concentrations ( 480 treatments ) and yielded 102 toxicity hits ( mean Z-score > 3 ) based on nuclei counting ( Fig 5C ) . In contrast Tr_Tox_CNN and Tr_Tox_RCNN identified 127 and 126 toxic wells , respectively ( Fig 5D and 5E ) . Z-scores computed for the transferred Tr_ ( R ) CNN models were plotted independently for all compounds screened ( S6 Fig ) . These results further demonstrate the superior performance and sensitivity of deep Tox_ ( R ) CNN toxicity predictions over classical screening endpoints based on nuclei staining .
There is a need to incorporate highly predictive toxicity assays into primary in vitro high-throughput screening in order to reduce the attrition of drug candidates at later phases of the drug discovery pipeline . In vitro cytotoxicity assessment is normally limited to measuring the number of viable cells per well . However , single-cell readouts make better outputs that avoid sources of experimental error such as non-homogeneities in cell-dispensing , drug-induced proliferative effects , and heterogeneous responses of different cell sub-populations , which could be misinterpreted if only well averages are examined . The toxicity research field has therefore been directed towards finding novel cell labels and readouts that distinguish between different cytotoxicity mechanisms [8–13] . Nevertheless , the use of toxicity reporters has not gained broad acceptance because it adds experimental complexity , thus reducing throughput and increasing screening costs . Here , we have established tools that predict cell toxicity based on the analysis of fluorescently labeled nuclei . These tools outperformed outputs that rely on toxicity reporters or cell counting . Since nuclei staining is common to most high content cell based assays , this tool has broad applicability for toxicity prediction in HCS , even in pre-accomplished screens , as demonstrated here . Over recent years , deep learning approaches have been successfully deployed in computer vision tasks and constitute the state-of-the-art tools for supervised machine learning . Several key features give deep learning approaches an advantage over other machine learning methodologies for toxicity analysis . First , by learning to represent data with multiple levels of abstraction in an unsupervised manner ( i . e . without human-based programming ) it avoids cumbersome knowledge-based feature engineering . This makes deep learning approaches independent of prior in depth knowledge of the target phenotypes , and therefore more suitable for broad toxicity prediction of multiple drugs with differing mechanisms by HCS . Second , their ability to learn intricate patterns improves recognition , feature extraction , and classification from noisy images , providing accuracy . This has led to deep technology outperforming other machine learning methods and human-based analysis , as demonstrated for several challenging biomedical applications [23–28] . Third , using transfer learning methods , networks can be updated to classify new datasets with limited training data [26 , 27 , 30] , and thus makes them suitable for predicting toxicity in pre-accomplished screenings . Accordingly , the deep-learning approaches presented here successfully predict toxicity of a broad spectrum of toxicity-associated patterns in HCS images of fluorescently labeled nuclei , proving to be more sensitive than established toxicity reporters and readouts for the recognition of pre-lethal toxicity . The Tox_ ( R ) CNN models suitably predicted toxicity in cell lines , nuclear stains and for compounds different from the ones used for training . Moreover , for very different cell lines , pre-trained Tox_ ( R ) CNN models can be subject to a simple transfer-learning approach that does not require toxicity controls , increasing the performance of toxicity predictions; this was successfully achieved here in the pre-accomplished screening with primary pancreatic cancer associated fibroblasts ( panCAF ) . Although CNN models were more sensitive and transferable and performed better , the use of RCNN models is justified by their independence from external segmentation and image cropping . We also demonstrate the utility of CNNs for extracting knowledge ( as feature maps ) that allows comparative analysis of different drugs in a screen . Previous cell-based toxicity screening approaches have combined multi-parametric image analysis of fluorescently labeled nuclei with the use of toxicity reporters in advanced machine learning pipelines [14–18] . The main strength of the tools presented here is their unique ability to predict toxicity based exclusively on nuclei staining , which offers the advantage of improving affordability and applicability of toxicity prediction . HCS is now an established tool for phenotypic drug discovery; in this setting , the deep-learning approaches presented here will promote a better use of HCS technology for toxicity assessment . The relevance of the deep learning approaches presented here relies on their high potential to enable sensitive and efficient compound prioritization based on detection of pre-lethal toxicity . They thus provide affordable cytotoxicity counter-screens for high throughput primary campaigns and could allow academic screening centers and pharma companies to discard cytotoxic compounds during primary screening and hit-to-lead drug development campaigns , thereby increasing the efficiency of drug discovery .
Mouse cardiac muscle HL1 cells purchased from Merck Millipore were grown on fibronectin ( 25μg/ml ) /gelatin ( 1mg/ml ) coated dishes with 10% fetal bobine serum ( FBS ) in Claycomb medium ( Sigma-Aldrich ) . Mouse embryonic ventricular endocardial cells ( MEVEC ) were kindly provided by Dr . de la Pompa [31] and cultured on 0 . 1% gelatin coated flasks in 10% FBS supplemented DMEM . EAHY926 cells were kindly provided by Dr . Edgell and maintained in 10% FBS supplemented DMEM . PanCAF were obtained from Dr . Hidalgo and cultured in RPMI with 20% FBS . Primary pig cardiac fibroblasts were isolated from fresh surgical samples by collagenase tissue digestion as described in [32] . Fibroblasts were grown to 80% confluency in flasks containing supplemented DMEM . All culture media were supplemented with 10% FBS ( except PanCAFs , which had 20% FBS ) , 100 U/ml penicillin , 100 μg/ml streptomycin and 2 mM L-glutamine and refreshed every 2–3 days . Mycoplasma tests were performed bimonthly for all cell line cultures . Fluorescence staining reagents including DAPI ( 4′ , 6-diamidine-2-fenilindol ) , Hoechst 33342 ( H42 ) , Mitotracker Orange and Cell Event 3/7 Caspase Green , were purchased from Invitrogen . Dimethyl sulfoxide vehicle ( DMSO ) ; Acetaminophen ( ACETA ) ; Doxorubicin ( DOXO ) ; carbonylcyanide p-trifluoromethoxyphenylhydrazone ( FCCP ) ; Sunitinib; Staurosporine ( STAUR ) ; Paclitaxel ( TAXOL ) ; Imatinib; Thapsigargin; Gemcitabine; Quercetin; Atenolol; Simvastatin; Genistein; Vinblastine; Monensin; Anagrelide; Epirubicin; Etoposide and Lovastatin were from Sigma . Ciclophosphamide ( CICLO ) , 5-Fluorouracil ( 5FLUO ) and Sunitinib malate ( SUNIT ) were from Tocris . Indacaterol and Bortezomib were kinldy provided by Dr . Blanco , Experimental Therapeutics Programme at CNIO . Cells were seeded on 384-well plates ( 5000 cells/well , otherwise specified ) , after 24h compounds were added to wells at N serial concentrations in 4 replicate wells . Plate was designed to avoid well distribution effects by using layout shown in S1A Fig , which places negative controls oriented asimetrically in distant positions within the plate ( usually center and right ) . Additionally , in experiments designed for training deep learning models , double of wells with extreme doses were included in the plate simetrically distributed at top and bottom locations . To avoid evaporation-related edge effects external rows were filled with PBS and not used for the assay . Compounds were dissolved in DMSO ( final concentration of 1% across the entire assay , otherwise specified ) . Cells were maintained in culture with compounds for 24h , and then stained for Caspase 3/7 and/or MitoTracker prior fixation with 4% PFA for 10 min at RT and final nuclei staining . Imaging of fluorescently labelled nuclei and toxicity probes was performed with an Opera automated confocal microscope ( Perkin Elmer ) fitted with an NA = 0 . 7 water immersion objective at a magnification of 20× . S1 Table summarizes all experiments used in this work , including information about cell lines , stainings , treatments , and number of images and cells . Image processing algorithm was developed using Definiens Developer version XD2 . 4 ( Definiens AG , Germany ) . Nuclei and cytoplasmic regions were first segmented based on differential contrast of DAPI/Hoechst 33342 intensity . Cells with a nuclear size bellow 0 . 3 times or over twice the mean of nuclear sizes per field were considered as debris and reclassified as non-cellular regions . An ID was univocally assigned to each cell the number of cells per well was computed . Toxicity readouts based on fluorescent reporters were extracted as Caspase 3/7 nucleus:cytoplasm intensity ratio ( Casp nuc/cyto ) , and Mitotracker mean intensity in the cytoplasmic region ( Mito ) , where specified . Images from DAPI/Hoechst 33342 stained nuclei were cropped and saved individually assuring one cropped image per single mass-centered cell , thus conserving univocal cell ID . Four different strategies of cropping images were established ( Fig 1F ) : The four strategies were set to a fixed size of 50x50 pixel crops , thus guaranteeing a proper inclusion of complete nuclei based on nuclear sizes and image pixel sizes . To avoid off-centered nuclei crops , those nuclei with a distance to field image boundaries of less than 50 pixels were excluded from the cropping extraction and further analysis . Additionally , bounding box coordinates from segmented cells were also extracted for training automatic object detection by RCNN . We designed the toxicity convolutional neural network with an architecture and parameters adapted from a state-of-the-art CNN model , VGG [33] , with high performance in image classification and well-known layers , activations , and initializations , including several 3x3 convolutions before a max-pooling layer . However , the limited size of our image crops ( 50x50 instead of 224x224 used by VGG ) required a reduction in the depth of the network to prevent overfitting . Inspired by state-of-the-art architecture LeNet-5 [34] , which was designed to work with 32x32 images , we constrained the network to 2 convolutional + max pooling groups of layers . Most of parameters were inherited from VGG model ( batch size , initial learning rate , weight initialization… ) , since the lack of a general reference standard did not allow us to perform systematic optimization of parameters . Number of epochs was set up to 120 since performance of the network reached stability ( hold-out validation using 33% of training data during 300 epochs with batch size of 256 ) while maintaining a reasonable computing time and preventing from potential overfitting . Kernel sizes of 3 and 5 were also tested reaching very similar results ( AUC values for caspase-based evaluation were 0 . 9513±0 . 0065 and 0 . 9509±0 . 0053 for CNN Nuc model with kernel sizes of 3 and 5 , respectively; p-val = 0 . 9 ) , so we fixed them to 3 such as in original VGG architecture , which also speeded up computations x1 . 63 . Final Tox_CNN network architecture comprises 4 convolutional and 2 fully connected layers ( Fig 1E ) to classify single-cell images as healthy or toxicity affected . The Rectified Linear Unit ( ReLU ) activation function is applied between each layer except the output dense layer , which uses a softmax activation function to provide a separate probability for each of the classes . We used ReLu as activation function since Sigmoid and Tanh can result in the so-called vanishing gradient problem . Convolutional layers convolve a 3×3 kernel over some input to produce 32 , 32 , 64 , and 64 feature maps , respectively . To reduce the number of features and the computational complexity of the network , we introduced two max-pooling layers with a window size of 2×2 after convolutional layers 2 and 4 . Additionally , to avoid overfitting , we included two dropout blocks after convolutional layers 2 and 4 ( probability of 25% ) , and another one next to the first fully connected layer ( probability of 50% ) . Dropout deactivates some neurons randomly with a specified probability during the weight update cycle . The final max-pooling layer is then flattened and followed by two densely connected layers with 512 and 2 features . Finally , we applied a softmax activation function to the output of last fully connected layer to calculate the probability for each class label . The total number of parameters to learn is equal to 4 , 031 , 458 , most of them belong to the first fully connected layer . We used ADADELTA algorithm [35] to adjust the learning rate automatically . To increase the number of data and avoid overfitting , we augmented images by applying random rotations in the range of [0° , 20°] , horizontal shifts in the range of [0 , 0 . 2 × Imagewidth] , vertical shifts in the range of [0 , 0 . 2 × Imageheight] , and horizontal/vertical flips , where Imagewidth and Imageheight show width and height of input images , respectively . By default , the modifications were applied randomly , so not every image will be changed every time . Images were normalized to zero mean and unit variance before feeding them into the network . We used state-of-the-art Faster Region-Based CNN [29] ( RCNN ) to both detect and classify cells as healthy or toxicity affected from entire images . Faster RCNN is composed of two modules: a Regional Proposal Network ( RPN ) and a RCNN network . RPN is a Fully Convolutional Network ( FCN ) [36] which proposes square regions within an image that may contain objects of interest without considering their classes while RCNN network classifies the object proposals from RPN into one of the classes ( or background ) , and refines the bounding boxes’ coordinates of the final proposals . We used the original Faster RCNN architecture [29] without any significant modifications except for the number of outputs for the classification and regression layers since we have two classes . Therefore , the classification layer has 18 ( 9×2 ) outputs , and the regression layer has 36 ( 9×4 ) outputs ( coordinates ) . Number of iterations was fixed to 750 , 000 . To train the network for single-cell toxicity prediction , a set of cells need to be labelled as healthy and toxicity affected . The uncertainty of the toxicity state of individual cells due to lack of bona-fide toxicity reporters hamper the possibility of creating a pure and clean training dataset . Therefore , cells were labeled according to the known or expected toxic response of different extreme treatments; a set of untreated cells ( or cells under harmless treatment ) were labelled as healthy , and a set of cells treated with known toxic compounds at the highest concentrations were labelled as toxicity affected . This labelling strategy minimize the amount of manual supervision needed to perform the tedious task of creating a large annotated dataset for training , avoiding also the use of any toxicity labels such as fluorescent reporters that always provide partial information since they are unable to detect all the toxic effects . These healthy and toxicity affected sets are initially conformed by all cells in a selection of wells correspondent to the appropriate extreme treatments . Training sets for both classes ( healthy and toxicity affected ) were balanced , where indicated , by removing all cells from randomly selected entire images . The final output is a field-based treatment-driven training dataset that represents two groups/classes with the ( expected ) highest mean difference in terms of toxicity state . A training-strategy avoiding overfitting was undertaken to properly deal with minimized , yet still present mislabeling of cells affected by toxicity in conditions with harmless or no treatment , as well as resilient cells in extreme harmful conditions . Training of both Tox_CNN and Tox_RCNN models need the outputs obtained from the image processing routine developed in Definiens ( see Image processing section ) . CNN approach used cropped single cell 50 × 50 pixel images; and RCNN approach used full fields ( 683 x 507 pixel images ) and bounding-box coordinates of cells . The initial experiment pursuing the comparison of CNN performance from the different cropping strategies used images from one experimental plate ( Experiment #1 ) , where the correspondent trainings were performed in parallel and repeated five times . The creation of central Tox_ ( R ) CNN mixed models involved images from 7 HL1 experiments ( Experiments #1–7 ) including several conditions: untreated cells , cells treated with DMSO , and cells treated with up to 8 drugs with known toxic effects at different concentrations . All drugs used demonstrated toxic effects in previous experiments where dose curves were fixed . Three of these experiments were designed in a way that allows the classifier to learn that healthy cells can also grow at low densities . Training dataset was created as detailed in the previous section , labelling single cells from wells without any treatment as healthy , and cells from treated wells with highest concentrations of the 8 available compounds as toxicity affected . In each plate , only half of these wells per condition were sparsely selected for training , and the rest were bound to test ( S1A Fig ) . In total , Tox_CNN mixed model was trained using 739 , 727 cells ( image crops ) covering 8 compounds ( toxicity affected ) and untreated cells ( healthy ) . Bounding-box coordinates of training cells , together with the correspondent 7 , 489 ( balanced ) and 10 , 883 ( unbalanced ) entire images ( fields ) were used to train the Tox_RCNN mixed models . S2 Table summarizes the number of instances ( crops or field images ) and experiments used for training each model . For CAF screening , we used deep transfer learning [37] to adapt Tox_ ( R ) CNN mixed models to a different cell line improving toxicity prediction . With this strategy , an existing pre-trained network is fine-tuned , avoiding training an entire neural network from scratch and reusing low level feature-detectors already learned . Therefore , we froze the weights of the first two convolutional layers in both ( R ) CNN approaches and retrained the rest of the layers to adapt to this new dataset . For this screening , since there is no prior information about the dose-response curve and expected toxicity , we used a different strategy to create the training set , but again following the guidelines detailed before ( see Label-free toxicity annotation of images ) . First , we selected drugs with at least one concentration with a significant toxic effect scored as a significant reduction of the number of cells per well ( Z-score>3 ) . Then , for each selected drug , cells treated with the two highest concentrations of that drug were included in the training set , and labelled as toxicity affected . Since untreated wells are not included in regular screenings , cells from half of the DMSO-treated wells were included in the healthy training set , which corresponds to the harmless condition in this specific screening . This resulted in a training set with 150 , 529 instances ( 6 , 057 field-images ) . S2 Table includes information about the number of instances ( crops or field images ) and experiments used to create the transferred models . We followed this general strategy for generating training sets , to ensure that it can be used in any assay where no prior information about toxic effects of compounds included in the screening is available . We used 25 epochs to re-train the Tox_CNN model , and 45 , 000 iterations to update the weights of the Tox_RCNN model; conforming the transferred models for toxicity prediction in PanCAF screening ( Tr_Tox_ ( R ) CNN models ) . We evaluated the performance of Tox_ ( R ) CNN models on several independent experiments . Cells out of the training set coming from experiments that were partially used for training were also employed for testing purposes: treatments with intermediate concentration of drugs were never included in training set , and not all wells from extreme treatments were selected for training . Tox_CNN model classified crop images at the input as healthy or toxicity affected based on the probabilistic scores obtained at the output for both classes . Tox_RCNN model return object detections which were classified as healthy , toxicity affected or background ( considered as non-cell detections and discarded from further analyses ) in entire field images . S3 Table summarizes experiments and number of instances ( crops or field images ) tested with the different models , and references to figures including the corresponding results . Well-based toxicity measurements were constructed from Tox_ ( R ) CNN predictions by computing the percentage of cells classified as healthy in each well . Standard toxicity measurements obtained from fluorescent reporters ( Caspase 3/7 and Mitotracker ) were aggregated in well-based values by computing the mean . Nuclei count obtained by image processing and RCNN-derived counting of nuclear detections were also reported for each well . For each type of toxicity measure ( x ) , we computed Z-scores by subtracting the mean ( μ ) and then dividing by the standard deviation ( σ ) of negative control ( DMSO treated wells ) : Z-Score=x-μσ Finally , we adjust the sign of the outputs to get increasing values for toxic effects , thus obtaining final well-based toxicity readouts that allow direct comparisons . The previously explained uncertainty of the toxicity status of individual cells under the wide range of different compound treatments hampers the proper evaluation of CNN models for parameter optimization and comparison of models obtained with different cropping strategies ( CNN Nuc , CNN Nuc_Ring , CNN Cell , CNN All , and the combination of all the previous , CNN 4crops ) . For that reason , assessment of model performance was tackled in different ways: Tox_CNN features from the first fully connected layer ( 512 features/sample ) were extracted for all cells in 4-plate screening of HL-1 cells with 24 drugs . Features obtained from cells that were either untreated , DMSO-treated , or treated with a non-extreme toxic concentration of the drugs ( 25 μM ) ( except for Taxol in which a 0 . 781 μM concentration was used ) were analyzed by PCA to select the 50 most informative features for further processing . Mean well-based features were later normalized ( Z-score with respect to untreated condition ) and aggregated in order to obtain mean values per treatment/condition . Finally , these feature vectors were assembled by hierarchical clustering [44] to generate the hierarchical tree ( clustergram function in MATLAB with Euclidean distance metric and weighted average linkage ) . Clusters along both dimensions ( features and treatment/condition ) were displayed in a heatmap of feature vector values including dendrograms representing the multilevel hierarchy obtained . Half maximal effective concentration ( EC50 ) is the concentration of a drug which induces a response halfway between the observed baseline and maximum effect of that drug . For the present work , the drug responses used for the EC50 calculations are the different toxicity effects , as indicated . EC50 were computed by fitting a Hill Equation sigmoid curve to the dose-response values and estimating the correspondent EC50 Hill Equation parameter [45] . Dose-response values and adjusted curves are displayed in Fig 2 and S2 Fig in such way that visual increasing values depict toxic effects for all toxicity measurements in order to allow proper comparison . Response axes are fixed to [1; 0] for CNN predictions , [4 , 000; 0] for nuclei count , [0 . 5; 3 . 5] for the Caspase ratio and [500; 100] for Mitotracker measurements . EC50 values were not calculated in cases where standard sigmoidal curve fitting was inappropriate within the existing range of drug doses ( Fig 2E and 2F ) . We used Keras [46] on the Theano backend [47] to develop the Tox_CNN model presented here . To construct the Tox_RCNN model we used a python implementation of Faster RCNN [29] which is developed on Caffe [48] . All experiments were run on a standard PC with an Intel Xeon CPU E5-2643 @ 3 . 30 GHz , 32 GB working memory , and using a 16 GB NVIDIA Quadro K4000 GPU to speed up the computations . Training time for the Tox_CNN model was 11 hours; transfer learning took 2 hours . Training time for the Tox_RCNN mixed model and transfer learning were 187 hours and 9 hours , respectively . Boxplots and well-based dot plots were created with NCSS Statistical Software ( version 11 ) , and mean Z-score plots were depicted with Python . MATLAB ( R2017a ) was used to perform hierarchical clustering and to compute dose-response curve adjustments and EC50 calculations . All figures displaying ( R ) CNN-derived information include only results from test sets; otherwise indicated . Figures showing information from initial CNN models associated with different cropping strategies show either the results from 5 independently trained models or the results of a representative model/cropping-strategy performing with the median AUC value ( out of five ) . | Visualization of nuclei using different microscopic approaches has for decades allowed the identification of cells undergoing cell death , based on changes in morphology , nuclear density , etc . However , this human-based visual analysis has not been translated into quantitative tools able to objectively measure cytotoxicity in drug-exposed cells . We asked ourselves if it would be possible to train machines to detect cytotoxicity from microscopy images of fluorescently stained nuclei , without using specific toxicity labeling . Deep learning is the most powerful supervised machine learning methodology available , with exceptional abilities to solve computer vision tasks , and was thus selected for the development of a toxicity quantification tool . Two convolutional neural networks ( CNN ) were developed to classify cells based on health status: Tox_CNN , relying on prior cell segmentation and cropping of nuclei images , and Tox_RCNN which carries out fully-automated cell detection and classification . Both Tox_ ( R ) CNN classification outputs provided sensitive screening readouts that detected pre-lethal toxicity and were validated for a broad array of toxicity pathways and cell assays . Tox_ ( R ) CNN approaches excel in affordability and applicability to other in vitro toxicity readouts and constitute a robust screening tool for drug discovery . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"fluorescence",
"imaging",
"medicine",
"and",
"health",
"sciences",
"nuclear",
"staining",
"pathology",
"and",
"laboratory",
"medicine",
"engineering",
"and",
"technology",
"signal",
"processing",
"drug",
"screening",
"toxicology",
"toxicity",
"pharmaceutics",
"crops",
... | 2018 | Tox_(R)CNN: Deep learning-based nuclei profiling tool for drug toxicity screening |
The development of new adjuvants enables fine modulation of the elicited immune responses . Ideally , the use of one or more adjuvants should result in the induction of a protective immune response against the specific pathogen . We have evaluated the immune response and protection against Trypanosoma cruzi infection in mice vaccinated with recombinant Tc52 or its N- and C-terminal domains ( NTc52 and CTc52 ) adjuvanted either with the STING ( Stimulator of Interferon Genes ) agonist cyclic di-AMP ( c-di-AMP ) , a pegylated derivative of α-galactosylceramide ( αGC-PEG ) , or oligodeoxynucleotides containing unmethylated CpG motifs ( ODN-CpG ) . All groups immunized with the recombinant proteins plus adjuvant: Tc52+c-di-AMP , NTc52+c-di-AMP , CTc52+c-di-AMP , NTc52+c-di-AMP+αGC-PEG , NTc52+CpG , developed significantly higher anti-Tc52 IgG titers than controls . Groups immunized with c-di-AMP and Tc52 , NTc52 or CTc52 showed the highest Tc52-specific IgA titers in nasal lavages . All groups immunized with the recombinant proteins plus adjuvant developed a strong specific cellular immune response in splenocytes and lymph node cells with significant differences for groups immunized with c-di-AMP and Tc52 , NTc52 or CTc52 . These groups also showed high levels of Tc52-specific IL-17 and IFN-γ producing cells , while NTc52+CpG group only showed significant difference with control in IFN-γ producing cells . Groups immunized with c-di-AMP and Tc52 , NTc52 or CTc52 developed predominantly a Th17 and Th1immune response , whereas for NTc52+CpG it was a dominant Th1 response . It was previously described that αGC-PEG inhibits Th17 differentiation by activating NKT cells . Thus , in this work we have also included a group immunized with both adjuvants ( NTc52+c-di-AMP+αGC-PEG ) with the aim to modulate the Th17 response induced by c-di-AMP . This group showed a significant reduction in the number of Tc52-specific IL-17 producing splenocytes , as compared to the group NTc52+c-di-AMP , which has in turn correlated with a reduction in protection against infection . These results suggest that the Th17 immune response developed after immunizing with NTc52+c-di-AMP could have a protective role against T . cruzi infection . Groups NTc52+c-di-AMP , Tc52+c-di-AMP and NTc52PB , were the ones that showed better protection against infection with lower parasitemia and weight loss , and higher survival .
Recent estimations indicate that about 6–7 million people are infected worldwide with Trypanosoma cruzi [1] , a protozoan parasite that is the etiological agent of Chagas disease [2] . Vectorial transmission takes place in endemic countries where the vector , a triatomine insect , is present . In Latin America , one hundred million people are at risk of infection , and about 56 , 000 new infection cases and 12 , 000 deaths are registered annually [3] . Geographic distribution of Chagas disease spread in the last decades due to migration , with more than 300 , 000 people being infected in the United States , as indicated by CDC estimations [4] . T . cruzi new infections in North America , Europe and Asia are mainly the consequence of transfusion of contaminated blood , congenital transmission and organ transplantation . The infection has an initial acute stage followed by a chronic stage where up to 30% of patients develop cardiac alterations and 10% develop digestive , neurological or mixed alterations [1 , 5] . Currently , only two drugs are used for treatment: Nifurtimox and Benznidazole . Both are effective in the acute stage of infection , lose effectiveness in the advanced phase , and have important side effects associated to the treatment [5 , 6] . Therefore , efforts are focused not only in transmission control and the search for more efficient and less toxic drugs , but also in the development of prophylactic and therapeutic vaccines . In the last years , vaccine development was focused on some antigens that proved to be promising candidates employing different adjuvants , immunization protocols and strategies , including recombinant proteins , DNA vaccines and viral vectors ( reviewed in [7 , 8 , 9] ) . Tc52 is a T . cruzi protein with glutathione transferase activity [10] which has two domains with similar size: the N-terminal ( NTc52 ) and the C-terminal ( CTc52 ) domains [11] . Many immunomodulatory properties were described for Tc52 that make it an interesting target for vaccine development . It binds to macrophages and dendritic cells and , in the presence of IFN-γ , it induces iNOS expression and NO synthesis by macrophages [12] . Also , Tc52 inhibits the proliferation of splenocytes induced by mitogen , an activity for which probably the C-terminal domain is responsible [13 , 14] . Tc52 is essential for the parasite , since the double knock-out is lethal [15] . Thus , Tc52 seems to be a promising vaccine candidate against T . cruzi infection . Therefore , vaccine prototypes were assayed using this antigen [16 , 17] . We have also tested Tc52 as a vaccine candidate using different protocols . We demonstrated that NTc52 gives better protection than CTc52 and similar to full-length Tc52 when assayed as a DNA-delivery system carried by attenuated Salmonella [18] . We then evaluate the role of Tc52 DNA in a multi-component vaccine prototype with promising results [19] . We also developed a DNA-prime/protein-boost vaccine strategy focusing on NTc52 [20] . In this work , we look for a safe vaccine with higher acceptance . To this end , instead of a DNA-based vaccine using an attenuated bacterium as carrier , we use purified protein Tc52 produced in Pichia pastoris , an organism with GRAS ( Generally Recognized As Safe ) status conferred by the USA FDA [21 , 22] . Since infection by T . cruzi could take place by vector-mediated breaks in skin and mucosa , by transfusions and organ transplants , by congenital infection and by oral route through the gastrointestinal tract [23] , we analyzed vaccine candidates that induce both systemic and mucosal immunity . The novel mucosal adjuvant bis- ( 3’ , 5’ ) -cyclic dimeric adenosine monophosphate ( c-di-AMP ) is a bacterial second messenger , involved in sensing DNA integrity , control of cell wall and potassium homeostasis [24] . This compound promotes expression of type I IFN and TNF through the activation of a STING/TBK1 axis . The co-administration of an antigen and c-di-AMP by intranasal route in BALB/c mice induced both systemic and mucosal antibody responses , and a balanced Th1/Th2/Th17 cellular specific immunity pattern [25] . Another mucosal adjuvant , a pegylated derivate of the α-galactosyl-ceramide ( αGC-PEG ) , induces a dominant Th2 immune response [26] , and more interestingly , it inhibits Th17 cell differentiation by activating NKT cells [27] . The role of IL-17 and Th17 cells in T . cruzi infection are still not well characterized . Though , reports suggested that IL-17 could have a protective role in infection and tissue damage control [28–32] . In this work we use c-di-AMP as an adjuvant for intranasal immunization with the recombinant proteins Tc52 , NTc52 and CTc52 . Also , we test the co-administration of NTc52 together with c-di-AMP and αGC-PEG to evaluate the ability of αGC-PEG to block Th17 response and its effect in the conferred protection against T . cruzi infection .
T . cruzi epimastigotes ( RA strain ) were grown in LIT medium as previously described [18] . The T . cruzi bloodstream trypomastigotes ( RA strain ) and the recombinant Tulahuen strain expressing β-galactosidase ( Tul-β-Gal ) [33] were isolated from the blood of infected mice . Full-length Tc52 and its N-terminal domain ( NTc52 ) were cloned and expressed in the yeast P . pastoris using the pPICzα-A expression vector as previously described [20] . Tc52 C-terminal domain ( CTc52 ) , corresponding to the Tc52 amino acid residues 224 to 245 [11 , 18] , was similarly cloned and expressed . The DNA sequence encoding for CTc52 was amplified from T cruzi RA strain epimastigote genomic DNA and cloned in the pPICzα-A vector . For that purpose the following oligonucleotides were used: the forward primer 5’CGACTGGAATTCGCTCCTGGCTATGTACTTTTTGTT3’ containing an EcoRI restriction site ( underlined ) , and the reverse primer 5’ACTAGCGCGGCCGCTCAGTGATGGTGATGGTGATGCAATGACCATGTGACGTGC3’ , with a NotI restriction site and a sequence encoding a His6 tag ( both underlined ) . Cloning , clone selection , and protein expression and purification were performed as described for NTc52 and Tc52 [20] . Cloning and expression of NTc52 in pcDNA3 . 1 plasmid , and transformation of attenuated Salmonella enterica serovar Typhimurium aroA SL7207 , was performed as previously described [18] . For the immune response analysis , nine groups ( 6 animals/group ) of inbred female 6- to 8-week-old C3H/HeN ( H-2K haplotype ) mice were immunized 3 times every 10 days by intranasal ( i . n . ) route . Group I ( GI ) : Control ( received PBS alone ) ; GII: Tc52 , GIII: NTc52 , GIV: CTc52 , GV: Tc52+c-di-AMP , GVI: NTc52+c-di-AMP , GVII: CTc52+c-di-AMP , GVIII: NTc52+c-di-AMP+αGC-PEG , GIX: NTc52+ODN-CpG . Mice were immunized with the same number of molecules ( 0 . 42 nmol ) of each recombinant protein according to the predicted molecular weight: Tc52 ( 52 . 2 kDa ) 21 . 9 μg , NTc52 ( 26 . 9 kDa ) 11 . 3 μg , CTc52 ( 25 . 6 kDa ) 10 . 7 μg . The amounts of adjuvants used per mice were: c-di-AMP 5 μg , αGC-PEG 15 μg , and ODN-CpG 1826 20 μg . These amounts were selected according to previous results [20 , 25 , 28 , 34 , 35] . To analyze the protection conferred against T . cruzi challenge , seven groups of mice ( 5 animals/group ) were immunized 3 times every 10 days as follows: GI: control with c-di-AMP alone , GII: control with αGC-PEG alone , GIII: Tc52+c-di-AMP , GIV: NTc52+c-di-AMP , GV: CTc52+c-di-AMP , GVI: NTc52+c-di-AMP+αGC-PEG , GVII: NTc52+αGC-PEG , and GVIII: NTc52PB . We included the group GVIII as positive control , since it was our previous best prime-boost protocol using the antigen Tc52 [18 , 20] . Briefly , GVIII mice were immunized with 2 oral doses of 109 CFU of Salmonella SL7207 delivering the construction pcDNA3 . 1-NTc52 ( SNTc52 ) , followed by 2 doses of NTc52+ODN-CpG 1826 by intradermal route . To ensure that the last immunization was at the same time for all groups , the immunization protocol for GVIII started 10 days before the other groups . Twenty days after the last immunization , mice were infected by intraperitoneal route with 103 T . cruzi bloodstream trypomastigotes of the highly virulent RA strain . Experiments with animals were approved by the Review Board of Ethics of the Faculty of Pharmacy and Biochemistry ( UBA , Argentina ) and in agreement with the local government of Lower Saxony , Germany; and conducted in accordance with the guidelines established by the National Research Council [36] . Serum samples , nasal lavages and saliva samples were collected 20 days after the last immunization for the measurement of antigen-specific IgG and IgA , respectively . Measurement of antigen-specific IgG and IgA was carried out by Enzyme Linked Immune Assays ( ELISA ) as described previously [18] . Plates were coated with recombinant Tc52 ( rTc52 , 2 μg/ml ) . For the measurement of total IgA in nasal lavages , plates were coated with anti-IgA polyclonal antibody ( 2 μg/ml ) . IgA specific titers were normalized to 10 μg of total IgA/well . Proliferation of spleen and cervical lymph nodes cells after stimulation with antigen rTc52 was evaluated by 3H-thymidine incorporation as previously described [18] . The results were expressed as proliferation index ( PI ) , defined as the cpm ( counts per minute ) in the presence of the antigen/cpm in the absence of the antigen . The number of IFN-γ , IL-4 and IL-17 producing spleen cells in response to antigenic stimulus was determined by ELISPOT . Ninety-six well plates were sensitized with capture antibodies ( anti-IFN-γ , anti-IL-4 , or anti-IL-17 ) for 16 h , at 4°C , and then blocked ( with RPMI medium supplemented with 10% fetal bovine serum , FBS ) . Splenocytes ( 2 x 105 or 4 x 105 cells/well ) were added in presence or absence of Tc52 ( 0 . 5 μg/well ) , and incubated at 37°C , 5% CO2 , for 24 h for IFN-γ and 48 h for IL-4 and IL-17 . Plates were washed and then incubated ( 2h , 25°C ) with biotinylated detection antibody and then with Avidin-HRP . As peroxidase substrate , 3-amino-9-ethil-carbazole ( AEC ) was used . Plates were scanned and spots were quantified in an ImmunoSpot CTL . Results were expressed as number of spots for 106 cells . Twenty days after the last immunization , mice were infected by intraperitoneal route with 1000 T . cruzi bloodstream trypomastigotes of the highly virulent RA strain . Weight and parasitemia were measured every 2 or 3 days after the challenge as reported [18] . The change in weight was expressed as a percentage ( W% ) and calculated as follows: W% = ( Wi—Wo ) x 100%/Wo , where Wo is the weight of each mouse immediately before infection , and Wi is its weight at day i post-infection . Survival was monitored daily . Statistical analyses were carried out with Prism software version 5 . 0 ( GraphPad , San Diego , CA ) , and R [37]; using a nonparametric Kruskal-Wallis test and Dunn’s posttest . The survival curves were analyzed with a log rank Mantel-Cox test . All the comparisons were done in reference to control groups ( PBS , c-di-AMP or αGC-PEG ) , except when indicated . P values of less than 0 . 05 were considered significant .
The antigen specific antibody responses developed after immunization were analyzed by ELISA in sera of all mice immunized . Plates were sensitized with full-length Tc52 instead of the immunizing antigen ( Tc52 , NTc52 or CTc52 ) , to evaluate the elicited antibodies that recognize their cognate epitope in the molecule exposed by the parasite . Mice immunized with recombinant proteins ( Tc52 , NTc52 and CTc52 ) plus c-di-AMP , and those immunized with NTc52 plus c-di-AMP+αGC-PEG or ODN-CpG , developed high titers of Tc52-specific IgG antibodies with significant differences compared to the control group . Mice immunized just with the proteins in the absence of any adjuvant did not show antibodies titers different than mice receiving PBS ( Fig 1A ) . Our main objective by giving the immunogen by intranasal route was to generate mucosal immunity including antibodies able to block the entrance of parasites by this route . Groups immunized with Tc52 , NTc52 or CTc52 in the presence of c-di-AMP showed titers of Tc52-specific IgA in nasal lavages which were significantly higher than in controls ( Fig 1B ) . Even when both c-di-AMP and αGC-PEG are potent mucosal adjuvants [25 , 26 , 34] , group VIII ( NTc52+c-di-AMP+αGC-PEG ) developed lower specific IgA titers than mice in the group VI ( NTc52+c-di-AMP ) . Nevertheless , no significant difference between these groups was observed . Although the ability of ODN-CpG as mucosal adjuvant was widely demonstrated [38–40] , we found that group IX ( NTc52+CpG ) developed anti-Tc52 IgA titers higher than the control ( median 19 . 5 and 3 . 0 , respectively ) , but the difference was not significant . This result corroborates our previous report that ODN-CpG is not able to elicit mucosal IgA against Tc52 [20] . The Tc52-specific cellular immune response was studied ex vivo in splenocytes and lymph nodes cells by proliferation assays ( Fig 2 ) . Mice from groups V to IX developed proliferative responses upon in vitro Tc52 reestimulation in splenocytes and in lymph node cells . For lymph node cells ( Fig 2A ) differences to the control were significant for GV ( p < 0 . 01 ) , GVI ( p < 0 . 001 ) and GVII ( p < 0 . 01 ) , whereas only GV ( Tc52+c-di-AMP ) and GVI ( NTc52+c-di-AMP ) spleen cells showed significant differences with respect to the control group ( p < 0 . 01 ) ( Fig 2B ) . As expected , c-di-AMP induced an immune response activating Th1 , Th2 and Th17 cells [25] . On the other hand , αGC-PEG inhibits Th17 cell differentiation by activating NKT cells [26 , 27] . Thus , we co-administered the adjuvants c-di-AMP and αGC-PEG in order to selectively block with the second adjuvant the Th17 response stimulated by the first one . With the aim to evaluate the type of immune response developed by the different immunization protocols , we focused on three cytokines: IL-17 , IFN-γ and IL-4 . To this end , the number of cytokine-producing cells in the presence of rTc52 was determined by ELISPOT . We found that recombinant proteins adjuvanted with either ODN-CpG or c-di-AMP elicited cells able to strongly secrete IFN-γ after stimulation . These were significantly more abundant ( p<0 . 001 ) than in animals receiving the recombinant proteins without adjuvant ( Fig 3 ) . When IL-17 was analyzed , every group adjuvanted with c-di-AMP elicited cells that strongly secreted IL-17 upon restimulation with rTc52 . In contrast , the group adjuvanted with ODN-CpG was not able to induce IL-17-secreting cells . When the result of each cytokine was analyzed as percentage of the sum of the three cytokines in each group , we found that the profile of cytokine-producing cells was similar for groups immunized with c-di-AMP containing vaccines , with values between 35–45% for IFN-γ and 40–59% for IL-17 . In contrast , groups immunized with ODN-CpG containing vaccines induced predominantly IFN-γ with ~88% of the total and practically no secretion of IL-17 . These results highlight the significant influence of c-di-AMP on the stimulation of a mixed Th17- and Th1-oriented immune response . Another important result was observed in mice immunized with NTc52+c-di-AMP+αGC-PEG , where we confirm that IL-17 secretion was partially blocked by the use of αGC-PEG as co-adjuvant ( Fig 3B ) , when comparing to mice immunized with NTc52+c-diAMP ( p < 0 . 01 ) . The groups that were included in the immunoprotection study differ little from those used to assess the immune response . Since immunization with NTc52 and CTc52 recombinant proteins without adjuvant did not induce a significant immune response , these groups were excluded from the evaluation of protection against parasite infection . However , two additional groups were included: ( i ) the NTc52+αGC-PEG group , to evaluate the protection generated by the NTc52 protein plus the αGC-PEG adjuvant alone , which induces a Th2-dominated immune response [26] , and ( ii ) a positive control group primed orally with attenuated Salmonella delivering the construction pcDNA3 . 1-NTc52 ( SNTc52 ) , followed by a boost of 2 doses of NTc52+ODN-CpG 1826 by intradermal route . This group ( NTc52PB ) was included as gold standard because it provided the best protection in previous experiments [20] . All immunized mice developed less parasitemia than control groups ( Fig 4 ) . Fig 4A shows the strong protection induced by Tc52 and its domains when c-di-AMP is used as adjuvant . When the area under the curve ( AUC ) is analyzed , the full-length antigen and the N-terminal domain showed lower AUC with significant difference against control ( c-di-AMP ) , p > 0 . 01 , with non parametric test ( Fig 4C ) . AUC from group CTc52+c-di-AMP did not show significant difference against control . Fig 4B shows the comparison of the adjuvants and protocols when the antigen used is NTc52 , which strongly reduces parasitemia . The reduction was especially noticeable for groups NTc52+c-di-AMP ( GIV ) , Tc52+c-di-AMP ( GIII ) and NTc52PB ( GVIII ) , which showed AUCs 6 . 71 , 6 . 48 and 6 . 11 times lower than the control group , respectively . The protection in terms of parasitemia was also analyzed at each point with nonparametric statistical tests ( Fig 4D ) . All immunized groups except NTc52+αGC-PEG and NTc52+c-di-AMP+αGC-PEG present significant differences compared to the control in at least one point . Groups that show better protection were the same as when AUCs were analyzed: GIII , GIV and GVIII . Taken both analyses together , the groups that showed better protection were: NTc52+c-di-AMP , Tc52+c-di-AMP and NTc52PB . Also , it is notable that αGC-PEG is not a good adjuvant for developing vaccines against T . cruzi infection using NTc52 as antigen . Group NTc52+αGC-PEG showed an AUC 3 . 11 times lower than control , and no significant differences , whereas the other immunization protocols with NTc52 gave better protection . Importantly , immunization with NTc52 in the presence of both adjuvants ( αGC-PEG and c-di-AMP ) induces less protection than immunizing with only c-di-AMP , suggesting the potential relevance of IL-17 in the control of T . cruzi infection when c-di-AMP is used . To analyze the protection conferred by vaccination , we also follow up the animal body weight every 2–3 days after infection ( Fig 5A ) . All immunized groups showed less weight loss than the control group . Fig 5AI shows that c-di-AMP by itself is unable to reduce the weight loss induced by infection . This adjuvant plus Tc52 or CTc52 was able to reduce the weight loss only in a small proportion of the animals . However , the reduction of weight loss was really significant ( p < 0 . 01 ) when c-di-AMP was used in combination with NTc52 . Fig 5AII shows the comparison of the adjuvants and protocols when the antigen used is NTc52 . At 25 dpi , the difference was significant only for the groups NTc52+c-di-AMP and NTc52PB ( p < 0 . 05 ) ;the NTc52+c-di-AMP group being the one that showed less body weight loss , behaving closely to uninfected mice . Moreover , at 25 dpi , this group also showed significant difference in body weight change with respect to the group vaccinated with NTc52+c-di-AMP+αGC-PEG ( p < 0 . 05 ) , in which a strong weight loss was observed , even higher than in controls immunized with just adjuvant . Survival was monitored daily post infection . All mice from both control groups ( c-di-AMP and αGC-PEG ) died by 35 dpi , whereas the survival of the other groups was 100% for Tc52+c-di-AMP , 80% for NTc52+c-di-AMP and NTc52PB , 60% for CTc52+c-di-AMP , and 40% for NTc52+c-di-AMP+αGC-PEG ( Fig 5B ) . Group NTc52+αGC-PEG showed 80% survival until 45 dpi , after which it fell to 60% . Survival remains unchanged until 100 dpi , when mice were sacrificed . Survival was significantly different between the control group and the Tc52+c-di-AMP group ( p<0 . 05 ) , whereas no significant differences were seen between the control group and the NTc52+c-di-AMP or NTc52PB groups ( p = 0 . 051 ) .
In this research project , we focused on the development of an intranasal vaccine , seeking to stimulate both systemic and mucosae pathogen-specific protective immunity . Previously , we demonstrated that the N-terminal domain of Tc52 ( NTc52 ) is a promising vaccine candidate [18 , 20] . Here , we have focused on this subunit administered with c-di-AMP as adjuvant , but we also evaluated the protective role of the adjuvant with full-length Tc52 and CTc52 as recombinant proteins . Twenty days after the last immunization , all groups that received a recombinant protein plus any of the adjuvants ( c-di-AMP , c-di-AMP+αGC-PEG or ODN-CpG ) developed high anti-Tc52 IgG titers , which were significantly different than in controls , as expected according to previous reports [25 , 26 , 34 , 41] . Nevertheless , the results were different when we analyzed the mucosal immunity elicited . Only mice immunized with Tc52 , NTc52 or CTc52 co-administered with c-di-AMP developed high Tc52-specific IgA titers in nasal lavages , showing the superior ability of c-di-AMP to induce mucosal immunity [25 , 34] . The group immunized with NTc52+c-di-AMP+αGC-PEG developed high IgA specific titers , significantly different than the controls . These titers were lower than in the NTc52+c-di-AMP group , but without statistical significance . We find that the use of c-di-AMP and αGC-PEG together did not improve mucosal immunity . Group NTc52+CpG developed anti-Tc52 IgA titers higher than the controls , but without significant differences . That contrasts to other reports that used ODN-CpG as an adjuvant in mucosal vaccines [39 , 40] , but correlates to our previous results observed in mice immunized with NTc52+CpG [20] . We have also analyzed cellular immune responses ex vivo in splenocytes and lymph node cells . Cells from all groups immunized with Tc52 , NTc52 or CTc52 , adjuvanted with c-di-AMP , αGC-PEG+c-di-AMP or ODN-CpG , showed a proliferation index over 2 . However , only groups immunized with c-di-AMP ( Tc52+c-di-AMP , NTc52+c-di-AMP , CTc52+c-di-AMP ) presented significant differences with respect to controls . The ability of c-di-AMP as mucosal adjuvant to induce specific cellular immunity was already described [25 , 34] . Nevertheless , this is the first work in which immunization with a protein antigen plus c-di-AMP or ODN-CpG ( and also c-di-AMP+αGC-PEG ) are tested together . In our protocol , using the same antigen ( NTc52 ) , c-di-AMP induced stronger cellular and humoral immune responses than ODN-CpG , as demonstrated by antigen-specific IgG and IgA titers , as well as proliferation assays . The profile of the cellular immune response was analyzed in splenocytes by ELISPOT , studying IFN-γ , IL-4 and IL-17 producing cells . Immunizations with c-di-AMP showed a cytokine profile related to a mixed Th17+Th1 immune response for all the antigens studied . In contrast , in mice receiving ODN-CpG the immune response elicited was predominantly Th1 . The co-administration of c-di-AMP and αGC-PEG with NTc52 induced a cytokine-producing cells profile of IFN-γ > IL-17 > IL-4 . The number of IL-17 spot forming units ( SFU ) was significantly different for groups NTc52+c-di-AMP and NTc52+c-di-AMP+αGC-PEG , showing the ability of αGC-PEG to inhibit , at least partially , the Th17 immune response induced by c-di-AMP . Ebensen et al . [26] have previously described in BALB/c mice ( H-2d ) immunized by i . n . route with β-galactosidase + αGC-PEG , a cellular specific profile of IL-4 > IFN-γ ( IL-17 was not assayed ) . These differences might be explained not only by the use of a different antigen , but also by the differences in the mouse strain . It is well known that BALB/c mice develop predominantly a Th2 profile , whereas C3H/HeN ( H-2k ) mice present a more balanced Th1/Th2 immune response . Protection against infection was evaluated challenging the immunized mice with a lethal dose of the highly virulent T . cruzi RA strain . The challenge was conducted intraperitoneally as a heterologous route for a more rigorous test of vaccine efficacy . Immunized mice showed protection as compared with controls , in terms of parasitemia , weight loss and survival . However , the groups Tc52+c-di-AMP , NTc52+c-di-AMP and NTc52PB were the ones that showed higher protection with significant differences with respect to the control groups in almost all parameters tested . In this case , we were more strict and analyzed if the protection elicited by immunization with c-di-AMP was able to improve the previous reported by us performance in a heterologous DNA-prime/protein-boost protocol , NTc52PB [20] . We find that NTc52+c-di-AMP is the group that stands better in all parameters , particularly in reduction of parasitemia levels and reduction of weight loss , showing not only significant differences alongside controls ( p<0 . 01 ) , but also similar weight changes compared to uninfected mice , with almost no reduction in body weight at 25 dpi . For comparisons it must be beared in mind that the NTc52PB protocol not only includes 4 doses ( one more than the other groups ) , but also involves the immunization with DNA and protein , and by different routes ( i . e . a more cumbersome protocol associated with considerable production constraints and difficult implementation logistics ) . As was expected , immunization with NTc52+αGC-PEG induced partial protection against infection since αGC-PEG , like α-GC , activates NKT cells and leads to a Th2 biased immune response [26 , 27 , 42] , whereas protection against T . cruzi infection is provided by a Th1 immune response [43 , 44] . We found that the Tc52 N-terminal domain represents a more protective antigen , giving the same or even more protection than the full-length Tc52 and clearly more than the C-terminal domain . The pattern of cytokine-producing cells was similar for those 3 groups ( Tc52+c-di-AMP , NTc52+c-di-AMP and CTc52+c-di-AMP ) : predominantly IL-17 and IFN-γ . This shows the notable influence of this adjuvant in the type of immune response developed . IL-17 is a pro-inflammatory cytokine which has a protective role against infection by many pathogens , inducing neutrophil recruitment as well as secretion of pro-inflammatory soluble factors . The role of IL-17 and Th17 cells in T . cruzi infection is still not well characterized . Nevertheless , reports suggested that IL-17 , secreted by Th17 or other cells , could have a protective role in infection and tissue damage control . Da Matta Guedes et al . [28] have also described the protective role of IL-17 modulating Th1 differentiation in the heart , and thus , controlling myocarditis . In accordance with that , recent reports demonstrate a correlation between IL-17 levels and cardiac function in patients with Chagas disease [32] . Furthermore , IL-17-/- mice infected with T . cruzi showed lower survival and higher blood levels of parasite and aspartate aminotransferase ( AST , a marker of muscle damage ) , than wild type mice [29] . A recent study with knock out mice for the cytokine receptor IL-17 RA demonstrated the important role of this receptor in protection against T . cruzi infection , recruiting IL-10-producer neutrophils that could modulate the IFN-γ-dependent inflammatory response , and thus , tissue damage . In T . cruzi infection , IL-17 could be produced by many cell types , including NKT , Tγδ , CD4+ Th17 and CD8+cells [30] . More recently , B cells were identified as the principal source of IL-17 in T . cruzi infection [31] . In this work we have studied the role of different immune response patterns , developed by different immunization strategies , in T . cruzi infection focusing on Th17 stimulation . We perform immunizations by intranasal route with the aim of inducing both systemic and mucosal immunity . Also , this immunization route induces per se a Th17 immune response [27] . This is the first work in which the immunization with an antigen and c-di-AMP or c-di-AMP + αGC-PEG is used in the development of a vaccine against a parasite infection , evaluating both the immune response and protection developed . It was demonstrated that αGC-PEG changes the cytokine profile induced by c-di-AMP , partially reducing the number of antigen-specific IL-17 producing cells in spleen . This change in the elicited immune response correlates with a sharp reduction in protection against infection , confirming the role of Th17 response in protection against T . cruzi infection . Closely related to our results , in a recent work the importance of both Th1 and Th17 immune responses for vaccine-induced immunity against Mycobacterium tuberculosis ( an intracellular pathogen ) was suggested [45] . This work emphasizes the importance of the Th17 immune response developed in the groups adjuvanted with c-di-AMP . Previously [20] , and also in this work , we have demonstrate that NTc52 in a DNA-prime/protein-boost vaccine strategy confers clear protection against T . cruzi infection . In the previous work , we focused on the Th1-Th2 immune response; the IL-17 specific response was not analyzed . In this work we have focused on c-di-AMP adjuvanting Tc52 , NTc52 and CTc52 , and also the modulation of the immune response by αGC-PEG . As we mentioned before , we have included the NTc52PB group as a gold standard in challenge assays , because in previous reports it was our best prototype vaccine based on Tc52 as antigen [20] . As concluding remarks , our studies highlight the potential of intranasal immunization with NTc52+c-di-AMP , which induced strong cellular and antibody ( both mucosae and systemic ) responses with a Th17+Th1 profile , as a prophylactic strategy against Chagas . This immunization protocol conferred indeed clear protection against lethal infection , as demonstrated by the parasitemia , survival and weight loss parameters . | Chagas disease is a parasitic disease caused by a protozoan parasite ( Trypanosoma cruzi ) which has a complex life cycle including insect vector and mammalians . In Latin America , 7–10 million people are infected , 100 million people are at risk of infection , and about 56 , 000 new infection cases and 12 , 000 deaths are registered annually . Migration spread the geographic distribution of the disease to North America and Europe . The infection in humans has an initial acute stage followed by a chronic stage where up to 30% of patients develop cardiac alterations and 10% develop digestive , neurological or mixed alterations . The acute infection is hardly detected and there is not drug to treat the chronic infection . Thus , there is an urgent need for prophylactic and therapeutic vaccines development . Several attempts to find a vaccine antigen has been made and the protein Tc52 is a good candidate . In a vaccine composition , as important as the antigen is the adjuvants , which are substances able to increase , improve or modified the immune response . This research provides information about the immune response and protection against Trypanosoma cruzi infection elicited by Tc52 or portions of this molecule using different adjuvants . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"enzyme-linked",
"immunoassays",
"immunology",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"vaccines",
"preventive",
"medicine",
"immunologic",
"adjuvants",
"protozoans",
"infectious",
"disease",
"control",
"vaccination",
"an... | 2017 | Immunization with Tc52 or its amino terminal domain adjuvanted with c-di-AMP induces Th17+Th1 specific immune responses and confers protection against Trypanosoma cruzi |
Gene Ontology ( GO ) has established itself as the undisputed standard for protein function annotation . Most annotations are inferred electronically , i . e . without individual curator supervision , but they are widely considered unreliable . At the same time , we crucially depend on those automated annotations , as most newly sequenced genomes are non-model organisms . Here , we introduce a methodology to systematically and quantitatively evaluate electronic annotations . By exploiting changes in successive releases of the UniProt Gene Ontology Annotation database , we assessed the quality of electronic annotations in terms of specificity , reliability , and coverage . Overall , we not only found that electronic annotations have significantly improved in recent years , but also that their reliability now rivals that of annotations inferred by curators when they use evidence other than experiments from primary literature . This work provides the means to identify the subset of electronic annotations that can be relied upon—an important outcome given that >98% of all annotations are inferred without direct curation .
Gene Ontology ( GO ) annotations are a powerful way of capturing the functional information assigned to gene products [1] . The organization of the GO in a Directed Acyclic Graph allows for various levels of assignment specificity , while the three ontologies—Biological Process , Molecular Function , and Cellular Component—capture three aspects of the gene product annotation . Some GO annotations are assigned by expert curators , either from experimental evidence in the primary literature ( experimental annotations ) , or from other evidence such as sequence similarity , review papers and database entries ( curated annotations ) . However , the vast majority ( >98% ) of available GO annotations are assigned using computational methods , without curator oversight [2] ( Fig . 1 ) . Uncurated—electronic—annotations are generally considered to be least reliable . Many users of GO annotations err on the safe side by assigning a lower rank/weight to electronic annotations or leave them completely out of their analyses [e . g . ] , [ 3]–[7] . However , there have been very few evaluations of the quality of electronic annotations . To our knowledge , the most relevant study to date assessed the annotation quality of only 286 human proteins [8] . Here , we provide the first comprehensive evaluation of electronic GO annotation quality . Based on successive releases of the UniProt Gene Ontology Annotation database ( UniProt-GOA ) , the largest contributor of electronic annotations [9] , we used experimental annotations added in newer releases to confirm or reject electronic annotations from older releases . We defined 3 measures of annotation quality for a GO term: 1 ) reliability measures the proportion of electronic annotations later confirmed by new experimental annotations , 2 ) coverage measures the power of electronic annotations to predict experimental annotations , and 3 ) specificity measures how informative the predicted GO terms are . After describing our new methodology in detail , we first consider changes in quality in UniProt-GOA over time . We then characterize the relationship between GO term reliability and specificity . Next , we consider possible differences in quality among the three ontologies , among computational methods used to infer the electronic annotations , and among the 12 best-annotated model organisms . Finally , we contrast electronic annotations with curated annotations that use evidence other than experiments from primary literature .
We first sought to evaluate general trends in the overall quality of UniProt-GOA . Four summary statistics—first and third quartile , median , and mean—allow us to describe the change in quality—specificity , reliability , and coverage—of successive UniProt-GOA releases ( Fig . 3 ) . Subsequent UniProt-GOA releases are improving with the addition of slightly more specific annotations on average ( Fig . 3 A ) . At the same time , new UniProt-GOA releases show steady and significant improvement in reliability , as indicated by the increase of all four summary statistics ( Fig . 3 B ) . By contrast , the coverage of annotations has decreased somewhat ( Fig . 3 C ) . Taken together , these indicators suggest a general improvement in the quality of recent UniProt-GOA releases . Next , we investigated the association between a GO term's specificity and reliability ( Fig . 4 ) . Previous works based on smaller datasets have observed a negative relation between the predictive power of computational annotation and the specificity of the assigned GO term [e . g . ] , [ 11]–[13] . Our results are consistent with these results to the extent that almost all general terms are stable ( Fig . 4 ) . Specific terms , however , span the whole range of reliability . We also observe that on average , reliability of electronic annotations hardly depends on their specificity: the variance of reliability increases with an increase in specificity , but the median stays largely constant . To assess the differences in annotation quality among the three ontologies , we analyzed the ontologies separately in terms of reliability , coverage , and specificity . On average , annotations associated with the three ontologies were similarly stable , but vary considerably in coverage ( Fig . 5 ) . Specifically , Biological Process ( BP ) terms had the lowest coverage , Molecular Function ( MF ) terms had the highest coverage , and Cellular Component ( CC ) terms were in-between . This is consistent with the notion that MF terms are easiest to assign , and BP terms hardest to assign [14] . Nevertheless , this difference in difficulty translates into variable coverage but very similar reliability , suggesting that the false-positive rate of electronic annotations is controlled effectively . To investigate differences in quality among the various sources of electronic annotations in UniProt-GOA , we repeated our analysis for each of them . The six sources can be classified in two main categories: mapping of keywords from other databases ( UniProtKB keywords , UniProt Subcellular Location terms , InterPro , and Enzyme Commission ) and the use of comparative genomics in functional annotation ( Ensembl Compara for eukaryotes and HAMAP2GO for microbial genomes ) ( Fig . 6 ) . Two sources of electronic annotations are restricted to single ontologies: the Enzyme Commission ( EC ) numbers map to MF GO terms , and subcellular location terms of the UniProt database map to CC GO terms ( Fig . 6 A/B ) . Both annotation sources are applied to a comparatively small number of terms , but their reliability is remarkably high: on this restricted set of GO terms , they outperform other sources of electronic annotation ( Fig . 6 , Fig . S2 in Text S1 , and Fig . S3 in Text S1 ) . The bulk of electronic annotations are inferred from the UniProt and InterPro databases ( Fig . S4 in Text S1 ) . With UniProtKB keywords , GO annotations are inferred using a correspondence table between Swiss-Prot keywords associated with UniProt entries and GO terms . Note that UniProt entries consist of a small minority of manually annotated entries ( “Swiss-Prot entries” ) and a large body of entries ( “TrEMBL entries” ) automatically annotated by a rule-based system ( “UniRules” ) . With InterPro , GO annotations are inferred from a correspondence table between InterPro sequence and structure signatures and GO terms . Despite similarities in the two approaches , UniProt-based annotations show considerably higher average reliability than their InterPro-based counterparts ( Fig . 6 C/F , horizontal lines ) . In terms of average coverage , the two approaches show similar performance ( Fig . 6 C/F , vertical lines ) . Substantial manual curation is involved in obtaining electronic annotations from the two sources that rely on comparative genomics: Ensembl Compara electronic annotations transfer experimental annotations among inferred one-to-one orthologs in a subset of model organisms , and HAMAP2GO electronic annotations rely on manually created rules to propagate experimental annotations within a family of microbial proteins . Despite the intricacies involved in the annotation pipeline , these two sources have the lowest mean coverage and reliability among the six analyzed sources ( Fig . 6 D/E ) . However , note that the HAMAP rules have taxonomic restrictions on propagation that are not included in the HAMAP2G0 pipeline . Hence , some aspects of HAMAP are not captured in UniProt-GOA , and therefore are not analyzed here . This overall low reliability—a consequence of many rejected annotations—indicates that GOA strategies based on comparative genomics are currently less reliable than approaches based on sequence features ( UniProtKB keywords and InterPro ) . To investigate the difference in electronic annotation quality among the model organisms , we repeated our analysis for each model organism separately . Overall , repeating the analysis confirmed our general findings above . However , we observed variations among organisms , both in the number of available annotations and their quality ( Fig . 7 , Fig . S5 in Text S1 , Fig . S6 in Text S1 , and Fig . S7 in Text S1 ) . Organisms with the largest number of changes—confirmations and rejections—tend to have the highest quality of annotation: the three unicellular organisms and the three mammals ( Fig . 6 , top and bottom rows , Fig . S7 in Text S1 ) . Experimenting , describing and interpreting results on unicellular organisms is arguably more straightforward than on multicellular organisms; it might explain the relatively high quality of electronic annotations for the three unicellular model organisms ( Fig . 7 , bottom row ) . The average quality measures for the three mammals—Homo sapiens , Mus musculus , and Rattus norvegicus—are comparably high ( Fig . 7 , top row ) , but many specific low-quality annotations somewhat reduce the means of reliability and coverage . Our observation that general GO terms tend to have higher reliability holds for each model organism . Nevertheless , assigning mainly general GO terms guarantees neither high reliability nor high coverage . We observe the worst electronic annotation quality on Gallus gallus , Danio rerio and Dictiostelium discoideum gene products , despite a mean specificity of 1 . 79 , versus 4 . 47 for mammals . To put the quality of electronic annotations in perspective , we contrasted them to curated annotations ( evidence codes RCA , ISS , TAS , NAS , and IC ) , i . e . annotations inferred by curators without direct experimental evidence ( Fig . 8 ) . Curated annotations contain annotations assigned using evidence codes perceived as of particularly high quality: for instance , del Pozo et al . [5] consider the TAS evidence code to “offer the highest confidence [along with the IDA evidence code]” . Buza et al . [6] rank TAS and IC evidence code second only to the group of annotation codes that rely on direct experimental evidence . In Benabderrahmane et al . [7] , TAS is the only evidence code to receive the weight of 1 . 0 . Compared to electronic annotations , it is not surprising that curated annotations have a considerably lower average coverage ( Fig . 8 , vertical lines ) . Indeed , the main appeal of electronic annotations is precisely that they scale efficiently to large quantities of data . But in terms of reliability , and contrary to current beliefs , curated annotations that use evidence other than experiments from primary literature do not fare better than electronic annotations ( Fig . 8 , horizontal lines , Fig . S9 in Text S1 ) . In fact , we observed a higher reliability for electronic annotations than for curated annotations ( 0 . 52 vs . 0 . 33 ) . A more detailed analysis revealed that the lower mean reliability of curated annotations in the 16-01-2008 UniProt-GOA release is mainly due to removed annotations with evidence code Reviewed Computational Analysis ( RCA ) ( Fig . S10 in Text S1 ) . The low reliability of RCA annotations is caused by the removal of many RCA annotations assigned to the M . musculus gene products ( Fig . S7 in Text S1 , yellow bar in the panel denoted Mus musculus ) ; these were removed as there were concerns about the veracity of results from some papers that had been annotated ( Emily Dimmer , personal correspondence ) . When we exclude annotations assigned using the RCA evidence code , the reliability of non-experimental curated annotations rises to 0 . 58 . But even then , the reliability of electronic annotations ( 0 . 52 ) remains competitive with that of curated annotations ( Fig . S11 in Text S1 ) .
To narrow the gap between the number of sequenced gene products and those with functional annotation , computational methods are indispensable [18] , [19] , even more so for the non-model organisms ( Fig . S4 in Text S1 ) . We introduced three measures to evaluate the quality of electronic annotations: one accounts for the specificity of the assigned GO term , and two—reliability and coverage—assess the performance of electronic annotation sources by tracking changes in subsequent releases of annotation files . Although the performance of electronic annotations varies among inference methods ( “sources” ) , the overall quality of electronic annotations rivals the quality of curated non-experimental annotations . This is not to say that the curators have made themselves redundant . On the contrary , as we highlight above , most electronic annotations heavily rely on manually curated UniProtKB keywords and InterPro entries . Moreover , given the essential role of curators in embedding experimental results into ontologies , so does the present study .
We used the January 2011 release of the OBO-XML file to obtain the GO terms , definitions and the ontology structure needed in the analysis . The file was downloaded from the GO FTP site http://archive . geneontology . org/latest-full/ . The annotations ( mappings of gene products to GO terms ) were downloaded from the European Institute for Bioinformatics ( EBI ) FTP site ftp://ftp . ebi . ac . uk/pub/databases/GO/goa/UNIPROT/ . Each file , created as part of the UniProt Gene Ontology Annotation ( UniProt-GOA ) project [9] , is a many-to-many mapping of UniProtKB IDs to GO terms . All dates mentioned in this study refer to the release date of these annotation files , not the date attribute of individual annotations . We analyzed 193 , 027 UniProtKB IDs; GO terms can be assigned to these sequences using any of the evidence or reference codes . The distribution of annotations among the 12 Gene Ontology Reference genomes [10] is shown in Fig . S6 in Text S1 . This set of model organisms has by far the largest number of high-quality experimental annotations , allowing us to make the most reliable estimate of the annotation quality ( Fig . S1 in Text S1 ) . The structure of the GO vocabulary is changing as a response to consistency checks , new biological insights , and intricacies involved in annotating various model organisms [20]–[22] . To account for these changes , for each pair of GO releases analyzed we only consider terms that are present in both releases . The source of an annotation is recorded in the evidence code ( http://www . geneontology . org/GO . evidence . shtml ) . We group GO evidence codes into 3 broad categories: 1 ) codes reflecting annotations assigned by curators using direct experimental evidence from the literature ( experimental evidence codes EXP , IMP , IGI , IPI , IEP , IDA ) , 2 ) codes reflecting annotations inferred by curators using other types of evidence ( curated evidence codes ISS , RCA , IC , NAS , TAS ) and 3 ) electronic evidence code ( IEA ) , denoting annotations which are inferred computationally ( Fig . 1 ) . Several evidence codes were not included in the analysis: they are either used to indicate curation status/progress ( ND ) , are obsolete ( NR ) , or there is not enough data to make a reliable estimate of their quality ( ISO , ISA , ISM , IGC , IBA , IBD , IKR , IRD ) . A reference code captures the source of an electronic annotation . We analyze six reference codes available in UniProt-GOA: three are based on cross-referencing keywords from other databases: UniProtKB keywords , UniProt Subcellular Location terms , and Enzyme Commission [23] , [24]; two are based on the propagation of annotations within a family of proteins: InterPro and HAMAP2GO [25] , [26]; one reference code uses comparative genomics in projecting experimental annotations to unannotated inferred one-to-one orthologs—Ensembl Compara [27] . When a ‘NOT’ qualifier accompanies an annotation , it explicitly states that the gene product is not associated with the respective GO term . A subtle use of the ‘NOT’ qualifier comes into play because the isoform distinctions are not reflected in the annotation files at this time; a gene product can be mapped to the GO term in a given spatial/temporal context , but the mapping is not valid in another context ( Judith Blake and Pascale Gaudet , personal correspondence ) . Such gene products will be mapped to one GO term twice—one accompanied by a ‘NOT’ qualifier and one without it . For consistency , we ignore all such occurrences . The 11-01-2011 UniProt-GOA release contains 493 gene products with such annotations . All analyses are performed on overlapping 3-year periods between 2006 and 2011 . Unless stated otherwise , we show the results associated with the most recent period ( 2008–2011 ) . The three measures of quality we introduced are specificity , reliability , and coverage . For clarity , the definitions are given and described for electronic annotations . Nevertheless , any subset of annotations can be analyzed this way , e . g . annotations assigned using one or a subset of evidence or reference codes . We measure the specificity ( opposite of generality ) of a GO term GOi with respect to its information content [10] , [28] , [29]:where freq ( GOi ) is the frequency of GOi among all annotations considered . To calculate the reliability for a GO term , we count all the confirmed and rejected electronic annotations associated with this term ( Fig . 2 A ) . An electronic annotation is confirmed if it is corroborated by a new ( added during the time interval ) experimental annotation . An electronic annotation is rejected if it is falsified by a new experimental annotation that comes with a ‘NOT’ qualifier , or if this electronic annotation has been removed in the later UniProt-GOA release . More formally , where is the set of confirmed annotations associated with term GOi and is the set of rejected and removed annotations associated with term GOi . To calculate the coverage for a GO term in a UniProt-GOA release , we count all the new experimental annotations in the later UniProt-GOA release correctly predicted by an electronic annotation in the earlier release , and those not correctly predicted ( missed ) by electronic annotations in the earlier release ( Fig . 2 B ) . More formally , where is the set of correctly predicted new experimental annotations associated with term GOi and is the set of “missed” new experimental annotations associated with term GOi . To calculate any of the measures of quality , we take into account the GO Direct Acyclic Graph ( DAG ) structure . To calculate the frequency of a GO term , we account for all annotations derived by inheritance . Consequently , the specificity of any child term is necessarily greater than or equal to the specificity of its parents . When calculating reliability , an annotation that is replaced by a more specific annotation ( a descendent ) is not considered rejected , as the descendent still implies it . Similarly , an annotation is confirmed by the arrival of an experimentally ascertained descendent , as the more specific term implies the more general term . Conversely , if an annotation is followed by the arrival of a less specific experimental annotation , only the subset of its ancestral terms implied by the less specific experimental annotation is deemed as confirmed; the rest is uninformative ( neither confirmed , rejected , or removed ) . All the results of the described analysis are available as Dataset S2 . The analysis was done using a combination of in-house Java classes , SQL queries to the custom database , and R scripts . Summaries were done using the plyr package of the R language [30]; all plots were created using the ggplot2 package of the R language [31] , and the interactive plots were created using the googleVis package of the R language; the respective R packages are available from the CRAN repository . REVIGO web server [32] was used to summarize the lists of GO terms and select those highlighted in the Results section . | In the UniProt Gene Ontology Annotation database , the largest repository of functional annotations , over 98% of all function annotations are inferred in silico , without curator oversight . Yet these “electronic GO annotations” are generally perceived as unreliable; they are disregarded in many studies . In this article , we introduce novel methodology to systematically evaluate the quality of electronic annotations . We then provide the first comprehensive assessment of the reliability of electronic GO annotations . Overall , we found that electronic annotations are more reliable than generally believed , to an extent that they are competitive with annotations inferred by curators when they use evidence other than experiments from primary literature . But we also report significant variations among inference methods , types of annotations , and organisms . This work provides guidance for Gene Ontology users and lays the foundations for improving computational approaches to GO function inference . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"functional",
"genomics",
"gene",
"function",
"genome",
"analysis",
"tools",
"genome",
"databases",
"biology",
"proteomics",
"gene",
"ontologies",
"biochemistry",
"proteomic",
"databases",
"genetics",
"genomics",
"computational",
"biology",
"genetics",
"and",
"genomics"
] | 2012 | Quality of Computationally Inferred Gene Ontology Annotations |
Incremental learning , in which new knowledge is acquired gradually through trial and error , can be distinguished from one-shot learning , in which the brain learns rapidly from only a single pairing of a stimulus and a consequence . Very little is known about how the brain transitions between these two fundamentally different forms of learning . Here we test a computational hypothesis that uncertainty about the causal relationship between a stimulus and an outcome induces rapid changes in the rate of learning , which in turn mediates the transition between incremental and one-shot learning . By using a novel behavioral task in combination with functional magnetic resonance imaging ( fMRI ) data from human volunteers , we found evidence implicating the ventrolateral prefrontal cortex and hippocampus in this process . The hippocampus was selectively “switched” on when one-shot learning was predicted to occur , while the ventrolateral prefrontal cortex was found to encode uncertainty about the causal association , exhibiting increased coupling with the hippocampus for high-learning rates , suggesting this region may act as a “switch , ” turning on and off one-shot learning as required .
In standard associative learning , an animal must repeatedly experience a number of pairings between a stimulus and a consequence before a particular stimulus pairing is fully learned . Learning is inevitably incremental . However , animals sometimes encounter outcomes that they have never experienced previously and from which it is necessary to learn rapidly in order to survive . In such cases , animals can learn on the basis of only a single exposure to a stimulus pairing , a situation described in the literature as one-shot learning . For example , in one-shot object categorization , humans demonstrate a capability to rapidly learn to recognize novel objects by means of a priori knowledge of object categories [1] . In jumping to conclusions , humans are known to undergo a rapid inference process because of an overestimation of the cost of acquiring more information [2–4] . In ( mis ) attribution , an outcome can be ( erroneously ) attributed to a ( wrong ) cause [4 , 5] . Owing to these findings , we have made progress in elucidating distinctive traits of one-shot learning at the behavioral level . However , we still have only a rudimentary understanding of the computational principles underpinning one-shot learning , and therefore , how this computational process unfolds at the neural level remains largely unknown . Much progress has been made in understanding the computational mechanisms underpinning incremental learning with algorithms such as the Rescorla-Wagner rule [6] , the probabilistic contrast model [7] , the associative learning model [8 , 9] , and Bayesian causal inference [4 , 10–12] , providing computational accounts for a wide variety of types of incremental learning . In these models , repeated experience of the same stimulus and outcome gradually cements the causal relationship between them until there is little left to learn . However , one-shot learning imposes a substantial challenge to these learning algorithms because such models are not optimized to facilitate learning from a single experience . One viable mechanism for switching between incremental and one-shot learning might be via the control of learning rates . It has been suggested in previous theoretical proposals that learning rate is modulated through changes in environmental uncertainty , such as volatility [8 , 13 , 14] . However , these prior proposals are not designed to account for one-shot learning because learning rates are adjusted only gradually in such frameworks based on detecting changes in environmental volatility or jumps in contingencies [13 , 14] . One process that is likely to contribute substantially to one-shot learning is episodic memory [15 , 16] , in which rapid associations between a context and an event can be formed [17–22] . This type of memory has long been known to depend at least in part on the hippocampal complex [23–27] , which has also been proposed to be both functionally and anatomically dissociable from other forms of memory involved in mediating more incremental types of associative learning [15 , 20 , 28 , 29] . However , while much evidence has accumulated in support of the notion of dissociable memory systems for one-shot and incremental learning [15 , 28 , 30–33] , almost nothing is known about how the brain is capable of switching between different types of learning strategy . In other words , how does the brain know when to deploy the episodic memory system as opposed to relying on incremental learning ? The aim of the present study is to test a novel computational framework that can account for when and how one-shot learning occurs over and above incremental learning , as well as to gain insight into how the brain is capable of implementing the switch between these different learning strategies . Our computational hypothesis is that the rates at which individuals learn to causally associate a stimulus with an outcome increases with the extent to which the relative amount of uncertainty in the causal relationship between that stimulus and outcome are left unresolved . Specifically , the more uncertainty there is about the causal relationship between a stimulus and an outcome , the higher the learning rate that is assigned to that stimulus in order to resolve the uncertainty . Stimulus-outcome pairs with very high uncertainty associated with them should elicit very rapid one-shot learning . At the neural level , the ventrolateral prefrontal cortex ( vlPFC ) has long been hypothesized to guide a control process that determines whether items are remembered or forgotten during episodic encoding [34–40] . It has further been established that interplay between the vlPFC and the hippocampus increases more when learning stronger associations than when learning weaker ones [36] and that these structures exhibit elevated connectivity during demanding tasks relative to less demanding ones [41] . Considering the functional similarity between episodic memory processes and one-shot learning , we hypothesized that areas involved in episodic memory processing such as the hippocampus might be selectively engaged under situations in which one-shot learning occurs . We further hypothesized that recruitment of the hippocampus might depend on uncertainty computations about the causal relationship mediated in parts of the prefrontal cortex—particularly the lateral prefrontal cortex , which has been implicated in human causal learning in previous studies [42–45] .
To test our hypotheses , we combined formal computational models with behavioral and neuroimaging ( functional magnetic resonance imaging [fMRI] ) data acquired from individuals performing a simple causal inference task , in which learning can occur from a single experience . Measuring neural signals for one-shot learning is challenging because by virtue of how rapidly it happens , there is very little time to collect data samples while it is going on in the brain . To resolve this issue , we developed a novel paradigm that enables us to assess one-shot learning ( Fig 1 ) . On each trial , participants are presented with a sequence of pictures . These pictures vary in the degree of frequency in which they are presented ( “novel cue” for the least frequently presented pictures and “non-novel cue” otherwise ) . After the sequence of pictures has been presented , participants will receive a monetary outcome , for which neither response nor choice is needed . This outcome will in some cases be to win a certain amount of money and on other occasions will involve losing a certain amount of money ( see Materials and Methods for more details ) . One type of outcome is presented frequently ( “non-novel outcome” ) , while the other type is presented only once ( “novel outcome” ) . After all the outcomes are presented in a round , participants are then asked to make ratings about how likely it is that each individual stimulus encountered during the round can cause an outcome ( “causal rating” ) . Participants were told that all rounds are independent of each other , and no participant reported noticing any dependencies between rounds while performing the task . Passive viewing of these sequential stimuli and outcome presentations allows us to test a pure effect of learning , without the confounding effects of choice behavior . Another important feature of this task is that the stimuli and outcomes do not necessarily occur contiguously—that is , multiple different stimuli are presented before each outcome is presented . This makes the present paradigm quite different from typical associative learning paradigms such as Pavlovian conditioning , in which a single cue is usually followed by an outcome ( or the absence of an outcome ) . The design of our task was structured so as to enable us to distinguish between incremental and one-shot learning . The task is designed , as will be described below , to enable us to test whether one-shot learning occurs when the amount of uncertainty in the causal relationship between a stimulus and an outcome is greater than that of other stimulus-outcome pairs . The computational model we propose to account for one-shot learning is based on a probabilistic instantiation of a causal learning model ( Fig 2 ) . It builds on the premise that one-shot learning is characterized by a dramatic increase of learning rate ( the rate at which new information is taken into account to update one’s current predictions ) . Such an increase in learning rate is hypothesized to occur when uncertainty in the causal relationship between a stimulus and an outcome is maximal , with the rate decreasing as the uncertainty is resolved . To implement this hypothesis , we constructed a normative Bayesian model for one-shot learning in which a Bayesian learner attempts to establish the causal relationship between a stimulus and an outcome ( “causal strength” [10 , 11] ) and the uncertainty about the causal strength ( “causal uncertainty” ) , whilst the degree of relative uncertainty about different possible causal relationships modulates the learning rate . Specifically , the model performs a probability estimate about the extent to which a particular stimulus has caused a given outcome ( see Fig 2A and Materials and Methods , section “Bayesian inference for causal learning , ” for more details ) . This is specified by the parameters of the posterior distribution , which is updated at the end of each trial when an outcome is presented . The mean and the variance of the posterior are referred to as the causal strength and the causal uncertainty , respectively . We speculate that the former is an observable variable reflected in actual rating patterns and the latter is a latent variable reflected in neural patterns that essentially weave those rating patterns . The amount of causal uncertainty for each stimulus and outcome pair is then compared and translated into the learning rate by means of the softmax function ( see Fig 2B ) [46] . This process of learning rate control assumes that the rate of learning to resolve remaining causal uncertainty about the stimulus-outcome pair at present increases with the amount of causal uncertainty left unresolved for this pair , compared to the amount of causal uncertainty for other pairs . For example , if causal uncertainties about all possible stimulus-outcome pairings were almost equal to each other , the brain would allocate learning rates evenly to all pairs , resulting in slow learning ( refer to “incremental learning” in Fig 2B ) . Conversely , a stimulus paired with an outcome in which the amount of causal uncertainty is significantly greater than other possible stimulus-outcome pairings results in very rapid learning , such that even within a single trial substantial learning occurs ( refer to “one-shot learning” in Fig 2B ) . Note that the model updates its posterior distributions about the stimulus-outcome causal relationships on a trial-by-trial basis ( when an outcome is presented ) while assigning learning rates to each individual stimulus-outcome pair on an event-by-event basis ( when each stimulus is presented ) . An assessment of the model’s viability and supporting simulation results demonstrating that the learning rate assignment effectively reduces the amount of causal uncertainty are provided in Materials and Methods and S1 Fig and S2 Fig . It is noted that our model reduces to a simpler heuristic model if we assume that causal uncertainty is high whenever the stimulus and the outcome novelty are high . This assumption leads us to consider an alternative hypothesis stating that the causal learning process is purely driven by the novelty of the stimulus-outcome pair . This can also be viewed as reflecting the operation of a simple heuristic causal judgment that a novel stimulus is rated as the most probable cause of a novel outcome if the novel stimulus is paired with that novel outcome . The event-by-event distinction between incremental and one-shot learning is made by simply reading out the learning rate for the stimulus presented at the time of each event . High learning rates at the time of stimulus event , reflecting the extent to which the relative amount of uncertainty in the causal relationship between that stimulus and outcome are left unresolved , imply that the model has high expectations that the stimulus event will be followed by an “informative” outcome event , by which time causal uncertainty is likely to be resolved . The one-shot learning we refer to here is thus an expectation or preparedness for an event that enables a decrease in the remaining amount of causal uncertainty , regardless of whether or not a novel outcome is presented at the end of the current trial . This allows us not only to test predictions of our causal uncertainty model above and beyond the distinction drawn by the novelty of the stimulus-outcome pair but also to dissociate neural processes pertaining to one-shot learning from those involved in incremental learning . Forty-seven adult participants ( 14 females , between the ages of 19 and 43 , mean = 25 . 8 , standard deviation = 5 . 2 ) performed the task in total . Twenty ( ten females , between the ages of 19 and 40 , mean = 26 . 1 , standard deviation = 5 . 3 ) among them were scanned with fMRI , and another 27 ( four females , between the ages of 20 and 43 , mean = 25 . 6 , standard deviation = 5 . 3 ) performed the task in follow-up behavioral experiments . One generic characteristic of our causal uncertainty model is that uncertainty will generally be high when a novel stimulus is paired with a novel outcome . To test for this effect , we created two round categories by stratifying rounds by the novelty of the stimulus-outcome pair . In the type1 rounds , a novel cue is paired with a non-novel outcome . In the type2 rounds , a novel cue is paired with a novel outcome . In both cases , non-novel cues are paired with both the novel and the non-novel outcomes . As expected , we found that a majority of participants rated the novel stimulus as the most probable cause of the novel outcome in the type2 round , as opposed to the type1 round ( Fig 3; p < 1e-4 , paired-sample t test with the causal rating dataset and one sample t test with the one-shot effect index ) . This one-shot learning effect occurs regardless of any delay between the novel stimulus and the novel outcome within a trial; we found no significant correlation between the distance between a novel stimulus and an outcome and the one-shot effect , which is defined as the causal rating for the novel cue minus the average causal ratings for the non-novel cues ( correlation coefficients = -0 . 002 , p = 0 . 95 ) . This implies that the model successfully predicts the effect of stimulus-outcome novelty . In two follow-up behavioral experiments , we examined if there is an effect of the sign of the outcome ( whether the novel outcome is a gain or a loss ) and of the magnitude of the outcome on the causal ratings ( see Materials and Methods for more details ) . While we found no effect of sign ( Fig 3—Experiment3; two-way repeated-measure ANOVA; F-test = 0 . 52 , p = 0 . 46 ) , a modest effect of outcome magnitude was found on the extent to which one-shot learning occurs when systematically examined ( Fig 3—Experiment2; two-way repeated-measure ANOVA; F-test = 4 . 31 , p = 0 . 04 ) , suggesting that participants do take the amount of outcome into account when making causality judgments . In both of the behavioral follow-up experiments , the one-shot learning effect was observed ( two-way repeated-measure ANOVA; F-test > 200 , p < 1e-20 ) . Our causal uncertainty model and the heuristic causal judgment mechanism could both equally well account for the behavioral results reported above . To further distinguish the predictions of these models , we attempted to demonstrate that our causal uncertainty model makes additional predictions about the causal ratings above and beyond the distinction made by the novelty of the stimulus-outcome pair . To do this , we created event categories by stratifying stimulus events by the learning rates predicted by our causal uncertainty model: one-shot learning events ( OS ) are defined as a collection of discrete stimulus events during which the learning rate of the model is greater than the 90th percentile , while the remaining events are deemed to correspond to incremental learning events ( IC ) ( see Fig 4A and Materials and Methods for more details ) ; our independent model-based analysis indicated that the 90th percentile threshold is a viable predictor for distinguishing between one-shot learning and incremental learning ( see S3 Fig for full details , including the rationale for the choice of the 90th percentile cut off ) . Second , we defined a one-shot learning round ( OS round ) as a round during which the model predicts occurrence of OS ( see the right of Fig 4A ) and the incremental learning round ( IC round ) as a round during which the model predicts no occurrence of OS ( see the left of Fig 4A ) . It is important to note that the type1 rounds do not necessarily overlap with the IC rounds nor do the type2 rounds overlap with the OS rounds ( see S4 Fig for more details ) . In the type1 round during which a novel cue is paired with a non-novel outcome , the one-shot effect index ( the causal rating for the novel cue minus the average causal ratings for the non-novel cues ) for the OS rounds is more negative than for the IC rounds ( see the left of Fig 4B; paired-sample t test , p < 0 . 01 ) , demonstrating that the extent to which participants rated the non-novel stimulus as the most probable cause of the novel outcome in the OS rounds is greater than in the IC rounds . Conversely , in the type2 round , the one-shot effect index for the OS rounds is more positive than for the IC rounds ( see the right of Fig 4B; paired-sample t test , p < 0 . 05 ) , demonstrating that the extent to which participants rated the novel stimulus as the most probable cause of the novel outcome in the OS rounds is greater than in the IC rounds . These findings demonstrate that additional variability in the causal ratings can be explained by our causal uncertainty model to a greater extent than by the predictions of the heuristic causal judgment model ( see S4 Fig for further predictions of the causal uncertainty model ) . This reiterates the point that one-shot learning is guided by causal uncertainty , which makes a distinction between the IC and the OS rounds rather than focusing on the novelty of the stimulus-outcome pair that distinguishes the type1 and the type2 rounds . To further evaluate our hypothesis that a computation about causal uncertainty best explains the behavioral data , we formally pitted our proposed causal uncertainty model against the heuristic causal judgment model . We found that the model version formulating our hypothesis performed significantly better than the heuristic causal judgment model in terms of producing a smaller mean squared error of the difference between the model predictions and participants’ actual causal ratings using a leave-one-out cross validation procedure to take into account effects of model complexity and overfitting ( paired-sample t test , p < 0 . 01; see Fig 4C and S5 Fig ) . The causal uncertainty model also exhibits rating patterns qualitatively more similar to subjects rating behavior than the alternative model ( see Fig 4B for patterns of causal strength of the best model and S6 Fig for patterns of the heuristic causal judgment model ) . This result provides even stronger support for our contention that our causal uncertainty model provides a better account of participants’ one-shot causal learning behavior than a simple heuristic approach ( also see S4 Fig for comparison between predictions of the causal uncertainty model and predictions of the heuristic causal judgment model ) . In addition , we also compared the performance of our causal uncertainty model against seven other alternative learning models typically used to account for incremental learning ( See S1 Table for the full list of models ) . These include the Rescorla-Wagner model [6] , the probabilistic contrast model [7] , the Pearce-Hall associative model [8] , variants of Bayesian latent variable models [47] , a Bayesian causal network learning model allowing for an identification of a combination of stimuli as a cause of a particular outcome and for an establishment of a causal relationship between different outcomes [48] , and a model of heuristic causal judgment that is designed to test the hypothesis that causal learning is driven by the novelty of the stimulus-outcome pair ( see Materials and Methods ) . Among alternative models , the Bayesian causal network learning model proposes that participants might use a more complex causal structure , such as a combination of cues causally linked to outcomes , than that deployed in our causal uncertainty model . Our causal uncertainty model outperformed each and every alternative model , including the Rescorla-Wagner model showing the second-best model fitness , both quantitatively ( paired sample t test at p < 0 . 05; see S5 Fig ) as well as in terms of the qualitative ratings patterns ( see S6 Fig ) . Collectively , these results provide further additional support for our model . To establish the neural computations underlying one-shot learning , we regressed each of our computational signals against the fMRI data ( for testing signals , see Materials and Methods ) . For a strict identification of regions responsible for uncertainty processing , we first tested for regions correlating with stimulus novelty , which was defined simply as the number of times a participant had encountered a particular stimulus in the task ( with a stimulus being most novel when first encountered ) . Next , we entered causal uncertainty into our analysis after adjusting for the effects of novelty so that areas found correlating with uncertainty are doing so above and beyond any effect of novelty . Novelty was positively correlated with activity in multiple areas—dorsal parts of prefrontal cortex , inferior parietal lobule , middle temporal gyrus ( p < 0 . 05 family-wise error ( FWE ) corrected , Fig 5A; S2 Table ) , and Caudate ( p < 0 . 05 cluster level corrected , S2 Table ) —and was negatively correlated with activity in fusiform gyrus extending to the parahippocampal gyrus ( p < 0 . 05 FWE corrected , Fig 5A; S2 Table ) . These results are highly consistent with previous findings implicating the medial temporal gyrus , parahippocampal gyrus , and fusiform gyrus in processing familiarity and novelty [49–53] . However , this interpretation might be complicated by the fact that , in our fMRI experimental design , the stimulus novelty could be associated with a growing association with a loss and a decreasing likelihood of being associated with the large novel gain . Above and beyond novelty , causal uncertainty was found to correlate with activity in multiple prefrontal areas , including vlPFC ( p < 0 . 05 FWE corrected , Fig 5A; S2 Table ) and dorsomedial prefrontal cortex ( p < 0 . 05 cluster level corrected , Fig 5A; S2 Table ) , consistent with our initial hypothesis . To determine which brain regions are engaged on events during which the model predicts the participant will implement one-shot learning , we ran a categorical analysis between event types ( Fig 5B ) . A significantly increased neural activation was found in multiple areas including hippocampus as well as fusiform gyrus ( p < 0 . 05 FWE corrected , Fig 5B ) on one-shot learning events compared to incremental learning events . We did not find any areas showing significantly increased activity on incremental learning events compared to one-shot learning events . One interpretation of these results is that neural systems previously implicated in incremental learning , such as the striatum [55 , 56] , may always be active during all learning scenarios ( including the one-shot case ) but the hippocampal system is additionally recruited when one-shot learning needs to take place . The selective recruitment is not solely driven by the detection of a novel stimulus , which is a situation in which participants try to consciously remember a novel stimulus in order to establish strong stimulus-outcome associations , as indicated by the low correlation between the stimulus category of the best-fitting model ( OS and IC ) and the category by the novelty type ( novel cue and non-novel cue ) ( Matthews correlation coefficient; mean = 0 . 29 , standard deviation = 0 . 1 across subjects ) . To further test our neural hypothesis about the role of hippocampus during OS , we subsequently ran an ROI analysis using an anatomically defined hippocampus mask [54] . The mean percent signal change , which quantifies how much the evoked BOLD response deviates from its voxel-wise baseline , was computed within the hippocampus ROI across all subjects . We found a significant increase in neural activity during OS but not during IC ( paired-sample t test p < 1e-8 , Fig 5C; also see S7 Fig for testing subregions of hippocampus ) . Importantly , when plotting hippocampal activity as a function of varying model-predicted learning rates throughout the experiment , we found evidence that the hippocampus is selectively recruited during very high learning rates ( above the 90th percentile ) and not for lower learning rates ( p < 1e-6; one-sample t test after Bonferroni adjustment for multiple comparisons across the ten learning bins ) . This suggests that the hippocampus gets switched on at high learning rates , when one-shot learning needs to take place , and that the hippocampus is not engaged when more incremental types of learning take place . These results firmly support our hypothesis that the hippocampal memory system contributes to events during which one-shot learning takes place , further suggesting that this region is relatively silent during incremental learning . In order to further characterize how the vlPFC , the region we found to most prominently encode causal uncertainty , interacts with the hippocampus during one-shot learning , we also ran a connectivity analysis ( see Materials and Methods for technical details ) . We computed correlations between the neural signals in vlPFC and hippocampus for different learning rates and found that functional coupling between the vlPFC and the hippocampus was high during very high learning rates but not during learning rates associated with more incremental learning ( Fig 6 ) . Both the patterns of the vlPFC activation described above and of the connectivity results presented here indicate that the vlPFC may selectively interact with the hippocampus particularly under circumstances in which one-shot learning is warranted . This finding supports the possibility that the vlPFC may effectively operate as a switch to turn on the hippocampus when it is needed during one-shot learning situations and , furthermore , leads us to understand the nature of the activity in hippocampus during one-shot learning , i . e . , that the hippocampus encodes causal uncertainty signal only during high learning rate events . We also tested whether our neural results might be accounted for by the following alternative factors .
At the neural level , our findings indicate evidence for involvement of a very specific neural system for the range of learning rates that would support one-shot learning according to our model . Specifically , activity in the hippocampus was ramped up for high learning rates ( 90th percentile or more ) relative to slower learning rates , in which , by contrast , the hippocampus showed no activity . Thus , the hippocampus appeared to be recruited in a switch-like manner , coming on only when one-shot learning occurred and being silent otherwise . Our findings support the theoretical proposition that episodic memory systems play a unique role in guiding behavior , distinct from the contribution of other systems involved in more incremental types of learning such as goal-directed and habitual instrumental control and Pavlovian learning [15 , 59] . It is worth noting that while previous studies have found hippocampal involvement in goal-directed learning , such as in our recent study in which the hippocampus was found to respond during the “planning” stage of a goal-directed action [60] , it is possible that the hippocampus is involved in contributing to two distinct computational processes . It is still an open question as to whether the mechanisms involved in one-shot learning are conceptually and neurally distinct from those being studied in goal-directed learning . The fMRI results suggest that parts of the prefrontal cortex including the vlPFC are involved in encoding uncertainty about the causal relationships between cues and outcomes . The ventrolateral prefrontal cortex has previously been implicated in memory encoding and explicit memory attribution [40 , 50 , 61–64] . Our findings provide new insight into the nature of the top-down control functions of vlPFC . In our computational model , the degree of causal uncertainty surrounding a cue-outcome relationship is used to dramatically adjust the learning so as to engage one-shot learning when required . The vlPFC was found to encode the causal uncertainty signal that in turn could be used to modulate learning rates , in a highly nonlinear fashion . One interpretation of these findings is that vlPFC uses knowledge about causal uncertainty to act as the controller of a switch , engaging episodic memory systems when learning needs to proceed from a single episode ( one shot ) as opposed to incrementally . This view is supported by our demonstration of the interactions between vlPFC and the hippocampus . A previous study found enhanced connectivity between the vlPFC/ dorsolateral prefrontal cortex ( dlPFC ) and the hippocampus in correct memory retrieval [64] , and this finding invited the speculation that lateral prefrontal cortex is recruited to guide explicit associative memory decisions . The present study greatly extends this proposal by providing a specific computational account for how connectivity between the two regions is controlled: the present results suggest that change in learning rate is an important factor involved in governing the degree of connectivity between these areas during one-shot learning . When learning rates were high ( above the 90th percentile ) , there was increased connectivity between vlPFC and hippocampus , which also corresponded to the selective increase in activity in the hippocampus during task performance , consistent with the possibility that vlPFC is acting to engage the hippocampus when it is required to facilitate one-shot learning . It is important to note that not all types of one-shot learning may be mediated by the hippocampus nor is the hippocampus likely the sole contributor to this process . In particular , taste aversion learning may depend on additional neural circuits [65 , 66] , and there is ongoing debate about whether or not the hippocampus is even necessary for taste-aversion learning [67] . Here , the outcomes used in the present task ( small monetary gains and losses ) are relatively inconsequential , as compared to taste aversion learning or learning with other highly biologically relevant outcomes . One important extension of the present work would be to also examine one-shot learning in circumstances involving more biologically relevant stimuli such as aversive tastes or pain in order to ascertain whether similar or distinct neural structures are implicated . In future work , it would also be worthwhile to determine the extent to which anxiety-related mechanisms might modulate hippocampal activity during one-shot learning when individuals are presented with highly aversive stimuli . In the present work , we also attempted to provide a specific computational account and to discount alternative explanations for our data . One very obvious possible alternative account is that participants may simply use a heuristic strategy in which the most novel stimuli are assumed to be responsible for causing any given outcome , as opposed to using the more sophisticated strategy of representing the causal uncertainty about a stimulus-outcome relationship . However , when we directly tested this heuristic strategy against out behavioral data , it did not account as well for the behavioral results as did our causal uncertainty model . Furthermore , we tested a number of other alternative models that are traditionally used to account for incremental associative learning . In all cases , the model we proposed was superior . This suggests that the present model is a highly parsimonious one with two core elementary features that are likely to be important elements of how the brain solves the problem of one-shot learning: the first is a computation of a representation of uncertainty about the causal relationship between events , and the second is the flexible adjustment of learning rates to accommodate rapid learning about those events . Those two key model features that correspond to the two mains signals we observed in the brain during task performance are , we suspect , likely to be an important component of any successful algorithmic approach to one-shot learning . The predictions of the present model together with the neural findings warrant an investigation of more challenging problems , including whether or not one-shot learning would occur when the amount of causal uncertainty keeps increasing in spite of a continual decrease in stimulus novelty . This refers to highly chaotic situations in which making observations does not necessarily guarantee resolving uncertainty in the causal relationships between stimuli and outcomes . In future work , testing for such effects would allow us to investigate how the causal inference process breaks down in such conditions . Taken together , these findings form the basis of a new understanding of the neural computations underlying the ability to learn from a single exposure to an event and its consequences . Developing a detailed account of when rapid learning takes place and which brain areas are engaged in this process might subsequently open the window to better understanding situations under which rapid causal attributions are generated in a dysfunctional manner such as in misattribution , superstition , and delusional reasoning [2–5 , 68] .
Forty-nine adult participants ( 14 females , between the ages of 19 and 43 , mean = 25 . 8 , standard deviation = 5 . 2 ) were recruited in total . Two participants , who gave the same unchanging causal ratings in most of the trials , were excluded from our analysis . Out of these , 20 ( ten females , between the ages of 19 and 40 , mean = 26 . 1 , standard deviation = 5 . 3 ) participated in the fMRI study ( experiment 1 ) , 13 participants ( two females , between the ages of 20 and 43 , mean = 25 . 5 , standard deviation = 6 . 2 ) participated in experiment 2 ( behavioral only ) , and 14 participants ( two females , between the ages of 22 and 35 , mean = 25 . 8 , standard deviation = 4 . 4 ) participated in experiment 3 ( behavioral only ) . All participants gave written consent , and the study was approved by the Institutional Review Board of the California Institute of Technology ( Protocol number 12–359 ) . Participants were screened prior to the experiment to exclude those with a history of neurological or psychiatric illness . The design was the same as that of experiment 1 except that we used two kinds of outcome pairs ( non-novel , novel ) : ( 10 , -50 ) and ( 50 , -10 ) . Thirteen participants were asked to complete 40 rounds . The change of the outcome pair was made on a trial-by-trial basis . This allows us to test the effect of outcome amount on subject ratings and to see if the one-shot learning effect still stands up . The design was the same as that of the experiment 1 except that we used two kinds of outcome pairs ( non-novel , novel ) : ( 10 , -50 ) and ( -10 , 50 ) . Fourteen participants were asked to complete 40 rounds . The change of the outcome pair was made on a trial-by-trial basis . This allows us to test the effect of the sign of an outcome on subject ratings and to see if the one-shot learning effect still stands up . Functional imaging was performed on a 3T Siemens ( Erlangen , Germany ) Tim Trio scanner located at the Caltech Brain Imaging Center ( Pasadena , California ) with a 32-channel radio frequency coil for all the MR scanning sessions . To reduce the possibility of head movement related–artifacts , participants' heads were securely positioned with foam position pillows . High-resolution structural images were collected using a standard MPRAGE pulse sequence , providing full brain coverage at a resolution of 1 mm × 1 mm × 1 mm . Functional images were collected at an angle of 30° from the anterior commissure-posterior commissure ( AC-PC ) axis , which reduced signal dropout in the orbitofrontal cortex [69] . Forty-five slices were acquired at a resolution of 3 mm × 3 mm × 3 mm , providing whole-brain coverage . A one-shot echo-planar imaging ( EPI ) pulse sequence was used ( TR = 2800 ms , TE = 30 ms , FOV = 100 mm , flip angle = 80° ) . The SPM8 software package was used to analyze the fMRI data ( Wellcome Department of Imaging Neuroscience , Institute of Neurology , London , United Kingdom ) . The first four volumes of images were discarded to avoid T1 equilibrium effects . Slice-timing correction was applied to the functional images to adjust for the fact that different slices within each image were acquired at slightly different points in time . Images were corrected for participant motion , spatially transformed to match a standard echo-planar imaging template brain , and smoothed using a 3-D Gaussian kernel ( 6-mm FWHM ) to account for anatomical differences between participants . This set of data was then analyzed statistically . A high-pass filter with a cutoff at 129 s was used . A GLM was used to generate voxelwise statistical parametric maps ( SPMs ) from the fMRI data . We created subject-specific design matrices containing the regressors ( R ) in the following order: ( R1 ) a block regressor encoding the average BOLD response during the full duration of the rating submission phase , ( R2 ) a regressor encoding the average BOLD response at the time of each stimulus presentation ( 1-s duration ) , ( R3 ) a parametric modulator encoding the novelty of stimuli , ( R4 ) a parametric modulator encoding the posterior variance ( refer to section “Bayesian inference for causal learning: ( 2 ) Latent inhibition” ) , ( R5 ) a regressor encoding the average BOLD response at the outcome state ( 2-s duration ) , and ( R6 ) a categorical variable representing the novelty of the stimulus-outcome pair at the time of each outcome presentation ( 1 if both the stimulus and the outcome is novel , 0 otherwise ) . The order of the regressors is determined in a way that eliminates the variance of no interest . These regressors were orthogonalized with respect to the previous ones in order to prevent shared variance from being explained multiple times . All of the findings we report survive whole-brain correction for multiple comparisons at the cluster level ( corresponding to “+”; height threshold t = 3 . 53 , extent > 100 voxels , p < 0 . 05 corrected ) . We used this single basic statistical threshold throughout the analysis . The areas surviving the most stringent threshold , p < 0 . 05 FWE whole-brain corrected at the voxel level , are marked with “*” in S2 Table and also shown in Fig 5 ( cyan and yellow blobs in the statistical maps ) . In addition , for all the figures , in order to show the full extent of the activations , we used the following stratification: p < 0 . 05 FWE , p < 1e-5 uncorrected , and p < 1e-3 uncorrected . We define two types of events: OS are defined as a collection of stimulus events in which the learning rate of the model is greater than the 90th percentile , while the rest of the events are deemed to correspond to IC . The trials in which a novel cue is presented with a novel outcome amount to 10% of total trials ( 1/2 type2 round x 1/5 of novelty-matching trials = 1/10 ) , and during these trials the learning rate almost always rises to peak . Thus , we define one-shot learning threshold as the learning rate of 90th percentile . To test whether there is a functional coupling between the prefrontal area associated with uncertainty processing and the hippocampus modulated by the learning rate , we performed a physiological correlation analysis . The procedure is the same as a psychophysiological interaction analysis [70] except that the psychological variable is a combination of multiple boxcar functions , whose interval is given by the percentile of learning rate . We used the first eigenvariate of BOLD signals from the ventrolateral prefrontal cortex extracted from a 5-mm sphere centered at the coordinates of the cluster identified as correlating with causal uncertainty ( S2 Table ) . The extracted BOLD signal was deconvolved in order to retrieve the underlying neuronal signal . The deconvolved signal was then used as a parametric regressor for the GLM analysis . The onset times for this first parametric regressor correspond to a collection of events during which the learning rate is between the 1st and 10th percentile . After performing the GLM analysis , the average beta value is computed within the anatomically defined bilateral hippocampus ROI [54] . This analysis is repeated for each size bin ( 10th percentile ) . The average beta value represents correlation between the neural activity of the ventrolateral prefrontal cortex and hippocampus during the events whose intervals are given by binned percentile of learning rate . To see if our fMRI findings are an artifact of the order in which the regressors had been entered into the fMRI design matrix because of serial orthogonalization , we ran another GLM analysis in which the main regressors are not orthogonalized with respect to each other ( i . e . , by disabling serial orthogonalization ) . The results are summarized in S8 Fig . All of the results are highly consistent with what we have reported in our main results and survive corrected thresholds . This indicates that our main results are very robust to orthogonalization order . The followings are the list of the models used for model comparison ( refer to S1 Table for full details ) : | There are at least two distinct learning strategies for identifying the relationship between a cause and its consequence: ( 1 ) incremental learning , in which we gradually acquire knowledge through trial and error , and ( 2 ) one-shot learning , in which we rapidly learn from only a single pairing of a potential cause and a consequence . Little is known about how the brain switches between these two forms of learning . In this study , we provide evidence that the amount of uncertainty about the relationship between cause and consequence mediates the transition between incremental and one-shot learning . Specifically , the more uncertainty there is about the causal relationship , the higher the learning rate that is assigned to that stimulus . By imaging the brain while participants were performing the learning task , we also found that uncertainty about the causal association is encoded in the ventrolateral prefrontal cortex and that the degree of coupling between this region and the hippocampus increases during one-shot learning . We speculate that this prefrontal region may act as a “switch , ” turning on and off one-shot learning as required . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Neural Computations Mediating One-Shot Learning in the Human Brain |
As a component of the Cytosolic Iron-sulfur cluster Assembly ( CIA ) pathway , DRE2 is essential in organisms from yeast to mammals . However , the roles of DRE2 remain incompletely understood largely due to the lack of viable dre2 mutants . In this study , we successfully created hypomorphic dre2 mutants using the CRISPR/Cas9 technology . Like other CIA pathway mutants , the dre2 mutants have accumulation of DNA lesions and show constitutive DNA damage response . In addition , the dre2 mutants exhibit DNA hypermethylation at hundreds of loci . The mutant forms of DRE2 in the dre2 mutants , which bear deletions in the linker region of DRE2 , lost interaction with GRXS17 but have stronger interaction with NBP35 , resulting in the CIA-related defects of dre2 . Interestingly , we find that DRE2 is also involved in auxin response that may be independent of its CIA role . DRE2 localizes in both the cytoplasm and the nucleus and nuclear DRE2 associates with euchromatin . Furthermore , DRE2 directly associates with multiple auxin responsive genes and maintains their normal expression . Our study highlights the importance of the linker region of DRE2 in coordinating CIA-related protein interactions and identifies the canonical and non-canonical roles of DRE2 in maintaining genome stability , epigenomic patterns , and auxin response .
Iron-sulfur ( Fe-S ) proteins , which have Fe-S clusters mainly ligated to their cysteine residues , are ubiquitous in all three domains of organisms in nature: Bacteria , Archaea , and Eukarya . Depending on Fe-S clusters as cofactors , which play roles in electron transport , catalysis and regulation of gene expression , Fe-S proteins participate in diverse biological processes including respiration , photosynthesis , amino acid and purine metabolism , DNA replication and repair [1 , 2] . The maturation of Fe-S proteins involves synthesis of Fe-S clusters first and then assembly of Fe-S clusters on recipient proteins . Three pathways have evolved in eukaryotes for the maturation process , including the SUF ( Sulfur mobilization ) pathway in the plastids , the ISC ( Iron-Sulfur Cluster assembly ) pathway in the mitochondria and the CIA ( Cytosolic Iron-sulfur cluster Assembly ) pathway in the cytoplasm [3–6] . The maturation of cytoplasmic and nuclear Fe-S proteins is accomplished through a part of the ISC pathway and the CIA pathway . In yeast , the reactions carried out by the cysteine desulfurase complex Nfs1-Isd11 in the ISC pathway produce persulfide [7] , which is transported from the mitochondria to the cytoplasm , in the form of glutathione trisulphide , by the ATP-binding cassette transporter Atm1 [8] and serves as the source of sulfur . Iron is provided via unknown mechanisms . In the cytoplasm , Fe-S clusters are assembled on the P-loop NTPases Cfd1–Nbp35 heterotetramer scaffold complex [9–11] . The newly synthesized iron-sulfur cluster is then transferred to recipient proteins with the assistance of the CIA targeting complex , which is composed of Cia1 , Cia2 and Met18 [5 , 12–15] . Nar1 interacts with both Cfd1–Nbp35 scaffold complex and the CIA targeting complex and likely couples the two steps of Fe-S protein maturation [5 , 16] . Plants lack the homolog of yeast Cfd1 and NBP35 performs the scaffolding function in a homodimeric form . However , plant homologs of other yeast ISC and CIA proteins , including NFS1 [4] , ISD11 [4] , ATM3 ( homolog of yeast Atm1 ) [17 , 18] , CIA1 [4] , AE7 ( homolog of yeast Cia2 ) [19] , MET18 [20–22] and NAR1 [23] , are present and functional studies revealed that ATM3 , AE7 and MET18 act in the CIA pathway as their yeast counterparts . Dre2 ( Derepressed for Ribosomal protein S14 Expression , also called CIAPIN1 ) was also identified as a component important for the CIA pathway in yeast [24] . Upstream of Nbp35 , Dre2 forms a complex with the diflavin reductase Tah18 . Electron transfer from NAPDH to Dre2 provides reducing power to Fe-S cluster assembly [25] . Dre2 itself contains a [2Fe-2S] cluster and a [4Fe-4S] cluster [24 , 26] , which cannot be transferred to Fe-S proteins [25] . The assembly of Fe-S clusters on Dre2 does not rely on Tah18 , Nbp35 , Nar1 and Cia1 [25] . Studies in human cells revealed that the [2Fe-2S] clusters of anamorsin , human homolog of Dre2 , can be transferred from a protein complex comprising human cytosolic monothiol glutaredoxin-3 ( GRX3 ) and BOLA2 [27 , 28] . Plant homologs of yeast Tah18 , Dre2 and human GRX3 are ATR3 [29] , DRE2 and GRXS17 [30] , respectively . DRE2 interacts with GRXS17 [30] , ATR3 [29] , and NBP35 [31] . However , CIA-related functional studies of DRE2 in plants remain undone . Intriguingly , it was found that epigenetic activation of maternal FLOWERING WAGENINGEN ( FWA ) in the central cell and endosperm , requires DRE2 , but not other CIA proteins , and DRE2 also localizes in the nucleus [32] , suggesting that DRE2 plays non-CIA roles . DRE2-null mutants , like null mutants for other CIA proteins , except MET18 , are embryonic lethal and no viable alleles of DRE2 have previously been found . However , study of CIA-related functions of DRE2 and discovery of novel non-CIA roles of DRE2 demand viable alleles . In this study , we created hypomorphic Arabidopsis mutants for DRE2 using the CRISPR/Cas9 system . Our genetic and biochemical evidence supports that the CIA-dependent role of DRE2 is important for the maintenance of genome and epigenome stability . Importantly , we find that DRE2 is involved in auxin response that may be independent of its CIA role . Our study reveals multiple CIA-dependent and -independent functions of DRE2 in plants .
Because complete loss of DRE2 functions leads to embryonic lethality [32] , which precludes investigation of DRE2’s functions , we set out to create dre2 hypomorphic mutants . For this purpose , we designed sgRNA1 to sgRNA3 targeting the first exon ( corresponding to the N-terminal Methyltransferase-like domain [33] ) , the sixth exon ( corresponding to the C-terminal CIAPIN1 domain [33] ) and the junction of the fourth intron and exon ( corresponding to the start of linker region ) , respectively ( S1A and S1B Fig ) . Mutant lines carrying sgRNA1 or sgRNA2 developed chlorotic leaf spots or stripes reminiscent of cell death ( S1C Fig ) . However , we were unable to obtain homozygous lines , indicating that these homozygous lines have lethal defects . Fortunately , we obtained two homozygous dre2 mutant lines from the transformants expressing sgRNA3 . The first line , named as dre2-3 , harbors a 27-nucleotides deletion in the fourth exon ( Fig 1A ) , which causes a deletion of the amino acids ‘IKAKKPSWK’ ( Fig 1B ) . The second line , named as dre2-4 , harbors a deletion of the 3′ splicing site ‘AG’ in the fourth intron ( Fig 1A ) , which disrupts splicing of the DRE2 mRNA ( Fig 1C ) . We cloned DRE2 cDNA from dre2-4 . Seven splicing variants were identified from 50 clones we sent for sequencing . Two of them are deletion variants which deletes amino acids ( DRE2Δ6 and DRE2Δ75 ) . The rest of them all contain premature stop codons , which are resulted from either deletion ( DRE2Δ85 , DRE2Δ112 and DRE2Δ127 ) , insertion ( DRE2+20 ) or one intron-retention ( DRE2+221 ) ( Fig 1D; S1F and S1G Fig ) . No normal DRE2 transcripts was identified from those clones , suggesting that the primary 3′ splicing site ‘AG’ is important for correct splicing of DRE2 mRNA ( S1F Fig ) . DRE2 transcripts are slightly decreased in dre2-3 , while reduced to about 40% in dre2-4 ( Fig 1E; S1D Fig ) . Like mutant lines carrying sgRNA1 or sgRNA2 ( S1C Fig ) , dre2-4 has chlorotic mosaics and wrinkled leaves ( S1E Fig ) . To verify that the CIA-dependent function of DRE2 is at least partially compromised , we analyzed the activities of aldehyde oxidases ( AO ) [17] , a representative Fe-S containing enzyme , in dre2-3 and dre2-4 . In-gel AO activity staining revealed that the AO activities are indeed lower in dre2-3 and dre2-4 than that in Col-0 ( Fig 1F ) . However , we did not find seed abortion in dre2-3 and dre2-4 . We further introduced the pFWA::ΔFWA-GFP reporter into dre2-4 . The expression of ΔFWA-GFP in the central cell and endosperm was normal in dre2-4 ( S2 Fig ) , suggesting that dre2-4 does not affect epigenetic activation of FWA . Thus , dre2-3 and dre2-4 may only lose some of DRE2’s functions . The CIA pathway mutants , including the ae7 , met18 and grxs17 mutants , are characteristic of a compromise in DNA repair and accumulation of DNA damage [19 , 20 , 22 , 30] . To test whether the dre2 mutants have the same defects , we conducted comet assays using the alkaline unwinding/neutral electrophoresis ( A/N ) protocol , which measures DNA SSBs as well as double strand breaks ( DSBs ) ( Fig 2A ) . Based on relative signal intensities of the comet tail , we classified a broad spectrum of DNA damage into four levels: 1% , 10% , 30% and 50% . The majority of cells had 1% DNA damage and less than 2% cells had 30% and higher level of DNA damage in Col-0 . However , the percentages of cells with 1% DNA damage dramatically decreased and the percentages of cells with 30% and 50% DNA damage increased to 65% or higher in dre2-3 , dre2-4 and met18-2 ( Fig 2A and 2B ) , suggesting accumulation of DNA damage in these mutants . In line with the comet assay results , dre2-4 showed higher sensitivity to MMS , a DNA double strand break ( DSB ) -inducing chemical , than Col-0 ( S3A Fig ) . We further detected the expression levels of multiple genes involved in DNA Damage Response ( DDR ) , including PARP1 , PARP2 [34] , PCNA [35] , BRCA1 , GR1 , DMC1 , RAD51 , RAD54 , TSO2 and BARD1 [36] in Col-0 , dre2-3 and dre2-4 without or with the application of MMS . We found that PARP1 , PARP2 , PCNA , BRCA1 , DMC1 and RAD51 were activated in dre2-3 and dre2-4 even without MMS treatment , whereas the rest were activated only after MMS treatment ( Fig 2C ) . Multiple cyclin genes , including Cyclin-B1-1 , Cyclin-B1-2 , Cyclin-A1-1 , Cyclin-A2-4 and Cyclin-B2-4 were also upregulated in dre2-3 and dre2-4 ( S3B Fig ) , suggesting that the accumulation of DNA damage and activation of DDR in these mutants arrested cell cycle progression . The short root phenotype is often associated with activation of DNA damage and cell cycle arrest . We also observed that dre2-3 and dre2-4 developed shorter roots than Col-0 ( Fig 2D and 2E ) . It appeared that dre2-4 is a stronger allele compared to dre2-3 , as dre2-4 had more DNA damage , more significant upregulation of DDR and cyclin genes and more severe short root phenotype ( Fig 2 ) . To exclude the possibility that these phenotypes are off-target effects arose from the CRISPR/Cas9 strategy , we introduced the 35S::DRE2 transgene into dre2-4 . The upregulation of DDR genes and cyclin genes and the short root phenotype in dre2-4 were all rescued by overexpression of wild type DRE2 in dre2-4 ( Fig 2D and 2E; S4 Fig ) , suggesting that these phenotypes result from DRE2 dysfunction . The CIA pathway mutants , including the ae7 , met18 and nbp35 mutants , have DNA hypermethylation at specific loci due to impaired activity of ROS1 [19 , 20 , 22 , 31] , a DNA glycosylase which catalyzes the first step of active DNA demethylation [37] . To assess the effects of mutation of DRE2 , an early-acting CIA factor , on DNA methylation patterns , we performed whole genome bisulfite sequencing on dre2-4 . In total , we identified 1459 differentially methylated regions ( DMRs ) in dre2-4 , including 951 hyper-DMRs and 508 hypo-DMRs ( S5A Fig; S1 Dataset ) . Of the 951 hyper-DMRs , 633 ( 66 . 6% ) are overlapping with those in ros1-4 ( Fig 3A; S5A Fig ) . Hyper-DMRs unique to dre2-4 are mainly hyper-methylated in CG context ( Fig 3A ) . The CG methylation levels of these dre2-4 specific hyper-DMRs in ros1-4 are also higher than those in Col-0 , suggesting that hyper-methylated regions in dre2-4 and ros1-4 actually overlap to a larger extent ( Fig 3A ) . Heat maps show that the methylation levels of dre2-4 hyper-DMRs in met18-2 and ros1-4 are also higher than that in Col-0 in CG , CHG , and CHH contexts ( Fig 3B ) . Analysis of the distribution of the dre2-4 hyper-DMRs in different genomic regions ( gene body region , intergenic region , TEs out of gene region and TEs overlapping with gene region ) revealed that more than 60% of the hyper-DMRs are distributed in gene body regions in dre2-4 ( S5B Fig ) much more than that of ros1-4 . Collectively , these results suggest that mutation of the early-acting CIA factor DRE2 has a similar effect on overall DNA methylation patterns as mutation of the late-acting MET18 . We further examined whether ROS1 promoter , hypermethylation of which induces ROS1 expression [38 , 39] , is hypermethylated in dre2-4 , as in met18-2 . We found similar hypermethylation of ROS1 promoter in dre2-4 and met18-2 ( S5C Fig ) . Results of Locus-specific bisulfite sequencing confirmed this ( Fig 3C ) . As a result of ROS1 promoter hypermethylation , ROS1 transcripts are upregulated in dre2-3 , dre2-4 , met18-2 and ros1-4 ( Fig 3D ) . The upregulation of ROS1 expression was also rescued by overexpression of wild type DRE2 in dre2-4 ( S5D Fig ) . Despite upregulation of ROS1 transcripts , the DNA hypermethylation phenotype of dre2-4 exists and persists , suggesting impaired ROS1 activity . To investigate the underlying cause for defects in the dre2-3 and dre2-4 mutants , we carried out yeast two-hybrid ( Y2H ) assays to assess whether crucial protein-protein interactions are affected in dre2-3 and dre2-4 . As characterized above , a deletion mutant of DRE2 , referred to as DRE2-3 , is produced in dre2-3 , whereas , two deletion mutants of DRE2 , referred to as DRE2Δ6 and DRE2Δ75 , are produced in dre2-4 ( Fig 1 ) . Alignment of the deletion mutants with wild type DRE2 revealed that all of the deletions are located within the unstructured linker region that separate the N-terminal methyltransferase domain and the C-terminal CIAPIN1 domain [40] ( S1B Fig ) . They start from Isoleucine139 ( Ile139 ) , but they end at different lysines ( Ks ) ( S1G Fig ) . We first detected the interactions of these deletion mutants with TAH18 and NBP35 . All of the deletion mutants interacted with TAH18 , like wild type DRE2 ( Fig 4A ) . However , they showed much stronger interaction with NBP35 compared with wild type DRE2 ( Fig 4A ) . We next examined the interactions of these deletion mutants with GRXS17 . While DRE2Δ6 interacted normally with GRXS17 , DRE2-3 and DRE2Δ75 lost interaction with GRXS17 ( Fig 4B ) . This is consistent with previous findings that the unstructured linker region of anamorsin in human cells reinforces its interaction with GRX3 [27] . Together , our results suggest that the linker region in DRE2 inhibits DRE2-NBP35 interaction but promotes DRE2-GRXS17 interaction . To uncover more biological functions of DRE2 , we performed RNA-seq to identify the differentially expressed genes between 12-day-old Col-0 and dre2-4 seedlings . Three biological replicates were performed and over 50 M high quality reads were obtained from the RNA library constructed from each sample . More than 90% of the reads could be uniquely mapped to the Tair10 Arabidopsis genome . In total , we identified 1593 genes that were significantly upregulated and 1111 genes that were significantly downregulated ( fold-change>2 , q<0 . 01 ) in dre2-4 as compared to Col-0 ( S2 Dataset ) . Gene Ontology ( GO ) analysis of the upregulated genes revealed 77 enriched biological processes ( FDR<0 . 05 ) , including ‘response to ionizing radiation’ , ‘response to gamma radiation’ , ‘cell cycle’ and ‘DNA replication’ ( S6A and S6C Fig ) . This is in accordance with our results that mutations in DRE2 cause accumulation of DNA damage and constitutive activation of DDR ( Fig 2 ) . GO analysis of the downregulated genes revealed 22 enriched biological processes ( FDR<0 . 05 ) , among which ‘response to auxin stimulus’ is enriched ( Fig 5A; S6B and S6D Fig ) . To confirm that DRE2 is involved in auxin responses , we first performed RT-qPCR to detect the expression levels of DRE2 and other CIA pathway components after treatment with IAA , the natural auxin . While the expression of other CIA pathway proteins were mildly induced , the expression of DRE2 was greatly induced by IAA , especially at early stage after IAA treatment ( Fig 5B ) . Secondly , we tested whether the growth of the primary root , which is inhibited by auxin , is affected in the dre2 mutants . The effect of IAA treatment on primary root length was much weaker in the dre2 mutant compared to that in wild type and met18-2 ( Fig 5C and 5D; S7 Fig ) , indicating that the dre2 mutations confer auxin insensitivity . Third , we tested whether the increase of lateral root formation , that is promoted by auxin [41 , 42] , is affected in dre2-3 , dre2-4 and met18-2 . There is no significant difference of the lateral root numbers between Col-0 and the three mutants without the application of IAA ( Fig 5C and 5D ) . IAA treatment greatly increased the lateral root numbers in Col-0 and the three mutants . The increase of lateral root number in met18-2 is comparable to that in Col-0 . However , the increase of lateral root numbers in dre2-3 and dre2-4 mutants was significantly less than that in Col-0 ( Fig 5C and 5D ) . Together , our results suggest that DRE2 participates in auxin response . This could be a CIA-independent function of DRE2 . Previous studies showed that DRE2 mainly localizes in the cytoplasm , like other CIA pathway proteins , but weak DRE2-GFP signal can be detected in the nucleus [32] . To ascertain the subcellular localization of DRE2 , we generated pDRE2::DRE2-GFP transgenic plants in Col-0 background . Consistently , we found that DRE2-GFP mainly localized in the cytoplasm in the differentiation zone of root ( S8A Fig ) . Strong nuclear DRE2-GFP signal could be detected after Leptomycin B ( an exportin inhibitor ) treatment ( S8A Fig ) , suggesting that DRE2 shuttles between the nucleus and the cytoplasm . In the meristematic zone of root , DRE2-GFP appeared to be more abundant in the nucleoplasm than in the cytoplasm ( S8B Fig ) . Unexpectedly , strong DRE2-GFP signal could be detected in the nucleolar cavity ( S8B Fig ) . We further performed subcellular fractionation experiments with HSP90 and Histone H3 as cytoplasmic and nuclear protein markers , respectively . We found that DRE2 was present in both the cytoplasm and the nucleus ( Fig 6A; S9 Fig ) . Importantly , when we separated the nuclear fraction into nucleoplasm and chromatin-bound fractions , we found that DRE2 was associated with chromatin ( Fig 6B; S9 Fig ) . To further visualize the nuclear localization of DRE2 , we immunostained nuclei isolated from Col-0 and the DRE2-HA transgenic plants . The DRE2-HA signal did not overlap with compacted heterochromatin regions that are intensely stained by DAPI , but overlapped with open euchromatin regions that are weakly stained by DAPI ( Fig 6C ) , suggesting that DRE2 is specifically associated with euchromatin . To test whether DRE2 can directly bind auxin responsive genes , we performed chromatin immunoprecipitation ( ChIP ) -qPCR using pDRE2::DRE2-GFP transgenic plants . Our results revealed that DRE2 were enriched at some auxin responsive genes , including SAUR16 , IAA14 , PIN4 and PIN7 ( Fig 6D ) . In dre2-4 mutants , the transcript levels of SAUR16 , IAA14 and PIN4 were reduced , while PIN7 remained unaltered ( Fig 6E ) . These results suggest that DRE2 binding is important for the expression of auxin responsive genes .
For most of the essential genes , no homozygous cell lines or mutants can be obtained . This impedes the functional studies of these genes . RNAi and CRISPR/Cas9 are two technologies for gene silencing , with the former knock down genes and the latter knock out genes or introduce deletions/mutations . Because RNAi only reduces the expression of genes , it is a better choice for investigating the functions of essential genes . However , RNAi-mediated reduction of gene expression sometimes is not sufficient to result in desired molecular and/or phenotypic changes . For instance , the RNAi line for NBP35 have unaltered activities of Fe-S enzymes in leaves [31] . Another drawback of RNAi is that it often has off-target effects and may produce false positive results [43] . CRISPR/Cas9 is effective for editing of specific genomic regions and has less off-target effects . Thus , it could be ideal for creation of hypomorphic mutants for essential genes . Usually , sgRNAs targeting exons are designed to introduce insertion or deletions mutations in protein-coding regions of genes . Different from the popular strategy , we designed a sgRNA targeting an intron-exon junction region of DRE2 to create mis-splicing mutants for DRE2 . We successfully created mis-splicing mutants for DRE2 and obtained viable homozygous lines . The mutant lines were found to exhibit hallmark features of the CIA pathway mutants . Moreover , we found that DRE2 participates in auxin response independently of its CIA role using these mutant lines . Our results suggest that generating viable mutant lines using CRISPR/Cas9 technology is a good strategy for studying essential genes . To clearly identify DRE2 as a CIA factor , we measured the activities of AO in dre2-3 and dre2-4 . Like the atm3 mutant [17] , dre2-3 and dre2-4 have low AO activities , suggesting that the CIA function of DRE2 is conserved in plants . Like the ae7 , met18 , atm3 and grxs17 mutants , the dre2 mutants have accumulation of DNA damage and activation of DDR . In eukaryotes , Fe-S clusters have been found to be assembled on DNA replication/repair proteins , including Pol α , δ , ε , and ζ , which are DNA polymerases for DNA replication and repair [44] , PriL , which is a subunit of eukaryotic primase for DNA replication and telomere maintenance [45 , 46] , XPD and FancJ , which are DNA helicases involved in nucleotide excision repair [47] , RTEL1 , which is a DNA helicase involved in homologous recombination [48] . The accumulation of DNA damage and activation of DDR in the ae7 [19] , met18 [15] , atm3 [19] and grxs17 mutants have been proposed to be caused by defective assembly of Fe-S clusters on plant counterparts of these proteins . It is possible that the DNA damage phenotype of the dre2 mutants is also a result of defective assembly of Fe-S clusters on DNA replication/repair proteins . Previous findings indicate that different Fe-S proteins are differentially affected by certain CIA mutation , future investigations are needed to determine which DNA replication/repair proteins are preferentially affected by DRE2 . The ae7 , nbp35 and met18 mutants have low activity of ROS1 and whole genome bisulfite sequencing analysis revealed DNA hypermethylation at hundreds of loci in met18 [19 , 20 , 22 , 31] . The dre2 mutants also exhibit DNA hypermethylation and the pattern of DNA hypermethylation in dre2-4 resembles that in met18-2 ( Fig 3B ) . Our results suggest that the action of DRE2 , an early acting CIA factor , is also required for full enzymatic activity of ROS1 . ROS1 transcript levels did not decrease in the dre2 mutants , excluding the possibility that decreased expression of ROS1 caused DNA hypermethylation . AE7 and MET18 interact with ROS1 and the physical interactions may mediate the transfer of Fe-S cluster to ROS1 and contribute to ROS1 activity [19 , 20 , 22 , 31] . As DRE2 is not a component of the CIA targeting complex , it is less likely that DRE2 directly interacts with ROS1 . Impairment of Fe-S assembly on ROS1 is less likely a result of disrupted DRE2-ROS1 interaction . As electron transfer from DRE2 provides reducing power to Fe-S cluster assembly , impairment of Fe-S assembly on ROS1 could be due to inadequate reducing power . The DRE2 protein contains a methyltransferase-like domain at the N-terminal , a CIAPIN1 domain at the C-terminal , and an unstructured linker in the middle ( S1B Fig ) . In human cells , the unstructured linker is the only region of anamorsin that tightly interacts with Ndor1 ( human homolog of Tah18 ) [49] and can stabilize the interaction with GRX3 [27 , 28] . The dre2-3 and dre2-4 mutations , we created by CRISPR/Cas9 , lead to amino acid deletions in the unstructured liker region . Consistent with studies in human cells , our Y2H results revealed that the linker is required for DRE2’s interaction with GRXS17 . However , in contrast to promoting DRE2-GRXS17 interaction , the linker region was found to inhibit DRE2’s interaction with NBP35 . The opposing functions of the linker region on DRE2-GRXS17 interaction and DRE2-NBP35 interaction lead us to propose that before mature , DRE2 exposes its linker region to facilitate DRE2-GRXS17 interaction and Fe-S cluster transfer from GRXS17 to DRE2 . Binding of Fe-S cluster and maturation of DRE2 may change the conformation of DRE2 and hide the linker region , thus disrupting DRE2-GRXS17 interaction to facilitate DRE2-NBP35 interaction and Fe-S cluster assembly on cytoplasmic or nuclear Fe-S proteins . Because the GRX3-anamorsin interaction promotes [2Fe-2S] cluster transfer [27] , loss of DRE2-GRXS17 interaction before Fe-S transfer to DRE2 in dre2-4 may lead to reduced Fe-S binding on DRE2 , eventually leading to the CIA-related defects in dre2-3 and dre2-4 . The CIA-related defects in dre2-3 and dre2-4 may also arise from stronger interaction between DRE2 and NBP35 . The CIA function of DRE2 is considered to be executed by cytoplasmic DRE2 . Previous studies found that DRE2 also localizes in the nucleus [32] . In this study , we confirmed the nuclear localization of DRE2 and found that it can associate with chromatin ( Fig 6A and 6B ) . Further immunostaining results indicate that DRE2 is associated with euchromatin ( Fig 6C ) . The nuclear localization , especially the euchromatin-association , of DRE2 suggests that DRE2 plays non-CIA roles . Previous studies revealed one of the non-CIA roles that DRE2 plays is activation of the imprinting gene FWA in the central cell and endosperm [32] . We found that dre2 mutant lines created by sgRNA1 or sgRNA2 and dre2-4 show chlorotic leaf spots or stripes reminiscent of cell death . This phenotype was detected only in the dre2 mutants , but not in other CIA pathway mutants , suggesting that DRE2 may regulate signaling pathways related to cell death independently of its CIA role . Through RNA-seq analysis , we found that auxin response was disrupted in dre2 mutants . Because only DRE2 was greatly induced by auxin treatment and only dre2 mutants , but not Col-0 and met18-2 , have alleviated inhibition of primary root growth and less increase of lateral root number , we propose that the involvement of DRE2 in auxin response might be independent of its CIA role . Our ChIP-qPCR results revealed that DRE2 directly binds multiple auxin responsive genes and DRE2 binding is required for their optimal expression . Thus , it is very likely that DRE2 participates in auxin response by controlling the expression of auxin responsive genes , although we could not exclude the possibility that the CIA-dependent function of DRE2 indirectly ( via a FeS protein ) contributes to the auxin response . Interestingly , the knockout line and RNAi line for GRXS17 also display reduced auxin sensitivity , albeit only at high temperature . Specifically , the inhibition of primary root length by auxin is alleviated in the knockout line and RNAi line for GRXS17 . This was attributed to elevated levels of reactive oxygen species in these lines [50] . Our finding that DRE2 , a downstream Fe-S transfer target of GRXS17 , participates in auxin response suggesting that malfunction of DRE2 could be a cause of altered auxin response in these lines . Together , our findings define the nuclear localization of DRE2 , its association with chromatin and its modulation of auxin responsive gene expression . It will be interesting to search for nuclear DRE2-interacting proteins and other nuclear DRE2 functions in the future .
All Arabidopsis materials used in this study are in Columbia background . The met18-2 ( SALK_147068 ) mutant has been described previously [19 , 22] . The dre2-3 and dre2-4 mutants were generated by an egg cell-specific promoter-controlled CRISPR/Cas9 system [51] using sgRNA-3 ( S1 Table ) . For horizontally-grown seedlings , surface-sterilized seeds were germinated on 1/2 Murashige-Skoog ( MS ) medium ( 7‰ Agar and 1% Suc ) at 22°C with 16 h of light and 8 h of darkness . For vertically-grown seedlings , surface-sterilized seeds were germinated on 1/2 MS medium ( 12‰ Agar and 1% Suc ) also at 22°C with 16 h of light and 8 h of darkness . The seedlings were harvested for experiments or transplanted into soil and grown at 22°C with the same photoperiod . To generate the pDRE2::DRE2-GFP and pDRE2::DRE2-HA transgenic plants , DRE2 genomic DNA fragment with its native promoter was amplified from Col-0 genomic DNA by PCR and cloned into the pCAMBIA1300 or pCAMBIA1305 vectors for plant transformation . For complementation of dre2-4 , DRE2 coding sequence ( CDS ) was amplified from Col-0 cDNA by PCR and cloned into the pCAM2300-35S-Ocs vector for plant transformation . Agrobacterium tumefaciens strain GV3101 carrying different constructs were used to transform the wild type or mutant plants via the standard floral dipping method [52] . Primary transgenic plants were selected on 1/2 MS plates containing 25 mg/L hygromycin ( pCAMBIA1300 and pCAMBIA1305 ) or 50 mg/L kanamycin ( pCAM2300-35S-Ocs ) . Homozygous transgenic lines were used for further experiments . The primers used for PCR are listed in S1 Table . Enzyme extraction and AO activity staining were performed as previously described [53–55] . Twelve-day-old seedlings were ground into fine powder and 100 mg of powder were suspended in 400 μL of extraction buffer ( 50 mM Tris-HCl ( pH 7 . 5 ) , 1 mM EDTA , 1 mM sodium molybdate , 10 μM flavin adenine dinucleotide , 2 mM DTT , protease inhibitor ( one tablet/100 mL ) and Polyclar AT ( 0 . 2 g/g fresh weight ) ) . After centrifugation at 15 , 000 rpm for 20 min , the supernatant was concentrated by filtration using Amicon Ultra-0 . 5 Centrifugal Filters ( 3K ) to achieve a final volume of 50 to 100 μL . The concentrated samples were used as crude enzyme preparations for native PAGE ( 6% ) . After electrophoresis , the gel was immersed in 0 . 1 M potassium phosphate ( pH 7 . 4 ) for 5 min , and then the activity of AO was determined by developing in a reaction mixture containing 0 . 1 M potassium phosphate buffer ( pH 7 . 4 ) , 1 mM 1-naphthaldehyde , 0 . 1 mM phenazine methosulfate , and 0 . 4 mM MTT at room temperature ( about 25°C ) for 1 h in the dark . Total RNA was extracted from 12-day-old seedlings grown on 1/2 MS medium using the TRIzol reagent ( Invitrogen , 15596026 ) . About 2 μg of total RNA was used for first-strand cDNA synthesis with the 5X All-In-One RT MasterMix ( abm , G485 ) following the manufacturer’s instructions . The cDNA reaction mixture was then diluted 10 times , and 1 μL was used as a template in a 20 μL PCR reaction with PowerUp SYBR Green Master Mix ( Applied Biosystems ) . The RNA transcript levels were determined by quantitative RT-PCR . TUB8 was used as an internal control . The primers used for PCR are listed in Table S1 . Comet assay was performed by using the Trevigen comet assay kit ( Trevigen , 4250-050-K ) following the manufacturer’s instruction . Briefly , nuclei extracted from 22-day-old rosette leaves at 1 x 105/mL were combined with molten LMAgarose at a ratio of 1:10 ( v/v ) at 37°C . Samples ( 50 μL ) were immediately pipetted onto CometSlide . The slides were placed flat at 4°C in the dark for 30 min and then immersed in prechilled Lysis Solution ( Trevigen , 4250-050-01 , ) at 4°C for 60 min . After lysis , the slides were immersed in freshly prepared Alkaline Unwinding Solution , pH >13 ( 200 mM NaOH , 1 mM EDTA ) for 20–60 min at room temperature in the dark . Electrophoresis was in Alkaline Electrophoresis Solution , pH >13 ( 200 mM NaOH , 1 mM EDTA ) at 21 volts for 30 min at 4°C . After washing in dH2O and 70% ethanol , samples were dried at ≤ 45°C for 10–15 min . DNA was stained by diluted SYBR Green I ( Trevigen , 4250-050-05 , ) in refrigerator for 5 min . The slides were dried completely at room temperature in the dark and then viewed by Olympus BX53 fluorescence microscope . The Comet Score software ( http://www . autocomet . com ) was used to evaluate the levels of DNA damage . About 3 μg genomic DNA was extracted from 14-day-old seedlings using Hi-DNAsecure Plant Kit ( TIANGEN , DP350-03 ) and then sent for bisulfite treatment , library preparation and sequencing ( illumina Hiseq 4000 , PE100 ) by the Beijing Genomics Institute ( Shenzhen , China ) . Data analysis was performed according to Wang et al . , 2016 [22] . About 100 ng of genomic DNA was modified using the BisulFlash DNA Modification Kit ( Epigentek , P-1026-050 ) according to the manufacturer’s instructions . An aliquot ( 1 μL ) of bisulfite-treated DNA was used for PCR using the TaKaRa EpiTaq HS ( for bisulfite-treated DNA ) ( Takara , R110A ) with gene-specific primers ( S1 Table ) in a reaction volume of 20 μL . The PCR products were cloned into the pClone007 Blunt vector kit vector ( TSINGKE , TSV-007B ) and at least 20 independent clones of each sample were sequenced . The CDSs of TAH18 , NBP35 and GRXS17 were amplified by PCR and then subcloned into pGBK-T7 ( Clontech , 630443 ) . The CDSs of DRE2 , DRE2-3 , DRE2Δ6 and DRE2Δ75 were amplified by PCR and then subcloned into pGAD-T7 ( Clontech , 630442 ) . For protein interaction analysis , two combinatory constructs were transformed simultaneously into the yeast strain AH109 ( Kept by our lab ) and tested for Leu , Trp , Ade , and His auxotrophy according to the manufacturer’s protocols . The primers used for PCR are listed in S1 Table . Leptomycin B treatment was performed as previously described [56] . Briefly , pDRE2::DRE2-GFP transgenic plants were grown vertically on 1/2 MS-medium plates for 5 days and then transferred to 3 mL of liquid 1/2 MS-medium ( pH 5 . 7 ) . After 48 h , Leptomycin B was added to a final concentration of 2 . 5 mM/L and incubated for 8–24 h . FWA-GFP in the central cell and endosperm was detected using an Olympus BX53 fluorescence microscope equipped with an Olympus DP80 digital camera [57] . DRE2-GFP in the root was detected using the Nikon’s modular A1+/A1R+ confocal laser scanning microscope system ( National Center for Protein Sciences at Peking University , Beijing ) . Tissues of 14-day-old DRE2-HA transgenic plants ( about 1 g ) were ground into fine powder and then suspended in 2 mL of Honda buffer ( 400 mM sucrose , 2 . 5% Ficoll , 5% Dextran T40 , 25 mM Tris-HCl ( pH 7 . 5 ) , 10 mM MgCl2 , 0 . 5% Triton X-100 , 0 . 5 mM PMSF , 10 mM β-mercaptoethanol and protease inhibitor ( 1 tablet/100 mL ) ) The sample was centrifuged at 10 , 000 g for 20 min at 4°C . The homogenate was filtered through a double layer of Miracloth twice and then centrifuged at 1 , 500 g for 5 min at 4°C . The supernatant was further centrifuged at 10 , 000 g for 10 min at 4°C and the cytoplasmic fraction ( supernatant ) was collected . The pellet ( nuclear fraction ) was washed three times with Honda buffer and one time with PBS buffer ( 1 mM EDTA , 137 mM NaCl , 2 . 7 mM KCl , 10 mM Na2HPO4 , 2 mM KH2PO4 ( pH 7 . 4 ) ) and then resuspended in 200 μL of glycerol buffer ( 20 mM Tris-HCl ( pH 8 . 0 ) , 75 mM NaCl , 0 . 5 mM EDTA , 0 . 85 mM DTT , 50% glycerol , 0 . 125 mM PMSF , 10 mM β-mercaptoethanol and protease inhibitor ( 1 tablet/100 mL ) ) . Another 200 μL of nuclei lysis buffer ( 10 mM HEPES ( pH 7 . 6 ) , 1 mM DTT , 7 . 5 mM MgCl2 , 0 . 2 mM EDTA , 300 mM NaCl , 1 M urea , 1% NP-40 , and 0 . 5 mM PMSF , 10 mM β-mercaptoethanol and protease inhibitor ( 1 tablet/100 mL ) ) was added and the sample was vortexed gently . The mixture was incubated on ice for 5 min and then centrifuged at 13 , 200 g for 2 min at 4°C . The supernatant was collected as the nucleoplasmic fraction . The pellet was washed one time with PBS buffer and collected as the chromatin-associated fraction . The cytoplasmic marker HSP90 was detected by a rabbit anti-HSP90 polyclonal antibody ( Santa Cruz Biotechnology , at-115 ) . The nuclear marker Histone H3 was detected by a mouse anti-H3 monoclonal antibody ( EASYBIO , BE7004-100 ) . The cytoplasmic , nucleoplasmic and chromatin-associated protein markers AGO1 was detected by a rabbit anti-AGO1 polyclonal antibody ( provided by Dr . Xiaoming Zhang , Institute of Zoology , CAS ) [58] . DRE2-HA was detected by a mouse anti-HA monoclonal antibody ( Sigma , H3663 ) . Immunofluorescence staining was performed using leaves from 21-day-old DRE2-HA transgenic plant as previously described [59] . Nuclei preparations were incubated with a mouse anti-HA monoclonal antibody ( Sigma , H3663 ) overnight at room temperature and then incubated with goat anti-mouse Alexa Fluor 488 secondary antibody ( Invitrogen , A-11001 ) for 2 h at 37°C . DNA was counterstained using DAPI in Prolong Gold ( Invitrogen , P36931 ) . Pictures were taken using the Nikon’s modular A1+/A1R+ confocal laser scanning microscope system ( National Center for Protein Sciences at Peking University , Beijing ) . Total RNA samples were used to generate RNA libraries for deep sequencing ( HiSeq XTEN , Illumina ) . Quality control was performed with FastQC ( v0 . 11 . 5 ) . Adapters and low quality reads were removed by cutadapt ( v1 . 11 ) . The trimmed and quality filtered clean reads were mapped to the Arabidopsis reference genome ( TAIR10 , https://www . arabidopsis . org ) under the guide of annotation from Araport11 ( https://www . araport . org ) using Tophat2 ( v2 . 1 . 1 ) . Reads were then sorted , indexed and compressed by samtools ( v1 . 5 ) and bigwig files were generated by bam2wig . py with option–u ( skip non-unique hit reads ) from RseQC ( v2 . 6 . 4 ) . Cuffdiff ( v2 . 2 . 1 ) was used to quantify the gene expression level and identify differentially expressed genes ( DEGs ) with option–u ( use 'rescue method' for multi-reads ) . DEGs were filtered in CummeRbund ( v2 . 16 . 0 ) under the criteria of FDR<0 . 05 and log2 fold change > 1 . Gene ontology ( GO ) analysis was performed by AgriGO ( http://bioinfo . cau . edu . cn/agriGO/analysis . php ) with the default parameters . Heatmaps were generated by R package ComplexHeatmap ( 1 . 12 . 0 ) . All statistical analysis analyses were performed in R ( 3 . 3 . 1 ) . Chromatin immunoprecipitation ( ChIP ) assays were performed according to published protocols [60 , 61] with minor modifications . Briefly , seedling tissues ( about 1 . 5 g ) were ground in liquid N2 and suspended in 20 mL of ChIP extraction buffer I ( 10 mM Tris-HCl ( pH 8 . 0 ) , 10 mM MgCl2 , 400 mM sucrose , 0 . 1 mM PMSF , 1 mM DTT , protease inhibitor ( 1 tablet/100 mL ) ) . The nuclei were fixed with 1% formaldehyde for 10 min at 4°C and then neutralized with 0 . 125 M glycine for 5 min . The nuclei were washed 4 times with ChIP extraction buffer II ( 10 mM Tris-HCl ( pH 8 . 0 ) , 10 mM MgCl2 , 250 mM sucrose , 1% Triton X-100 , 0 . 1 mM PMSF , 1 mM DTT , protease inhibitor ( 1 tablet/100 mL ) ) until they turned white , then layered on top of ChIP extraction buffer III ( 10 mM Tris-HCl ( pH 8 . 0 ) , 2 mM MgCl2 , 1 . 7 M sucrose , 0 . 15% Triton-X100 , 0 . 1 mM PMSF , 1 mM DTT , protease inhibitor ( 1 tablet/100 mL ) ) and centrifuged at 12000 rpm for 1 h . The purified nuclei were then suspended in nuclear lysis buffer ( 50 mM Tris-HCl ( pH 8 . 0 ) , 10 mM EDTA , 1% SDS ) , and sonicated for 28 cycles ( 30s on , 30s off ) . After centrifugation , the chromatin supernatant was diluted 10 times with dilution buffer ( 16 . 7 mM Tris-HCl ( pH 8 . 0 ) , 167 mM NaCl , 1 . 2 mM EDTA , 1 . 1% Triton X-100 ) . An anti-GFP antibody ( Abcam , ab290 ) was used for ChIP assays . After washing , the reversal of crosslinking and phenol chloroform extraction , the purified DNA was suspended in 50 μL of ddH2O and diluted 5 times . A 1 μL aliquot was used for quantitative PCR . The RNA-seq data for Col-0 and dre2-4 and the whole-genome bisulfite sequencing data for dre2-4 was deposited at NCBI ( SRP153123 ) . The Col-0 and ros1-4 whole-genome bisulfite sequencing data used were from NCBI ( SRP119887 ) . | The Cytosolic Iron-sulfur cluster Assembly ( CIA ) pathway is essential for the maturation of Fe-S proteins localized in the cytosol and the nucleus . As an important component of the CIA pathway , DRE2 is essential from yeast to mammals . To study the CIA-related functions of DRE2 and further explore novel non-CIA roles of DRE2 in Arabidopsis , we for the first time created two homozygous dre2 hypomorphic mutants using the CRISPR/Cas9 technology . The dre2 mutants exhibit hallmark features of the CIA pathway mutants indicating CIA-dependent functions of DRE2 in Arabidopsis . Unexpectedly , we find that DRE2 participates in auxin response and nuclear DRE2 directly binds multiple auxin responsive genes and regulates their expression , suggesting that DRE2 plays CIA-independent roles . Our findings significantly expand our understanding of the biological functions of DRE2 in eukaryotes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"biotechnology",
"engineering",
"and",
"technology",
"hormones",
"dna",
"damage",
"plant",
"science",
"dna",
"replication",
"plant",
"hormones",
"genetically",
"modified",
"plants",
"epigenetics",
"dna",
"seedlings",
"molecular",
"biology",
"techniques",
"plants",
"dna"... | 2019 | Canonical cytosolic iron-sulfur cluster assembly and non-canonical functions of DRE2 in Arabidopsis |
Buruli ulcer ( BU ) is a neglected tropical disease caused by Mycobacterium ulcerans . The tissue damage characteristic of BU lesions is known to be driven by the secretion of the potent lipidic exotoxin mycolactone . However , the molecular action of mycolactone on host cell biology mediating cytopathogenesis is not fully understood . Here we applied two-dimensional electrophoresis ( 2-DE ) to identify the mechanisms of mycolactone's cellular action in the L929 mouse fibroblast proteome . This revealed 20 changed spots corresponding to 18 proteins which were clustered mainly into cytoskeleton-related proteins ( Dync1i2 , Cfl1 , Crmp2 , Actg1 , Stmn1 ) and collagen biosynthesis enzymes ( Plod1 , Plod3 , P4ha1 ) . In line with cytoskeleton conformational disarrangements that are observed by immunofluorescence , we found several regulators and constituents of both actin- and tubulin-cytoskeleton affected upon exposure to the toxin , providing a novel molecular basis for the effect of mycolactone . Consistent with these cytoskeleton-related alterations , accumulation of autophagosomes as well as an increased protein ubiquitination were observed in mycolactone-treated cells . In vivo analyses in a BU mouse model revealed mycolactone-dependent structural changes in collagen upon infection with M . ulcerans , associated with the reduction of dermal collagen content , which is in line with our proteomic finding of mycolactone-induced down-regulation of several collagen biosynthesis enzymes . Our results unveil the mechanisms of mycolactone-induced molecular cytopathogenesis on exposed host cells , with the toxin compromising cell structure and homeostasis by inducing cytoskeleton alterations , as well as disrupting tissue structure , by impairing the extracellular matrix biosynthesis .
Buruli ulcer ( BU ) is a neglected tropical disease caused by Mycobacterium ulcerans infection [1] . Infection usually starts in the subcutaneous tissue and initially gives rise to non-ulcerative lesions . Histologically , increasing areas of necrosis contrast with the smaller central zone , in which acid-fast bacilli concentrate [2] during both an intracellular phase as well as extracellularly [3] , [4] . With disease progression , necrosis advances , radiating from the focus of infection and involving all cells and structures in its path [5] . If left untreated , necrosis extends to the corium and the lesion breaks down into a severe ulcer . In the ulcerative stage of the disease bacteria disseminate and become predominantly extracellular [3] , [4] , being found throughout the necrotic tissue [5] . The treatment of BU consists primarily in a lingering antibiotic protocol with a combination of rifampicin and streptomycin [6] , however surgical resection of infected skin is still necessary for advanced stages [7] . Moreover , the frequent delay in treatment seeking hampers disease management and increases morbidity [8] , with serious long-term sequelae [9] . Prevention is also difficult as little is known about disease transmission [10] , [11] , [12] , [13] , [14] and no vaccine is currently available [15] , [16] . M . ulcerans pathogenicity and the tissue damage characteristic of BU are mediated by its toxin mycolactone , a potent cytotoxic and immunosuppressive polyketide-derived macrolide [2] , [17] , [18] , [19] , [20] , [21] , [22] . Mycolactone is produced as a mixture of congeners , with one major form , which is conserved within a given geographical area [23] . Mycolactone A/B is the main variant produced by African isolates; Australian isolates produce mycolactone C [23] and the Chinese isolate MU98912 used in this study produces mycolactone D [24] . Regarding mycolactone's action , in vitro studies mainly performed in the mouse fibroblast L929 cell line have shown that the toxin diffuses passively through the plasma membrane [25] . Further studies also show that cells incubated with the toxin display a distinctive cytopathicity , characterized by early actin cytoskeleton rearrangement , cell round-up and detachment from the bottom of the well , and an arrest in the G0/G1-phase [17] , [26] , culminating in an apoptotic cell death [19] . Recently , Guenin-Macé et al . unveiled that the toxin targets the actin-cytoskeleton regulator Wiskott-Aldrich syndrome protein ( WASP ) , inducing its hyperactivation [27] , and Hall et al . described that mycolactone inhibits co-translational translocation of proteins into the endoplasmic reticulum ( ER ) , thus inhibiting the production of nearly all proteins that transit through the ER [28] . However , despite these advances , the molecular action of this toxin on the host cell biology that drives its pathogenesis is not fully understood . This work had the purpose of conducting a characterization of the proteome of mycolactone-treated cells , in order to better understand the effects of this toxin on host cell biology . At first , we performed a kinetic characterization of mycolactone's cytopathic , cytostatic and cytotoxic effects on L929 cells . Based on this , specific incubation times and toxin doses were chosen for the proteomic study by two-dimensional electrophoresis ( 2-DE ) . Functional studies were performed in both in vitro and in vivo models to verify our findings in mycolactone-exposed cells and investigate their role in BU pathogenesis . The data obtained showed that cytoskeleton and collagen biosynthesis are severely affected by mycolactone , supporting the involvement of cytoskeleton on mycolactone-induced cytopathogenicity and identifying a new activity of the toxin on the decrease of the collagen content in M . ulcerans-infected tissues .
The time- and dose-dependent kinetics of the cytostatic and cytotoxic activities of mycolactone were investigated by an integrated analysis of cell cycle and cell death in L929 cells . Doses of mycolactone were selected through a pre-screening MTS assay based on the concentrations reported for human ulcer exudates ( 0–300 ng/mL ) [29] . MTS assay showed a threshold around 15 ng/mL , above which mycolactone is progressively cytotoxic reaching a plateau at 50 ng/mL ( data not shown ) . Therefore , the range of mycolactone concentrations tested in this study was narrowed to 12 . 5–50 ng/mL . Results presented in figure 1 show that the ethanol ( vehicle ) equivalent ( <0 . 002% ) , as well as the mycolactone concentration below the threshold ( 12 . 5 ng/mL ) , had no detectable cytopathic ( rounding and detachment ) or cytotoxic effects . Cytotoxic doses of mycolactone ( >12 . 5 ng/mL ) induced detachment , cell cycle arrest in G0/G1 phase at 48 h and 72 h of treatment and the appearance of a sub-G0/G1 population , compatible to apoptotic cells , more evident at 72 h ( figure 1A ) . Consistent with cell cycle data , annexin-V/PI assays revealed an annexin-V+/PI− population for the highest mycolactone concentrations ( 25 and 50 ng/ml ) at 72 h ( figures 1B ) , indicative of apoptotic cells . It was previously reported that cells incubated with mycolactone re-grow when mycolactone is removed from the medium , indicating that mycolactone's effect might be reversible [17] . To further investigate the reversibility of mycolactone's effect , cells were incubated with different concentrations of mycolactone for 48 h and afterwards washed and incubated in fresh media for an extra 48 h period ( figure 1 , 48 h+48 h ) . We found that cells that had previously been incubated with the lowest cytotoxic concentration of the toxin ( 25 ng/mL ) re-adhere , recover the normal cell cycle progression ( figure 1A , 48 h+48 h ) and appeared to overcome the cytotoxic stimulus , since no increase in the sub-G0/G1 or annexin-V+/PI− populations was observed ( figures 1A and 1B , 48 h+48 h ) . On the other hand , cells that had been previously incubated with the highest cytotoxic dose ( 50 ng/mL ) , while remaining in suspension , were not able to overcome the initial stress induced by mycolactone and became committed to death ( figures 1A and 1B , 48 h+48 h ) . These data demonstrate that the reversibility of mycolactone's effect occurs within a window of concentrations around 25 ng/mL . Overall , within the range of studied mycolactone concentrations , we found doses that did not induce observable cytotoxic effects , doses that induced a reversible stress , and doses that irreversibly triggered an apoptotic cell death . To further characterize our model , the kinetics of the cytopathic effects , namely cytoskeleton alterations and cell round-up and detachment , were also assessed . In cells incubated with mycolactone , we observed not only the previously described alterations for actin [26] , but also changes in the tubulin cytoskeleton , which appeared bended into a microtubule hank ( figure 2A ) . Within 12–18 h of exposure to the toxin , actin ultrastructures ( stress fibers and lamellipodia ) were lost , and , although still attached , most of the cells were completely round-up by 18–24 h ( figures 2A and 2B ) . At 24 h , half of the cells were already in suspension , while the remaining cells eventually detached in the following 12 h [26]: detachment being probably a consequence of cell round-up and loss of adhesion structures . To study the effects of mycolactone on host cell biology , the total cellular proteome from mouse fibroblast L929 cells incubated with 50 ng/mL of mycolactone ( dose that triggers a commitment to apoptotic cell death ) or the ethanol equivalent ( control ) , was separated by 2-DE . To establish a temporal perspective of mycolactone's action , three incubation times were chosen: 24 h ( when cells are detaching , but viable and mounting a response to the mycolactone-induced stress , associating with the first detectable consequences of mycolactone on the fibroblast proteome ) ; 48 h ( when cells become committed to death , coinciding with the onset of an apoptotic population ) ; and 48 h+48 h , ( when most of the cells are in the process of apoptotic cell death , with only 40% of viability ) ( figure 1 ) . The comparison of control cells proteome at different time-points did not reveal any significantly changed spots , showing that ethanol ( vehicle ) represents a suitable control . In contrast , the comparison between the proteome of control and mycolactone-treated cells at each time-point revealed a time-dependent increase in the number of changed spots , with 4 spots changed at 24 h , 10 at 48 h and 20 at 48 h+48 h . All 20 spots were identified by mass spectrometry and found to correspond to 18 proteins comprising 5 up- and 13 down-regulated ( figure 3 and figure 4 ) . To reveal the cellular processes altered by mycolactone , the proteins were clustered into functional groups according to UniProt database ( www . uniprot . org ) . The major groups comprised ( i ) cytoskeleton-related proteins ( Dync1i2 , Cfl1 , Crmp2 , Actg1 , Stmn1 ) ; ( ii ) stress response proteins ( Hspa1b , Uba52 ) ; and ( iii ) collagen biosynthesis enzymes ( Plod1 , Plod3 , P4ha1 ) ( figure 4 and figure S1 ) . This reveals that mycolactone results in an alteration of cytoskeleton-related proteins and down-regulation collagen biosynthesis enzymes . On the other hand , stress response proteins were up-regulated . An additional group ( caspase targets ) was created clustering proteins identified as caspase substrates ( Fdps , Psme3 , Btf3 ) through the web caspase substrates database CASBAH ( www . casbah . ie ) [30] . Consistent with being caspase substrates , these proteins were only decreased in the last time point ( 48 h+48 h ) when most of the cells were undergoing an apoptotic death process ( figure 1 ) . Four additional proteins were classified as miscellaneous proteins ( Prdx4 , BSA , Unc119b , Ftl1 ) ( figure 4 and figure S1 ) . Overall , the proteome of mycolactone-treated cells revealed that the intracellular structure ( cytoskeleton ) and the extracellular matrix ( collagen ) are severely affected by the M . ulcerans toxin . The proteomic study revealed several regulators and structural components of both microfilaments and microtubules affected by mycolactone treatment after 24 h ( Dync1i2 , Cfl1 ) , 48 h ( Crmp2 , Actg1 ) and 48 h+48 h ( Stmn1 ) ( figure 4 and figure S1 ) . Cytoplasmic dynein 1 intermediate chain 2 ( Dync1i2 ) is a non-catalytic subunit of the microtubule-associated molecular motor dynein , which is involved in the transport of elements of the Golgi apparatus , endosomes and lysosomes [31] . The here detected early ( 24 h ) down-regulation of Dync1i2 ( spot 1 ) suggests that this transport may be compromised in mycolactone-treated cells . Three other proteins altered in cells treated with the toxin are cytoskeleton regulators ( Cfl1 , Crmp2 , Stmn1 ) . Cofilin 1 ( Cfl1 , spot 4 ) , a well-established regulator of actin dynamics , promotes microfilament assembly or disassembly depending upon the concentration of Cfl1 relative to actin and other actin-binding proteins , as well as upon its phosphorylation status [32] . Collapsin response mediator protein 2 ( Crmp2 ) , which was identified in two spots ( spot 6 and 7 ) , is a multifunctional adaptor protein which can induce microtubule assembly by binding to αβ-tubulin heterodimers [33] , whereas stathmin ( Stmn1 , spot 18 ) has been described as a microtubule-destabilizing oncoprotein [34] . Interestingly , the isoelectric points ( pIs ) of spots 4 ( Cfl1 ) and 6 ( Crmp2 ) differed in the gel from their expected theoretical values ( figure 3 and figure 4 ) , suggesting posttranslational modification such as phosphorylation . Given that the regulatory activity of Cfl1 [32] and Crmp2 [33] can be modulated by phosphorylation , we studied the phospho-status of these proteins with the Pro Q Diamond phosphostaining . Indeed , the analysis of the phosphoprotein stained gel revealed that both spots were phosphorylated ( figure S2 ) . Thus , mycolactone increases the phospho-Cfl1 at 24 h , decreases both phosphorylated and non-phosphorylated forms of Crmp2 at 48 h , and decreases Stmn1 in the latest time point ( 48 h+48 h ) . In addition to the alteration of cytoskeleton regulators , the proteomic study revealed that the cytoskeleton is also altered on its structural components . Actin gamma ( Actg1 ) , a component of microfilaments , is down-regulated at 48 h in mycolactone-treated cells . Overall , these results show that cells exposed to mycolactone undergo a process of cytoskeleton remodeling involving regulators and structural components , providing a novel molecular basis for the effect of mycolactone on this organelle . Two stress response proteins ( Hspa1b , Uba52 ) were up-regulated upon treatment with mycolactone . Spot 10 ( figure 4 and figure S1 ) was identified as a fusion protein ( Uba52 ) consisting of N-terminal ubiquitin and C-terminal 60S ribosomal protein L40 . The detected spot position in the gel ( figure 3 ) in comparison with the theoretical positions for the fusion protein ( pI 9 . 87/14 . 7 kDa ) , ubiquitin ( pI 6 . 56/8 . 6 kDa ) and the ribosomal protein ( pI 10 . 32/6 . 2 kDa ) suggest the presence of ubiquitin . Indeed , all three spot-specific peptides covered amino acids 13–55 revealing that ubiquitin is present . The here detected increase of free ubiquitin after 48 h of mycolactone treatment could result from an inhibition of ubiquitin ligases or from an up-regulation of its expression . To investigate this in more detail , protein ubiquitination was studied by western blot , which revealed that mycolactone exposure results in an increase of ubiquitinated proteins , more evident at 48 h and 48 h+48 h ( figure 5A ) . Therefore , these data show that rather than an inhibition of ubiquitin ligases , mycolactone induces an up-regulation of the ubiquitin/proteasome system ( UPS ) . The UPS and the lysosomal degradation system ( autophagy ) are the two main cellular degradative pathways . These systems crosstalk each other and the up-regulation of one may occur in response to a down-regulation/dysfunction of the other . Autophagy is known to be dependent on microtubule cytoskeleton [35] and dynein-driven transport [36] with dynein playing a role in the delivery of autophagosome contents to lysosomes during autophagosome-lysosome fusion [36] . Since microtubules ( Figure 2 ) and dynein ( figure 4 and figure S1 ) were found to be affected by mycolactone , we hypothesized that the mycolactone-induced cytoskeleton-related changes might impair the autophagic process leading to the up-regulation of the UPS . Therefore the role of mycolactone in autophagy was further investigated . During autophagy the cytosolic form of LC3 ( LC3-I ) is conjugated to phosphatidylethanolamine ( PE ) to form LC3-PE ( LC3-II ) , which is recruited to autophagosomal membranes . As the autophagosomes fuse with lysosomes to form autolysosomes , LC3-II is degraded together with the intra-autophagosomal components by lysosomal hydrolases . Thus , lysosomal turnover of the autophagosomal marker LC3-II reflects autophagic activity [37] . Processing of this marker was analyzed by western blot and immunofluorescence . Western blot revealed an increase of the autophagosome marker LC3-II in cells treated with mycolactone ( figure 5B ) compatible with an inhibition of autolysomes formation . In agreement , an increase of LC3-positive cytoplasmic vesicles upon toxin exposure was also detected with the immunofluorescence assay ( figure 5C , red-stained ) . To further understand these results , we treated L929 cells with different stimuli and added bafilomycin A1 to inhibit the autolysosomal degradation step [38] 2 hours before protein extraction ( figure 5D ) . The increase of LC3-II induced by mycolactone ( figure 5B and 5C ) , together with the lack of difference in LC3-II in cells treated with mycolactone in the presence or absence of bafilomycin A1 ( figure 5C ) , suggests a block of autophagy at the terminal stages [39] . Furthermore , the higher LC3-I levels observed in cells exposed to mycolactone , when compared with cells where autophagy was induced by rapamycin ( figure 5C ) , suggests that most probably autophagy is being induced due to cell detachment [40] or as a feedback response to the blockage of the autophagic terminal stage . These data indicate that mycolactone inhibits autophagosome-lysosome fusion and in turn impairs autophagy . Taken together , this reveals that mycolactone mediates up-regulation of the UPS and inhibition of autophagy . Since autophagy counteracts several stresses , including infection by intracellular pathogens [41] , [42] , [43] , mycolactone-induced impairment of autophagy may have implications for the progress of M . ulcerans infection . Proteomics identified several enzymes of collagen biosynthesis progressively down-regulated in mycolactone-treated cells: Plod1 ( 24 h ) , Plod3 ( 48 h ) and two isoforms of P4ha1 ( 48 h+48 h ) . Further studies showed that these proteins were transcriptionally down-regulated after 24 h of mycolactone exposure ( figure S3 ) , thus the differential down-regulation of the different proteins probably reflects different protein stability . These enzymes catalyze the hydroxylation of lysine ( Plod1 and Plod3 ) and proline ( P4ha1 ) residues , which is essential for the formation and stabilization of collagen fibers [44] , [45] . The here detected down-regulation of these enzymes suggests that collagen fibers stability may be compromised in mycolactone exposed cells . Interestingly , histopathological studies from the 1960's reported a collagen decrease in human BU lesions [5] , [46] , [47]; however this feature has been overlooked and it was never subject to studies to determine its cause or its relevance for BU . To investigate if mycolactone was responsible for a decrease in tissue collagen , an experimental model of BU disease , the mouse footpad infection with M . ulcerans , was used . Mice were challenged with virulent mycolactone-secreting ( MU98912 ) , avirulent mycolactone-negative ( MU5114 ) strains of M . ulcerans or PBS as a control . Pathology progression was assessed by measuring footpad swelling ( figure 6A ) and , at different time-points , footpads were collected for histological processing and collagen scoring ( figure 6B ) . The results showed that , in footpads infected with mycolactone-secreting M . ulcerans ( MU98912 ) , the progressive increase of pathology ( figure 6A ) was associated with a decrease of the collagen score ( figure 6B ) , preceding the breakdown of the lesion into an ulcer ( by day 40 post-infection ) . In contrast , infection with the mycolactone-negative strain ( MU5114 ) did not induce pathology nor did it alter the collagen content in infected footpads , similar to what was observed for the PBS-injected control group ( figure 6A and 6B ) . These results suggest that the decrease in collagen content is not a consequence of the infection or the elicited immune response , but rather caused by mycolactone . To verify this , mice were challenged with purified toxin or ethanol equivalent as control . The results showed that mycolactone induced footpad swelling associated with a decrease of the collagen score , while the vehicle did not ( figure 6D and 6E ) . Histological samples stained with Masson's trichrome showed a decay of collagen fibers in MU98912-infected or mycolactone-treated footpads , characterized by the disorganization and thinness of collagen fibers ( figure 6C and 6F ) . These results in the mouse model of infection show that the earlier described degeneration of collagen in BU lesions [5] , [46] , [47] , [48] is a consequence of the secretion of mycolactone by the infecting strain .
Even though , the methodology used has some limitations , since it excludes the analysis of transmembrane and secretory proteins , which were found to be severely affected by mycolactone [28] , this work is the first proteomic study on the effect of mycolactone on cells , unveiling important information about the toxin action . It has been known for years that the actin-cytoskeleton of mycolactone-treated cells suffers early structural rearrangements [26] . Recently , it was also shown that these changes were mediated by the mycolactone-induced hyperactivation of the actin-cytoskeleton regulator WASP [27] . In this study we show that mycolactone causes structural changes in microtubules and we identify several regulators and structural components of both microfilaments and microtubules affected by the M . ulcerans toxin . These data confirm the cytoskeleton as a major target of mycolactone and further specifies the mechanisms of the toxin's cellular action . Moreover , given the cytoskeleton's dynamic nature , with constant remodeling , it remained unclear how these changes contribute to the tissue damage characteristic of BU lesions . Since the proteomic data pointed likewise to an involvement of the UPS further experiments were performed confirming its mycolactone-dependent up-regulation . UPS and autophagy constitute the main intracellular processes of protein degradation taking part in the cellular protein quality control system . Thus , UPS and autophagy are critical in the maintenance of cellular homeostasis and their activities are strictly orchestrated . Moreover , perturbations in the flux through either pathway have been reported to affect the activity of the other system , and a number of mechanisms have been proposed to rationalize the link between the UPS and autophagy [49] . Therefore , it was investigated if the detected mycolactone-dependent changes affect autophagy . The here obtained data indicate that autophagosome-lysosome fusion is impaired in mycolactone-treated cells . Given that the delivery of autophagosome contents to lysosomes is dependent on microtubule cytoskeleton [35] and on dynein-driven transport [36] , the mycolactone-induced impairment of autophagy appears to occur secondarily to mycolactone-induced cytoskeleton alterations . Further evidence of a dysfunctional vesicle-lysosome fusion is given by another altered protein in mycolactone-treated cells . Proteomics revealed that the cell culture medium constituent BSA ( clustered on miscellaneous proteins ) was increased in cells treated with mycolactone . As degradation of extracellular proteins occurs in lysosomes [50] , the observed BSA accumulation suggests likewise that the endosomes-lysosome fusion may be compromised . Since the delivery of autophagosome to lysosomes [36] and retrograde transport of endosomes [31] is mediated by dynein the here observed down-regulation of one of its components ( Dync1i2 ) , suggest an impairment of dynein-driven transport upon mycolactone exposure . Further work is needed to explore the effect of mycolactone on cytoskeletal motors-mediated transport , however the down-regulation of a molecular motor subunit ( Dync1i2 ) together with the microfilaments' and microtubules' architectural changes , induced by the toxin , hint at a dysfunctional cytoskeletal motors-mediated transport within mycolactone-treated cells . Other cytoskeleton dependent functions , like phagocytosis [51] , [52] , cell motility [27] and cell shape [17] are also described to be impaired in mycolactone treated cells . Thus , these evidences imply that mycolactone induces a nonfunctional cytoskeletal-architecture , affecting cytoskeleton-dependent functions , with consequences for cellular homeostasis . Moreover , our proteomic study revealed several regulators and structural constituents of both actin- and tubulin-cytoskeleton ( Cfl1 , Crmp2 , Stmn1 and Actg1 ) affected by mycolactone . These alterations may reflect a cell feedback response to the abnormal cytoskeletal architecture as an attempt to restore the physiological cytoskeletal conformation and dynamics . In particular , the early alterations found on cofilin , a well-known regulator of actin dynamics [32] , and on dynein , recently found to play a role in the production of normal bundled stress fibers [53] , may represent an immediate cellular response to actin polymerization mediated by mycolactone-induced WASP hyperactivation [27] . Thus , a growing body of evidences supports a model in which the cytoskeletal disarrangement induced by mycolactone impairs multiple cytoskeleton-dependent cellular functions with cytotoxic consequences for the host cells . These cytoskeletal changes might have also implications early on infection , during the M . ulcerans intracellular phase , when the pathogen has to survive and proliferate inside the host cell [3] , [4] . Autophagy is being increasingly recognized as an important component of immunity , playing specific roles in shaping the immune system development , fuelling host innate and adaptive immune responses , and directly controlling intracellular microbes as a cell-autonomous innate defense mechanism . As an evolutionary counterpoint , intracellular pathogens have evolved to block autophagic microbicidal defenses and subvert host autophagic responses for their survival or growth [41] , [42] . Importantly , studies have implicated autophagy in the control of many pathogenic bacteria [43] , from which M . tuberculosis [54] should be highlighted here due to its genetic proximity to M . ulcerans . Thus , the mycolactone-induced impairment of autophagy , mediated by its action over the cytoskeleton , might represent a virulence mechanism of M . ulcerans to impair host cell immunity against intracellular pathogens . One of the main findings of this work is the identification of a novel activity of mycolactone , with the demonstration of its role in the decrease of collagen content in M . ulcerans-infected tissues . Collagen decrease in human BU lesions was described in the 1960's , in the first histopathological studies of this disease [5] , [46] , [47] , however , this phenomenon has been overlooked , even when , more recently , Guarner et al . described this feature as one of the most reliable criteria for the histopathological diagnosis of BU [48] . This previously unappreciated feature of the disease was never subject to studies to determine its cause or its relevance for BU progression and associated sequelae . Here we link , for the first time , the activity of mycolactone with the collagen reduction in M . ulcerans-infected tissues . Our results from the in vivo model show that collagen decrease is not a consequence of the infection or the immune response , but of the presence of mycolactone . In fact the inoculation of the purified toxin shows an association between decrease of collagen and the presence of mycolactone . In vitro , we showed that in L929 cells mycolactone transcriptionally down-regulated several ER-resident collagen-modifying enzymes ( Plod1 , Plod3 and two isoforms of P4ha1 ) . Additionally , Hall et al . described a post-transcriptional mechanism in which mycolactone blocks co-translational translocation of proteins into the ER , thus inhibiting the synthesis of the majority of ER-resident ( like the collagen-modifying enzymes ) and secretory proteins ( like extracellular matrix proteins ) [28] . Although these mechanisms have not been verified in vivo , altogether , these data suggest that mycolactone inflicts a transcriptional and post-transcriptional inhibition of the collagen biosynthesis pathway , which translate into a degeneration of collagen fibers in mycolactone-exposed tissues . Our data also show that the mycolactone-induced collagen degeneration precedes the breakdown of the lesion into an ulcer , suggesting that collagen decrease may be involved , together with cell death , in the tissue destructuration that culminates in the emergence of an M . ulcerans-induced ulcerative lesion . Moreover , it may be a mechanism of pathogen dissemination , given that in early lesions bacteria concentrate in a smaller central zone , while in advanced lesions bacilli are dispersed throughout the necrotic area . Finally , this collagen decay in BU lesions may also be implicated in the development of the sequelae characteristic of this devastating skin disease . BU has a very high morbidity rate associated with contractures [9] . Wound contraction is a natural mechanism by which open wounds close during the healing process , but also results in significant tissue distortion with loss of joint mobility and cosmetic disfigurement . Although the mechanism of wound contraction is not fully understood , it is associated with the abnormal generation of thicker collagen fibers [55] , [56] , [57] . Therefore , it is conceivable that during the healing process , fibroblasts and myofibroblasts repopulating the lesion overcome the collagen-deficiency through abnormal- or over-production of collagen leading to the extreme contractures characteristic of BU [58] . In fact , a recent paper by Andreoli et al . described an increase in activated myofibroblasts and an abundant production of extracellular matrix proteins in antibiotic-treated BU lesions [59] . Further studies are needed to test this hypothesis , but if proven correct the use of collagen-based materials as a bed for the skin graft , or even as a replacement in smaller legions , may decrease the contracture and thus the morbidity in BU patients . Overall , our results provide molecular and functional evidence of the impact of mycolactone on the cytoskeleton and cytoskeleton-dependent cellular functions , and extend our knowledge on the action of the M . ulcerans toxin to collagen biosynthesis , providing new perspectives on BU pathogenesis and paving the way for future therapeutic approaches .
M . ulcerans strains were selected from the Institute of Tropical Medicine collection in Antwerp , Belgium . MU5114 is a mycolactone-negative strain due to repeated subculturing , leading to the spontaneous loss of genes involved in mycolactone synthesis [23] , [60] . MU98912 is highly virulent for mice [22] and produces mycolactone type D [24] . The isolates were grown on Middlebrook 7H9 medium ( Becton , Dickinson and Company ) with 1 . 5% of agar at 32°C for approximately 6–8 weeks . For the preparation of the inoculum , M . ulcerans was recovered , vortexed using glass beads and diluted in phosphate-buffered saline pH 7 . 4 ( PBS ) to a final concentration of 1 mg/ml . Protocol for mycolactone extraction/purification was adapted from the one previously described [17] . Briefly , MU98912 was cultured in Dubos medium supplemented with 10% oleic acid-albumin-dextrose complex , at 32°C . At late exponential growth phase , bacteria were harvested and lipids were extracted with chloroform and methanol ( 2∶1 ) for 4 hours . The organic phase was separated from bacterial debris and hydrophilic components by addition of a 0 . 2 volume of water , followed by centrifugation . The organic phase was dried and resuspended in ice-cold acetone . The individual lipid components of the acetone-soluble lipid fraction were separated by chromatography using the CycloGraph system ( Analtech ) . The separated fractions were analyzed by thin layer chromatography , and the fractions corresponding to mycolactone were pooled , dried down , weighed , resuspended in absolute ethanol , and stored at −80°C under nitrogen atmosphere in the dark [61] . Purified mycolactone was analyzed by mass spectrometry ( MS detector Thermo LxQ linear ion trap ) and the presence of mycolactone D confirmed . Under these conditions , mycolactone was stable for at least three years . Mouse fibroblasts L929 cell line was cultured in Dulbecco's Modified Eagle Medium ( DMEM ) ( Gibco ) supplemented with 10% fetal bovine serum ( Gibco ) , 2 mM L-glutamine ( Gibco ) , 10 mM HEPES ( Gibco ) , 1 mM sodium pyruvate ( Gibco ) and antibiotic-antimycotic ( Gibco ) . Cells were expanded in 175 cm2 flasks ( Nunc ) until 90% confluence . Then , cells were plated in 12-wells plates ( Nunc ) at a density of 2 . 5×105 cells/well , with increasing concentrations of mycolactone or with ethanol equivalent ( <0 . 002% ) , as a control . Rapamycin ( Calbiochen ) and Bafilomycin A1 ( Sigma ) was used to induce autophagy and to block autolysosome degradation , respectively . At each time-point , cells were collected and a pool of adherent and suspended cells was made . Cells were rinsed and resuspended in PBS . Absolute ethanol was gently added until 70% final concentration . Cells were stored in this fixing solution at 4°C . When all time-points had been collected , cells were rinsed in PBS and incubated with staining solution ( 0 . 1% triton-X-100 , 20 µg/mL of propidium iodide , 250 µg/mL of RNase in PBS ) for one hour in a bath at 50°C , in the dark . Samples were analyzed by flow cytometry ( LSRII , BD ) . The protocol was done according to the manufacturer's instructions ( BD Pharmingen ) . At each time-point cells were collected and a pool of adherent and suspended cells was made . Cells were rinsed , stained and analyzed by flow cytometry ( LSRII , BD ) . L929 cells were allowed to adhere to coverslips ( Nunc ) overnight following the incubation in different conditions . Cells were rinsed and fixed in paraformaldehyde , for 1 hour , at room temperature . Cells were rinsed and stored in PBS at 4°C . When all time-points had been collected , cells were blocked with blocking solution ( 5% BSA , 0 . 1% triton-X-100 , 0 . 1% tween-20 in PBS ) and incubated overnight at 4°C with the mouse anti-tubulin antibody ( AA4 . 3 , developed by C . Walsh and obtained from the Developmental Studies Hybridoma Bank , developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa , Department of Biology ) . Cells were rinsed and incubated with secondary AF488 goat anti-mouse antibody ( Invitrogen ) and rhodamine-phalloidin conjugate ( Invitrogen ) for 1 hour , at room temperature , in the dark . Cells were visualized using a confocal microscope ( FV1000 , Olympus ) with ×60 objective . 3D remodeling was performed using Fluoview software ( Olympus ) . At each time-point , cells were collected and a pool of adherent and suspended cells was made . Cells were rinsed with PBS , resuspended in Lysis Buffer ( 50 mM Tris-HCl pH7 . 2 , 250 mM NaCl , 2 mM EDTA , 1% NP-40 , 10% Glycerol , protease inhibitor ( Roche #11873580001 ) and phosphatase inhibitor ( Roche #04906837001 ) ) and stored at −80°C , until protein was extracted at the end of the experiment . When all time-points had been collected , samples were thawed , incubated for 30 minutes at 4°C with agitation , sonicated in a ultrasonic ice cold bath for 1 minute until no agglomerate was seen and centrifuged ( 30 minutes , 14000 rpm , 4°C ) . The supernatant was considered the total protein extract . For Western Blot analysis , protein concentration was determined ( Thermo Scientific #23227 ) and aliquots stored at −80°C . Protein was precipitated in 80% ( v/v ) acetone and the protein pellet resuspended in urea buffer ( 7 M urea , 2 M thiourea , 4% ( w/v ) CHAPS , 0 . 15% ( w/v ) DTT , 0 . 5% [v/v] carrier ampholytes and Complete Mini protease inhibitor cocktail ) . The protein separation was done as previously described [62] . Briefly , 100 µg protein extract was diluted with urea buffer to a final volume of 420 µL and in-gel rehydration was performed overnight . IEF was carried out in IPG strips ( pH 3–10 , non-linear , 18 cm; GE Healthcare , Uppsala , Sweden ) with the Multiphor II system ( GE Healthcare ) under paraffin oil for 55 kVh . SDS-PAGE was done overnight in polyacrylamide gels ( 12 . 5% T , 2 . 6% C ) with the Ettan DALT II system ( GE Healthcare ) at 1–2 W per gel and 12°C . The gels were silver stained and analyzed with the 2-DE image analysis software Melanie 3 . 0 ( Gene-Bio , Geneva , Switzerland ) . To verify the reproducibility three biological replicates for each time point and condition as well as three technical replicates were analyzed ( all analyzed gels are in Supplementary Information ) . An expression change was considered significant if the intensity of the corresponding single spot differed reproducibly more than twofold and was reproducible for all three experiments . The expected spot position in the 2D-gel according to the known protein sequence was calculated with the Compute pI/Mw tool ( http://ca . expasy . org/tools/pi_tool . html ) . For the detection of phosphorylated proteins 400 µg of protein were separated by 2-DE , stained with Pro-Q Diamond Phosphoprotein Gel Stain ( Molecular Probes ) , according to the manufacturer's instructions , scanned to detect the phosphorylation signals , silver stained and rescanned . Images of both scans were matched with the 2-DE image analysis software Melanie 3 . 0 ( Gene-Bio ) . For the protein identification , 400 µg of protein were separated by 2-DE . Selected spots were excised , digested with trypsin ( recombinant; Roche ) , and prepared as described earlier [62] . In brief , the extracted and dried peptides were dissolved in 5 µl alpha-Cyano-3-hydroxycinnamic acid ( 98% , recrystallized from ethanol-water , 5 mg/ml in 50% acetonitrile and 0 . 1% TFA ) and 0 . 5 µl applied onto the sample plate using the dried-droplet method . Peptide masses were measured with a UltrafleXtreme MALDI-TOF/TOF ( Bruker , Billerica , MA , USA ) . Proteins were identified according to their spot-specific peptide mass fingerprint and/or peptide sequence with the bioinformatic tool BioTools Version 3 . 2 ( Bruker ) with the following search parameters ( tolerance: MS = 10–50 ppm , MS/MS = 0 . 5–0 . 9 Da , enzyme: Trypsin , engine: Mascot , database: NCBInr , modifications: Oxidation ( M ) ) . A protein identification was accepted if at least three major peaks matched to the protein with the highest score ( full MS and MS/MS data in Supplementary Information ) . If the protein spot was detected at a lower molecular mass than expected , suggesting processing or fragmentation , the spot-specific peptides in the mass spectrum were also analyzed to confirm which parts of the corresponding protein sequence matched with these peptides . If the mass spectrum of the spot lacked peptides observed for the complete protein and had a different position in the 2D gel than expected it was indicated as a protein fragment . Therefore , both the spot position observed by 2-DE and the specific peptides in the corresponding mass spectrum were analyzed to indicate a putative protein fragment . 40 µg of protein were resolved in a 12% SDS-PAGE and transferred to the 0 . 2 µm Nitrocellulose membranes ( Bio-Rad #170-4159 ) with the semi-dry Trans-Blot Turbo system ( Bio-Rad ) . Membranes were blocked and subjected to immunoblotting with GAPDH antibody ( CellSignaling #2118 ) , LC3A/B antibody ( CellSignaling #4108 ) or mono- and polyubiquitinylated conjugates antibody ( Enzo Life Sciences #PW8810 ) , followed by incubation with horseradish peroxidase linked secondary antibodies ( Southern Biotech ) . Bands were detected with SuperSignal ( Thermo Scientific #34095 ) in a Universal Hood II ( Bio-Rad ) and quantified with QuantityOne ( Bio-Rad ) . GAPDH was used as loading control . At each time-point , cells were collected and a pool of adherent and suspended cells was made . Cytospins were made ( Cytospin III , Shandon ) and cells were fixed in paraformaldehyde for 20 minutes at room temperature and stored in ethanol 96% at 4°C . When all time-points had been collected , cells were blocked with blocking solution ( 5% BSA , 0 . 1% triton-X-100 , 0 . 1% tween-20 in PBS ) and incubated overnight at 4°C with the LC3A/B antibody ( CellSignaling #4108 ) . Normal Rabbit IgG Control ( R&D Systems AB-105-C ) was used as isotype control . Cells were rinsed and incubated with secondary AF568 goat anti-rabbit antibody ( Invitrogen ) for 1 hour , at room temperature , in the dark . Cells were visualized using a confocal microscope ( FV1000 , Olympus ) with ×60 objective . Eight-weeks-old female BALB/c mice were obtained from Charles River ( Barcelona , Spain ) and housed under specific-pathogen-free conditions with food and water ad libitum . Mice were infected in the left hind footpad with 30 µL of M . ulcerans suspensions with 4 . 8 log10 AFBs , or 30 µL PBS as control . Footpad thickness was evaluated every 2–3 days . Mice were sacrificed weekly and footpads were harvested for histological studies . The in vivo studies were approved by the Portuguese national authority for animal experimentation Direção Geral de Veterinária ( ID: DGV 594 from 1st June 2010 ) . Animals were kept and handled in accordance with the guidelines for the care and handling of laboratory animals in the Directive 2010/63/EU of the European Parliament and of the Council . Footpads were harvested , fixed in 10% phosphate-buffered formalin and embedded in paraffin . Tissue sections were stained with hematoxylin and eosin ( H&E ) , analyzed by light microscopy with polarized light and the amount of dermal collagen fibers was blindly scored from 0 ( lowest ) to 4 ( highest ) independently by two persons in two independent experiments . Additionally , tissue sections were stained with Masson's trichrome and pictures were taken in a light microscopy . At each time-point , cells were collected and a pool of adherent and suspended cells was made . Cells were rinsed , resuspended in TRIzol Reagent ( Ambion ) and stored at −80°C , until total RNA was extracted , at the end of the experiment , according to the manufacturer's protocol . Reverse transcription was done with whole RNA using RevertAid H Minus First Strand cDNA Synthesis Kit ( Fermentas ) according to the manufacturer's instructions . qPCR was perform on the C1000TM Thermo Cycler ( Bio-Rad ) using TaqMan Gene Expression Assay ( AB Applied Biosystems ) ( Plod1: Mm01255769_m1; Plod3: Mm00478798_m1; P4ha1: Mm00803137_m1; B2m: Mm00437762_m1; Gapdh: Mm99999915_g1; Hprt: Mm00446968_m1 ) . Relative quantification was determined with CFX Manager Software ( Bio-Rad ) using B2m , Gapdh and Hprt as reference genes . Differences between the means of experimental groups were analyzed using the Prism version 5 . 0 software ( GraphPad ) . Percentage and fraction values were transformed to and analyzed as arcsin values . Differences were considered significant only with a P value<0 . 001 , in the in vitro studies; or with a P value<0 . 01 , in the in vivo study . | Buruli Ulcer ( BU ) is a neglected tropical disease caused by Mycobacterium ulcerans infection . It has been recognized for many years that BU pathogenesis is mediated by the potent exotoxin mycolactone; however , the molecular action of this toxin on the host cell biology that drives its pathogenesis is not fully understood . Here we present a proteomic-based study that explores the molecular action of mycolactone on host cells biology . Our results provide further molecular evidence for the cytoskeleton-disarrangement induced by mycolactone , and unveil its impact on cytoskeleton-dependent cellular functions . Moreover , we extend the field of action of this toxin to the biosynthesis of collagen , implicating mycolactone on the decrease of dermal collagen found on BU lesions . Given the dependence of M . ulcerans virulence on its toxin , these findings on mycolactone's molecular action on host cells and tissues are of major importance for the understanding of BU pathogenesis . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"tropical",
"diseases",
"bacterial",
"diseases",
"molecular",
"motors",
"emerging",
"infectious",
"diseases",
"neglected",
"tropical",
"diseases",
"cellular",
"structures",
"and",
"or... | 2014 | Proteomic Analysis of the Action of the Mycobacterium ulcerans Toxin Mycolactone: Targeting Host Cells Cytoskeleton and Collagen |
The tubular networks of the Drosophila respiratory system and our vasculature show distinct branching patterns and tube shapes in different body regions . These local variations are crucial for organ function and organismal fitness . Organotypic patterns and tube geometries in branched networks are typically controlled by variations of extrinsic signaling but the impact of intrinsic factors on branch patterns and shapes is not well explored . Here , we show that the intersection of extrinsic hedgehog ( hh ) and WNT/wingless ( wg ) signaling with the tube-intrinsic Hox code of distinct segments specifies the tube pattern and shape of the Drosophila airways . In the cephalic part of the airways , hh signaling induces expression of the transcription factor ( TF ) knirps ( kni ) in the anterior dorsal trunk ( DTa1 ) . kni represses the expression of another TF spalt major ( salm ) , making DTa1 a narrow and long tube . In DTa branches of more posterior metameres , Bithorax Complex ( BX-C ) Hox genes autonomously divert hh signaling from inducing kni , thereby allowing DTa branches to develop as salm-dependent thick and short tubes . Moreover , the differential expression of BX-C genes is partly responsible for the anterior-to-posterior gradual increase of the DT tube diameter through regulating the expression level of Salm , a transcriptional target of WNT/wg signaling . Thus , our results highlight how tube intrinsic differential competence can diversify tube morphology without changing availabilities of extrinsic factors .
Branched tubular networks , like our vasculature transport and exchange vital gases and nutrients along entire organisms . The branching patterns , tube structures and dimensions in these networks show considerable regional variations to meet the different needs of target organs and ensure optimal organ function and animal fitness [1–4] . Adaptations of branch morphologies to the tissue environments can be achieved by changing the local availability of extrinsic factors like guidance molecules and/or by intrinsic regional differences in tube cell competence to respond and modify signaling outcomes . Although the prominent roles of variations in extrinsic signals in organotypic branching become widely established [3 , 5 , 6] , the tube intrinsic mechanisms determining the differential responses of tube cells to signaling remain to be explored [7–10] . Despite the huge evolutionary distance of insects and mammals , the formation and maturation of the respiratory tube network in Drosophila melanogaster has served as a fruitful model system of branching morphogenesis [11 , 12] . Here , we use this system to evaluate the contribution of tube intrinsic , regionally differential competence in diversification of tube morphology . The fly respiratory network , also called the tracheal system ramifies extensively to deliver oxygen to each cell in the body ( Fig . 1A ) [13 , 14] . It derives from 10 primordial cell clusters specified in ectodermal para-segments ( PS ) 4–13 on each side of the body [15] . At stage 11 , the metameric cell clusters invaginate and begin to extend 6 primary branches . The tracheal metameres ( Tr1-10 ) interconnect into a network through the activities of 5 to 6 specialized fusion cells in each branching unit [16–18] . These cells find and adhere to their ipsilateral or contralateral counterparts in neighboring metameres to form continuous tubes . Tracheal branching morphogenesis is controlled by two conceptually different groups of extrinsic signals [6 , 19] . One is required in all primary branches while the other class specifies unique sets of primary branches . FGF/Branchless ( Bnl ) [20] is dynamically expressed in the surrounding tissues and serves as a general extrinsic branching signal by activating FGFR/Breathless ( Btl ) on the airway cells [21–23] . Btl activation initiates oriented cell migration and enhances branch elongation while it also organizes cell fate specification along each primary branch . The gradual specification of distinct fusion and terminal cell fates at the branch tips is essential for both branch fusion and further ramification branching [14 , 20] . The second class of signals includes BMP , Wnts and Hh proteins . BMP/Decapentaplegic ( Dpp ) is expressed in dorsal and lateral stripes along the length of the embryo [24] . Dpp signaling specifies and licenses dorsally ( dorsal branch , DB ) and ventrally migrating primary branches ( lateral trunk , LT and ganglionic branch , GB ) by inducing the expression of the TFs kni and knirps-related ( knrl ) in airway cells [24–26] . Wg is expressed in a repetitive pattern of transverse ectodermal stripes and together with other WNT signaling molecules specifies dorsal trunk ( DT ) identity [27–29] . It upregulates the expression of the TF salm [30] in the major airways of the network . salm promotes short and thick tubes by suppressing intercalation of the tube constituent cells [31 , 32] . On the other hand , kni/knrl can repress salm expression in DB [25] and promotes cell intercalation in the long and narrow tubes of dorsal ( DB ) and ventral branches [31] . Hh is also segmentally expressed in ectodermal stripes and modulates the airway branching , both indirectly through positive regulation of bnl expression [33] and directly by acting on terminal cell specification or extension [34 , 35] . The functions of hh in primary branching remain to be thoroughly studied [34 , 36 , 37] . Collectively , the differential primary branch identities established by the second class of extrinsic signals are intimately linked with the distinctive branching patterns and dimensions of individual primary branches [38] . The region-specific modification of serially homologous organs and appendices is a general theme in animal development [39–42] . The evolutionary conserved Hox gene complexes are key selector genes of tissue identities along the anterior posterior ( A-P ) axis of animals [39 , 41 , 43–45] [46–50] . In Drosophila , 2 groups of Hox genes , the Antennapedia complex ( ANTP-C ) and Bithorax complex ( BX-C ) confer regional differences to the body plan by graded expression of their products in register with para-segmental units [39 , 51] . The 3 protein-coding genes of the BX-C [52] are expressed in distinct and partially overlapping domains along the A-P embryonic axis . Ultrabithorax ( Ubx ) expression initiates in the cells of PS5 , abdominal-A ( abdA ) expression starts from PS7 and Abdominal-B ( AbdB ) from PS10 [44 , 53] . A connection between regional modification of airway morphology and Hox genes has been established already in the early studies of Hox-gene mutants . Upon loss of all BX-C genes the tracheal metameres Tr2-Tr10 become transformed to Tr1 [39 , 54] . However , the genetic and molecular mechanisms establishing the different branch morphologies along the airway tubes have been largely unexplored . Here , we focus on the regulation of 2 distinct morphological modifications along the DT major airways . We first analyze how the most anterior part of the DT diverges its branching pattern and tube size to generate long and narrow tubes targeting the head . We show that these regional modifications in cell behaviors are controlled by a combination of hh signaling and the airway intrinsic Hox code . We further investigate the mechanism of tube tapering in the central domain of the DT airways . We find that BX-C genes modulate the anterior-to-posterior gradation of DT tube diameter partly through regulating the expression of level of salm , a target of WNT/wg signaling . Our work highlights that the intrinsic Hox code locally modifies the outcomes of extrinsic signals to establish regionally different branching patterns and to coordinate tube shapes in register with the embryo axis .
The DT is a continuous tube running along the A-P axis of the embryo . It connects with the exterior through the narrow tube of the spiracular chamber in the posterior spiracle ( PSP ) [55] . The DT is constructed by the fusion of an anterior ( DTa ) and a posterior ( DTp ) branch from each tracheal metamere ( Tr1-Tr10 ) [13 , 17] . The DT airway encompasses several regional variations that provide a suitable system for the study of the interplay of external signaling with intrinsic factors during tube morphogenesis [13] . First , the most anterior end of the DT extends several specialized branches ( see below ) . Second , it shows a pronounced posterior to anterior diameter tapering contrasting the largely cylindrical shape of other primary branches in the network . Third , its most anterior metameric unit ( Tr1 ) lacks a DTa fusion cell , whereas the most posterior one ( Tr10 ) does not generate a fusion cell in its posterior branch ( DTp ) . More generally , Tr1 is distinct from the rest of the tracheal metameres because it encompasses more cells and branches to oxygenate the specialized organs of the head and thorax . The specialized branches of Tr1 include the cerebral branch ( CB ) targeting the brain , the pharyngeal branch ( PB ) to the anterior intestine , the ventral cephalic branch ( VC ) extending to epidermis and muscles and the ganglionic branch GB0 , which penetrates the ventral nerve cord . Among these , the CB and VC are directly linked to the anterior end of the DT airways . Despite the pronounced differences in the final branching patterns , branching in Tr1 is comparable to the common stereotypic primary branching of Tr2-Tr9 during stage 11 ( Fig . 1A ) . At stage 12 however , DTa1 , elongates further than other DTa branches and shifts dorsally . By stage 13 , visceral branch 1 ( VB1 ) and DTa1 co-segregate from the transverse connectives ( TC ) . Later , DTa1 extends dorsally and posteriorly and turns towards the brain , forming CB [13] . DTa1/CB develops very narrow and long tubes compared to the thick and short DT branches in posterior metameres . VB1 extends more anteriorly , forming the pharyngeal branch ( PB ) ( Fig . 1A ) . The gradual morphological diversification of CB/DTa1 compared to DTa braches in other metameres prompted us to examine the expression of branch identity TFs , salm and kni . In a “typical” , central metamere , salm expression is upregulated by wg/WNT signaling [27–29] and is detected in DB and DT at stage 11 [30] . Later , kni is induced in DBs by dpp , where it represses salm expression [24–26] . kni and salm are co-expressed in DB1/10 ( see below ) . In contrast to the DTa of other metameres , we found that kni expression is strongly upregulated in DTa1 from late stage 11 ( Fig . 1B ) . Concomitantly , salm is not detectable in DTa1 ( Fig . 1C ) although it becomes upregulated in DTp1 as in other metameres ( Fig . 1D ) [30] . To test the significance of the differential kni expression in DTa1 we analyzed mutants lacking both kni and its paralog knrl . In these chromosomal deficiency mutants , the abdominal segments are missing due to the early gap gene function of kni , while trunk development is rather normal [56] . In the trunk region of Df ( kni/knrl ) mutants , the formation of kni positive primary branches [24 , 25] is variably affected , ranging from complete absence to branch stalling [25] , while the salm-positive DT branches can form and fuse ( see below ) [57] . We noticed that in kni/knrl mutants , airway cells initiate branch outgrowth in the dorsal and ventral directions and respond to dpp by inducing the dpp-responsive kni reporter , kni- ( dpp ) -lacZ ( Fig . 1E ) . In these embryos , salm is ectopically detected in kni- ( dpp ) -lacZ positive cells in either dorsal or ventral cells near the wg stripe ( Fig . 1E ) . This suggests that dpp mediated kni/knrl induction suppresses salm induction by wg in both the dorsal [25 , 26] and ventral branches . However , in the first and the last metameres of Df ( kni/knrl ) mutants , salm expression additionally expands to nearly cover the entire metamere , including the putative CB/DTa1 ( Fig . 1E ) . This suggests that kni functions in DTa1 to repress salm . Additionally , the competence of tracheal cells to induce salm is differentially modulated in the terminal Tr1 and Tr10 metameres compared to the central metameres . Consistent with the notion that generalized reduction of wg/WNT signaling can bypass the requirement of dpp/BMP signaling during DB extension in Tr2-Tr10 [31] , btl-Gal4 [58] mediated overexpression of GFP fused to Axin ( Axn ) [59–61] , a negative regulator of wg signaling moderately rescues dorsal extension of the residual DBs ( DB1 and DB2 ) of Df ( kni/knrl ) mutants but does not appreciably rescue the extension of DTa1/CB ( Fig . 1J ) . In sharp contrast , btl-gal4 driven UAS-kni or UAS-knrl in Df ( kni/knrl ) mutants restores CB formation ( Fig . 1F-I ) . Thus , we conclude that kni induction and the resultant salm repression in DTa1/CB are essential for its formation and extension . In support for this , btl-gal4 driven UAS-salm in wild-type background significantly suppresses DTa1/CB formation but has little effect on the extension of DTa in Tr2-Tr10 [28 , 62] , which endogenously expresses salm ( Fig . 1K ) . Collectively , the results suggest that kni induction and the resultant salm repression in DTa1/CB are essential for its formation and extension . The diversified expression of kni in DTa1 compared to the remaining DTa branches could be regulated by differential expression of exogenous guidance factors around DTa1 and/or by intrinsic differences of competence among the DT1 cells . Firstly , to investigate which extrinsic factors are upstream of kni induction in DTa1 , we analyzed the expression or function of known , secreted airway branching regulators . At stage 11 , Tr1 like the rest of the tracheal metameres is surrounded by 6 patches of bnl expressing cells , prefiguring the stereotypic directions of the common primary branches ( S1A Fig . ) . This general pattern diversifies in the cephalic region of stage 12 embryos , where the DTa1 migrates toward a more dorsal bnl expression spot ( S1B Fig . ) . Despite this difference , kni induction in DTa1 is still detected in btl mutants ( S1C , D Fig . ) , excluding a major role of bnl in kni induction in DTa1 . dEGFR/faint little ball/torpedo ( top ) [63] encodes an RTK that is suggested to positively act on salm expression [64] upon binding the Spitz/EGF ligand [65] . In embryos mutant for rhomboid ( rho ) , encoding a protease required for generating the active Spitz [66 , 67] or for dEGFR , the expression of kni in the DTa1/CB still occurs ( S1E , F Fig . ) . This argues against a role of localized dEGFR activation in controling kni expression in DTa1 . dpp and wg are known inducers of kni/knrl in DB [24] and salm in DT [27 , 28] , respectively , in a “typical” metamere . Thus , variations of their expression in the Tr1 proximity might influence the specialized expression patterns of kni or salm in DTa1 . However , the expression of both dpp and wg is comparable around Tr1-3 ( S1G , H Fig . ) , arguing against an instructive role of these two factors in kni induction in DTa1 . Indeed , neither mutants of tkv , encoding one of the two dpp receptor subunits [68 , 69] nor arm mutants lacking an essential component of wg signaling [70–72] , showed major defects in kni induction and outgrowth of DTa1/CB ( S1I , J Fig . ) . hh is a signaling molecule that binds its receptor patched ( ptc ) , thereby relieving ptc-mediated inhibition of the 7 transmembrane domain protein smoothened ( smo ) [73 , 74] . hh is expressed in stripes in the ectoderm , abutting the anterior edge of the airway primordia at stage 10 and overlying the anterior part of the invaginated airway cells of each metamere at stage 11 ( S1K , L Fig . ) [34] . Glazer and Shilo showed that hh induces marker gene expression in the anteriorly migrating branches of central metameres [34] , arguing that hh signaling patterns the anterior primary branch fates of the “typical” , central metameres . We found that just after invagination of the airway primordial cells , expression of ptc , a transcriptional target of hh [75 , 76] is upregulated in the DTa1 precursors ( Fig . 2A ) , suggesting that hh signaling is active there . In hh mutants the dorsalward CB extension is hardly detectable and salm expression is expanded in the entire DT1 ( Fig . 2B-D ) . This suggests that hh signaling in DTa1 induces kni and thereby represses salm expression . Among the ectopically salm expressing cells in DTa1 , some cells also express kni while others do not . We suggest that in the absence of hh signaling , hh responsive kni induction is lost while dpp signaling may take over to induce kni expression in some salm positive cells . Such an ectopic activation of kni in the absence of hh could induce ectopic kni/salm-double positive cells resembling DB1 cells . This interpretation is consistent with the de-repression of kni- ( dpp ) -lacZ in Df ( kni/knrl ) mutants ( Fig . 1E ) . Thus , we conclude that hh signaling is required to induce kni and to repress salm in DTa1 . The earlier function of hh is also required for the maintenance of the striped expression of wg [75 , 77] , which induces salm expression in DT . Thus , the variability of salm expression either in DTa1 or in DTa/DTp of any metamere in hh mutants may partly reflect a reduction or loss of epidermal wg expression . In hh mutants , bnl expression guiding DB migration in central metameres is lost but bnl expression in surrounding cells guiding CB is still detectable at stage 12 ( S1M Fig . ) . Nevertheless , the dorsal extension of a CB-like branch is not detected in hh mutants at later stages ( S1N Fig . ) . To more directly address the effect of hh signaling in DTa1 we attempted to inactivate its components specifically in the airways . hh signaling modifies the transcriptional activity of cubitus interruptus ( ci ) [74] . In the absence of hh , Ci is proteolytically processed and acts as a repressor [78] , while upon hh pathway activation , smo mediated signaling suppresses this proteolysis and turns Ci into an activator [78–80] . The balance of loss of the repressor form and generation of the activator form of Ci determines the hh signaling outputs [79–81] . We generated embryos expressing dominant negative forms of ci , ciDN ( cirep [82 , 83] and/or ci75 [78 , 84] ) exclusively in the airways and assessed the expression levels of kni- ( dpp ) -lacZ ( a DB and LT/GB marker ) [25] and of salm-TSE-lacZ ( a reporter of salm expression ) [30] in metamere 1 . Both markers are ectopically induced in the DTa1 of these embryos at stage 13 ( Fig . 2E-H ) although the cell number in this branch did not significantly change ( 20 cells , standard deviation SD = 0 . 707 for 5 wild type embryos at stage 13 and 19 . 2 cells , SD = 0 . 447 for 5 btlX2> cirep embryos ) . We interpret that the incomplete inactivation of hh signaling in the airways by ciDN , partially transformed DTa1 cells to DTp1 . These cells are still receiving enough Dpp to express kni- ( dpp ) -lacZ . The weaker effects of ciDN expressing embryos compared to hh mutants may reflect ineffectiveness of CiDN or the delayed btl-Gal4 mediated expression [58] of CiDN , which starts slightly later than the initiation of hh action on DTa1 . Additionally , the airway-specific overexpression of CiDN or general hh inactivation in smo mutants frequently resulted in DTa1/VB1 co-segregation defects and CB misrouting at stage 16 ( Fig . 2I , J and S1O Fig . ) . A similar CB misrouting phenotype has been described in mutants of unplugged ( unpg ) encoding a TF expressed in CB [54] . Indeed , the expression of an unpg enhancer trap in the CB of wild type embryos is lost upon CiDN overexpression ( Fig . 2I , J ) . Collectively , these results identify a selective , direct role of hh signaling in inducing the distinct cell identities of DTa1 compared to the cells of the remaining DT branches . The transformation of DTa1 to DTp1/DB1 upon inhibition of hh signaling suggests that its overactivation may be sufficient to transform DTp1/DB1 to DTa1 . To examine this , we analyzed ptc mutants , where hh signaling is inappropriately activated due to the loss of ptc-mediated inhibition of smo [73 , 74] . In ptc mutants , the dorsal part of metamere 1 expresses Kni but not Salm already at late stage 11 ( Fig . 3A , B ) . Correspondingly at stage 13 , expression of salm-TSE-lacZ is specifically lost from metamere 1 , suggesting a defect in both DTp1 and DB1 specification ( Fig . 3C and S3A Fig . , note that in wild type , salm is expressed in DB1 and DTp1 ) . Consistent with a loss of the DB1 fate in ptc mutants , kni- ( dpp ) -lacZ expression in the dorsal part of metamere 1 is completely lost ( Fig . 3E ) while Kni protein is expressed in the whole distal part of Tr1 ( S2B Fig . ) . Although ptc mutants contain fewer airway cells [33] , presumably due to an early upregulation of wg [85] , a negative regulator of the airway primodia size [86] , reduction of cell number in CycA mutants [38] does not significantly abrogate DB1/DT1 fates ( S2D , H Fig . ) . This suggests that the effect of ptc on DB1/DT1 specification is more direct and not due to a general reduction in the number of airway cells . Because btl-Gal4 driven Cirep ( Fig . 3D , F ) or Ci75 can restore the expression of both kni- ( dpp ) -lacZ and sal-TSE-lacZ in the Tr1 cells of ptc mutants and because Ptc is expressed in all airway primordia including the entire Tr1 primordium ( Fig . 2A ) , these results suggest that overactivation of hh signaling in Tr1 abolishes the DTp1/DB1 fates . To further test the role of hh signaling in determining branch identities in Tr1 we analyzed the RNA expression of unpg . In control embryos at early stage 12 , unpg expression is strongly detected in DTa1 [54] and weakly in the anterior part of TC1 . Both of these regions correspond to hh signaling activation ( Fig . 3G ) . At stage 13 , unpg RNA is detected in CB and GB0/GB1 in Tr1 and also in the GBs of the more posterior metameres Tr2-Tr9 ( Fig . 3H ) [54] . Consistent with the loss of unpg-lacZ expression in CB upon CiDN overexpression , unpg RNA expression is lost in DTa1/CB of hh mutants ( Fig . 3K , L ) . Conversely in ptc mutants , it is expanded posteriorly to cover the positions of DB1/DTp1/TC1 ( Fig . 3I , J ) , indicating their transformation to CB-like fates . Consistent with the expanded unpg expression , we often detected a duplication of CB-like branches in ptc mutants ( Fig . 3P ) . We additionally noted that unpg expression in GB is lost in ptc mutants ( Fig . 3J ) while unpg is derepressed in LTa in hh mutants ( Fig . 3L ) , in accord with the notion that hh confers the anterior branch identity in the central metameres [34] . At stage 16 , ptc mutants show variable branching defects including stalled GBs , DBs and DT breaks [33] . Concomitantly with the loss of DTp1 fate marker ( salm-TSE-lacZ ) , DT1 and DT2 never fuse in ptc mutants ( Fig . 3N-P ) . In wild type , one of DTp1 cells takes the fusion cell fate , activates dys expression and attaches to a fusion cell in DTa2 ( S2E Fig . ) , [18] . In ptc mutants , dys is not activated in DTp1 while dys expression is variably expanded in more cells of the DT branches in posterior metameres ( S2E-G Fig . ) . This may reflect an increase of epidermal expression of wg [85] , an inducer of the fusion cell fates [27 , 28] . The failure of dys activation and DT1 fusion in ptc mutants is significantly rescued by btl-Gal4 mediated overexpression of Cirep ( S2I Fig . ) . Moreover , both loss of dys expression in DT1 as well as DT1/2 fusion defects are variably observed when dominant active Ci , Ciact [84 , 87] is overexpressed in the airway cells ( S2J Fig . ) . Notably however , the Ciact overexpression by btl-Gal4 did not diminish salm-TSE-lacZ or kni- ( dpp ) -lacZ expression . In summary , we suggest that hh signaling instructs the DTa1 fate at the expense of DB1/DTp1 fates . In DTp1 , hh signaling must be kept low to allow the proper selection of the fusion cell fate and subsequent DT1/DT2 branch fusion . The hh induction of kni expression in the DTa of wild type embryos as well as the loss of DT/DB fates in ptc mutants are confined to Tr1 . However , hh signaling outcomes are expected to be equally profound in the more posterior metameres of both wild type [34] and ptc mutant embryos [33] . The exclusive restriction of Hh responses within Tr1 implies the presence of an inhibitory mechanism preventing kni activation in the DTa branches of posterior metameres . The BX-C genes represent obvious candidate regulators of posterior metamere identity and modulators of hh signaling outcomes along the A-P axis of the airways . Indeed , unpg expression is de-repressed in progressively more posterior metameres in Ubx mutants and Ubx abdA AbdB triple mutants [54] . BX-C gene expression is graded along the A-P axis of the airway metameres ( S3A-D Fig . ) . Ubx expression starts in Tr2 ( PS5 ) and peaks at Tr3 ( PS6 ) ( S3A Fig . ) . abdA expression starts in Tr4 ( PS7 ) and peaks in Tr6 ( PS9 ) ( S3B Fig . ) while AbdB expression starts in Tr7 ( PS10 ) and peaks in Tr10 ( PS13 ) ( S3C Fig . ) . These expression patterns are in register with the expression of BX-C genes in the ectoderm [53] , which is the origin of the airway primordia . To explore the function of BX-C genes in DTa fates , we first monitored dys expression in various BX-C mutants . In Ubx mutants , single fusion cells are detected in Tr1 , Tr2 and Tr3 ( S3E , F Fig . ) suggesting that DTa2 and 3 are transformed to become DTa1/CB . abdA single mutants do not show dys expression defects in the DT ( S3G Fig . ) while a superfluous fusion cell in DTp10 is detected in both AbdB single and abdA AbdB double mutants ( S3H , I Fig . ) . This suggests that DT10 , which normally contains only a single fusion cell in its DTa branch , is transformed into a more anterior identity . Compared to Ubx single mutants , Ubx abdA double mutants have single fusion cells in Tr1–8 and often in Tr9 ( S3J Fig . ) . In Ubx abdA AbdB triple mutants , the DT stumps of all metameres contain single fusion cells ( S3K Fig . ) . This implies that DTa branches in progressively more posterior metameres are transformed to become DTa1 upon progressive loss of BX-C genes [39] . Any single gene of the BX-C is sufficient to suppress the DTa1 fate . Consistently , we detected expansion of kni expression and a loss of salm in the transformed DTa in Ubx single , Ubx abdA double and Ubx abdA AbdB triple mutants ( Fig . 4A-C ) . These phenotypes are often accompanied with the appearance of dorsally extending branches that are positive for Kni but negative for kni- ( dpp ) -LacZ , Salm and salm-TSE-LacZ , resembling the CBs of wild type embryos ( S3K Fig . ) [54] . The marker expression analysis in BX-C mutants suggests that in the posterior metameres , Ubx , abdA and AbdB interfere with the outcomes of hh signaling . If the antagonistic role of BX-C on hh-mediated kni induction reflects an essential function of the BX-C in posterior metameres , one might expect some rescue of the branching defects of BX-C mutants upon simultaneous loss of hh or kni/knrl . Indeed , salm expression is de-repressed in DTa1-9 of Ubx abdA hh and of Ubx abdA kni/knrl mutants ( Fig . 4D , E ) . Additionally , DT fusion is weakly restored in both the triple and quadruple mutants ( S3L , M Fig . ) . Taken together , we suggest that BX-C genes antagonize hh-mediated induction of kni in DTa branches . In addition to the airways , BX-C genes are expressed in many embryonic tissues . Where do they act to divert hh signaling from kni induction in the DT branches ? In lack of reagents for the reliable conditional inactivation of the BX-C genes in the airways , we monitored the effects of airway-specific ectopic expression of BX-C genes on DT1 cell specification . The first metamere does not express the BX-C genes ( S3A-D Fig . ) and thereby may provide a naïve environment for assessing the effects of their overexpression on marker gene activation [41] . btl-Gal4 mediated overexpression of any of the BX-C genes in wild type background , variably decreases kni expression in the DTa1 and concomitantly leads to increased salm levels at stage 13/14 ( Fig . 4F-J ) . At later stages , DTa1 branches are thick , resembling typical DTa branches of posterior metameres ( see below ) in agreement with the previously reported loss of CBs upon abdA overexpression [88] . Similarly , btl-Gal4 mediated overexpression of either Ubx or abdA restores the fusion defects of DT1 and DT2 in ptc mutants ( S3N , O Fig . ) . We detected that expression of both salm-TSE-lacZ and kni- ( dpp ) -lacZ is restored in the dorsal part of Tr1 of ptc mutants upon abdA overexpression ( S3P , Q Fig . ) . These results argue that the Hox code in the airway cells autonomously changes hh-signaling outputs in Tr1 both in wild type and in ptc mutants , where the hh pathway is hyper activated . Finally , we asked if transgenic expression of abdA or Ubx in the airways could restore the branch fusion defects along the entire DT of Ubx abdA double mutants . Again , this manipulation rescues the branch fusion phenotypes ( S3R-T Fig . ) arguing that BX-C genes autonomously shunt hh signaling from inducing kni in the DTa branches of all metameres to promote continuous DT formation . A common characteristic of biological tubes is the tapering of tube diameter along their length . The Drosophila larval airways receive air only from the PSP and distribute it anteriorly . Correspondingly , the tubes show a posterior to anterior tapering [38] , which presumably gradually increases the flow rates to the anterior and facilitates air diffusion from the PSP to the most distant anterior organs ( http://hyperphysics . phy-astr . gsu . edu/hbase/pfric . html ) . salm is a master selector gene for DT identity . Intriguingly , its expression levels in the DT tubes at stage 13/14 show a largely proportional decrease from posterior to anterior metameres matching the tapering of the airways ( Fig . 4F ) [30] . The graded diameter along the airway length also coincides with the graded expression of BX-C proteins along the A-P axis . To explore the potential regulatory roles of BX-C factors and salm in tube shaping , we first analyzed airway shapes in BX-C mutants and detected 2 kinds of effects of BX-C genes on tube diameter , metamere-autonomous and systemic ( see below ) . Consistent with graded AbdB expression in PS 10–13 , in AbdB mutants , tube diameter in DT7-10 lost its tapering and became narrower suggesting that the amount of AbdB proportionally controls the tube diameter . ( Fig . 5A , C , I , J and S1 Table ) . Similarly , in abdA mutants , the gradient of tube diameter in DT4-6 was lost and DT4-9 became narrower ( Fig . 5B , I , J and S1 Table ) suggesting again that AbdA levels proportionally control DT tube diameter . In abdA AbdB double mutants the shape of the airways is changed further ( Fig . 5D , I , J and S1 Table ) . The airways of Tr4-10 acquire a more cylindrical shape compared to the conically shaped tubes of wild type embryos . We suggest that in wild type embryos the gradient of abdA and AbdB activities would superimpose on a weak but clear , abdA and AbdB-independent gradient of DT tube thickness ( Fig . 5I and S1 Table ) . This may explain why abdA mutants , where PS7-9 ( DT4-6 ) are expected to transform to PS6 ( DT3 ) still show a clear difference in tube diameter between DT3 and DT4 and why AbdB mutants , where PS10-13 ( DT7-10 ) are expected to transform to PS9 ( DT6 ) show a distinct tube caliber in DT6 and DT7 ( Fig . 5I and S1 Table ) . Thus , in accord with their known functions in determining cell fates and morphogenesis in the embryonic ectoderm [39 , 44 , 53] , the BX-C genes control the tapering of airways along the A-P axis autonomously . We note however that there is also a systemic effect of BX-C mutations along the entire DT . In either abdA , AbdB single or in abdA AbdB double mutants , the diameter of the more anterior metameres , where corresponding BX-C genes are not expressed also show a slight reduction of tube diameter ( Fig . 5I and S1 Table ) . Among different possibilities , these results may suggest that the activities of abdA or AbdB control the hydrostatic pressure in the lumen to non-autonomously assure proportional growth of all the DT tubes [89] ( see below ) . The residual tapering of abdA AbdB double mutants implies a mechanism of A-to-P gradient formation independent of abdA and AbdB . Ubx could exert such a function in the absence of abdA and AbdB . We analyzed bxd113 or bxd100 mutants , where Ubx expression levels in PS5 become similar to that of PS4 but its ectodermal expression is lost from PS7 onwards [90] ( S4A Fig . ) . In these embryos , DT3 tube diameter approaches that of DT2 and there is also an overall reduction of tube diameter in more posterior metameres ( Fig . 5I , S4C-E Fig . and S1 Table ) , suggesting that the endogenous Ubx level controls DT diametric tube expansion . However , in abdA AbdB double mutants , Ubx levels are largely uniform in DT4-10 , which correspond to PS7-13 [91] ( S5B Fig . ) . This suggests the presence of an additional , BX-C-independent cue in DT tube shaping . In Ubx abdA AbdB triple mutants , the transformed Salm positive residual DT/TC branches in posterior metameres are slightly thicker than those in anterior metameres , further arguing for the existence of the postulated BX-C-independent mechanism in tube shaping ( S3K , S4F Figs . ) . Is there any causative link between the BX-C mediated DT expansion control and the A-to-P gradual increase of salm expression levels in the DT ( Fig . 4F ) [30] ? We noticed that Salm levels are reduced in central and posterior metameres of abdA , AbdB or abdA AbdB double mutants ( S4G-J Fig . ) . This reduction is largely consistent with the changes in shape and DT tube diameters in the corresponding mutants . Conversely , upon overexpression of AbdB , higher Salm amounts are detected in the DT of all metameres at stage 13 ( Fig . 4I ) . The DT branches of these embryos often stall and fail to fuse making evaluation of tube diameter difficult . Nevertheless , the DT diameter in all metameres appears comparable to the diameter of the most posterior DT branches ( Fig . 5G ) . Overexpression of Ubx or abdA renders Salm expression levels uniform in the anterior metameres ( Fig . 4G , H ) . Correspondingly , the DT diameter in the anterior metameres becomes thicker ( Fig . 5E , F , I , J and S1 Table ) . How does the Salm gradient along the DT A-P axis correlate to the graded DT tube expansion ? salm overexpression confers DT identity to other primary branches [28 , 62] . We noted that salm overexpression also generally expands tube diameter not only in the transformed branches [28] but also in the DT , which endogenously expresses salm ( Fig . 5H , I , J and S1 Table ) . The programmed secretion of luminal and apical proteins has been proposed to drive tube dilation of the Drosophila airways [92–95] . Tenectin ( Tnc ) is a luminal glycoprotein accumulating in the DT and hindgut tubes during diametric expansion [96] . Tnc overexpression in the airways drives DT tube dilation in a dose-dependent manner potentially through increasing hydrostatic pressure [89] and the tnc mRNA levels increase in a characteristic graded fashion along the A-P axis of the DT tubes in wild type embryos [89] . We found that the diametric increase caused by salm overexpression in the airways is accompanied by an increase in the luminal levels of Tnc and conversely , Tnc becomes undetectable in the tracheal tubes of salm mutants ( S4K-M Fig . ) . This suggests that Salm adjusts the graded expression levels of Tnc and presumably other proteins during tube dilation . Additionally , the diameter increase in the branches of salm over expressing embryos is still most pronounced in the posterior metameres . ( Fig . 5H , I , J and S1 Table ) . The accentuated tube enlargement in posterior metameres upon salm overexpression suggests a salm-independent control mode of A-to-P gradient of DT tube diameter . This is consistent with the observation that tube diameter of the remaining branches in salm mutants are still thickest in Tr10 ( S4N Fig . ) . In conclusion , our work suggests two interdependent mechanisms by which the tube intrinsic Hox code controls branch identity and tube shape in the Drosophila respiratory network ( Fig . 6A , B ) . Firstly , Hox genes autonomously divert extrinsic hh signaling from kni induction in the DT thereby generating continuous , salm positive DT airways . We suggest that this allows a second tier of DT tube shape regulation , where the Hox activity gradient both locally and systemically organizes graded tube dilation via salm dependent and independent modes . The Ci/BX-C circuit may control tube morphology directly by binding to the regulatory regions of kni and salm , as recent genome-wide TF binding studies suggest that kni is a direct target of ci [37] and that salm is a direct target of Ubx [97] . In our model , tube tapering and thereby luminal fluid flow are calibrated by the balance between extrinsic signals and the intrinsic Hox code . Since Hox selector genes are regionally expressed in other developing tubular organs like the mammalian lung [98] and vasculature [9 , 10] , a similar regulatory logic of tube branching and shaping may apply to other systems .
Flies kept over balancer chromosomes [99] were grown in standard medium . We obtained the appropriate genotypes by standard genetic crosses . For overexpression of genes , we used the Gal4/UAS system [100] . Mutant embryos were identified by the expression of twi-lacZ , ftz-lacZ , Ubx-lacZ or dfd-GFP [101] constructs inserted on balancer chromosomes . We identified mutants harboring Ubx or AbdB mutations by selecting embryos with previously reported phenotypes in the anterior spiracle or PSP . For collection of large amount of virgins , we used a Y chromosome harboring hs-hid construct developed by R . Lehmann and M . Van Doren [102] . See Flybase [103] for details of strains described below . Mutant strains . abdAM1 ( a gift from B . Gebelein ) [104] , AbdBM1 , AbdBM5 and Df ( 3R ) P115 as Df ( Ubx abdA AbdB ) ( gifts from I . Lohmann ) [105] , btl∆oh10 and btl∆oh24-1[106] , Df ( 5 ) as Df ( salm/salr ) ( a gift from M . Llimargas ) [28 , 107] , hh13C [108] , kni early rescue fragment; Df ( 3L ) riXT ( a gift from R . Schuh ) [25] , rhod38 ( a gift from D . Andrew and A . Salzberg ) [109 , 110] , rho7M ( a gift from J . Skeath ) [111] , topf24 ( a gift from K . Moses ) [112] . arm4 and CycA3 were obtained from Natinal Institute of Genetics ( NIG ) , Mishima , Japan . Ubx1 , abdAMX1 , abdAD24 AbdBD18 , Ubx1 abdAD24 AbdBD18[113] , Df ( 109 ) as Df ( Ubx , abdA ) , CycAC8LR1 , Df ( 3L ) Exel6115 as Df ( CycA ) , hhAC , Df ( 3L ) BSC448 homozygous or its transheterozygote over Df ( 3L ) riXT as Df ( kni/knrl ) , ptc9 , Df ( 2R ) Exel7098 as Df ( ptc ) , smo3 , tkv7 and Df ( 2R ) Exel6076 as Df ( top ) were obtained from Bloomington stock center ( BDSC ) , Indiana , USA . Enhancer trap strains . 1-eve-1 as trh-lacZ ( a gift from N . Perrimon ) [114] and unpgr37 ( a gift from F . J . Diaz-Benjumea , R . Urbach and G . M . Technau ) [115 , 116] . hhP30 [117] was obtained from BDSC . Enhancer reporter strains . kni- ( dpp ) -lacZ [25] and salm-TSE-lacZ [30] ( gifts from R . Schuh ) . Gal4 and UAS strains . btl-gal4 on 2nd and 3rd chromosomes ( gifts from S . Hayashi ) [58] , UAS-abdA ( a gift from F . J . Diaz-Benjumea ) [115] , UAS-ci75 and UAS-ciH4P ( gifts from S . Ishii ) [84] , UAS-cirep ( a gift from A . Moore ) [83] , UAS-kni and UAS-knrl ( gifts from R . Schuh ) [25] . UAS-AbdB , UAS-Axn-GFP , UAS-salm and UAS-Ubx were obtained from BDSC . Egg collection was done with apple/grape juice plate at 25°C . Embryos were bleached and fixed as previously described [118] for 15–30 minutes with a 1:1 mixture of heptane and a fix solution ( 3 . 7% formaldehyde , 0 . 1M Hepes pH6 . 9 , 2mM MgSO4 ) . Embryos were dechorionated with methanol and incubated in 0 . 1% PBT supplemented with 0 . 5% BSA . Staging of embryos was done as previously described [119] . For immunostaining the following primary antibodies were used: Guinea-pig anti-AbdA ( 1:500 , a gift from B . Gebelein ) [104] , rabbit anti-Dys ( 1:500 , a gift from L . Jiang ) [18] , Guinea-pig anti-Gasp ( 1:1000 ) [95] , mouse anti-Ubx ( 1:10 , a gift from R . White ) [120] , Guinea-pig anti-Kni ( 1:300 ) , ( developed by J . Reinitz and distributed by Y . Hiromi , East Asian Segmentation Antibody Center , Mishima , Japan ) [121] , rabbit anti-Salm ( 1:200 , gifts from R . Barrio and T . Cook ) [107 , 122] , rabbit anti-Tnc C-terminal ( 1:1000 , a gift from Z . A . Syed and A . Uv ) [89] , rat anti-Trh ( 1:200 , a gift from D . Andrew ) [123] and rabbit anti-Trh ( 1:50 ) . Mouse anti-Abd-B ( 1:10 , donated by S . Celniker ) [124] , mouse anti-Cut ( 1:10 , donated by G . M . Rubin ) [125] , mouse anti-Hnt ( 1:10 , donated by H . D . Lipshitz ) [126] , mouse mab2A12 ( anti-Gasp ) ( 1:5 , donated by M . Krasnow , N . Patel and C . Goodman ) [17 , 95] , mouse anti-Ptc ( 1:10 , donated by I . Guerrero ) [127] and mouse anti-PTP10D ( 1:10 , donated by K . Zinn ) [128] were obtained from Developmental Studies Hybridoma Bank ( DSHB ) , Iowa , USA . Commercially available antibodies were anti-LacZ ( E . coli . β-Galactosidase ) antibodies made in goat ( 1:500 , Biogenesis ) or rabbit ( 1:1000 , Capel ) and anti-GFP antibodies made in rabbit ( 1:500 , JL-8 Clontech ) , mouse ( 1:1000 , GFP20 Sigma ) or goat ( 1:500 , ab6673 Abcam ) . Donkey or goat biotin- or fluorescently labeled secondary antibodies made against the host species of primary antibodies were purchased from Jackson Laboratories . Streptavidin coupled with AMCA , FITC or Cy5 were used when necessary . For mab2A12 detection TSA amplification ( PerkinElmer ) was used . Double fluorescent labeling with RNA probe and antibody was carried out as described [129] . The following cDNA clones were used to make hybridization probes; bnl ( a gift from M . Krasnow ) [20] , dpp [108] and unpg ( a gift from P . A . Beachy ) [54] . wg , salm and kni clones were obtained from Drosophila Genomics Resource Center ( DGRC ) , Indiana , USA . Confocal images were taken by Biorad MRC1024 , Oympus Fluoview 1000 or Zeiss LSM780 . Images of controls and mutants taken by the same confocal microscopes were used for comparison . Images were processed by ImageJ and figures were prepared with Photoshop and Illustrator . Quantification of Trh positive cell number of CB was done for stage 13 embryos stained with antibodies against Trh and DE-cad , scanned at 40× magnification with 0 . 6 um intervals . For quantification of DT tube thickness , mid-late stage 16 embryos stained with antibodies against PTP10D , Gasp and DE-cad were scanned at 40× magnification with 0 . 6 um intervals . Using ImageJ , 3 points next to the bases of DB were selected from each Z-stacked image for each metamere to measure the maximal distance of PTP10D positive apical membranes perpendicular to the longitudinal tube axis . 3 embryos for each genotype were measured . Average , SD and p-value of student t-test were calculated by Excel . | Tubes are common structural elements of many internal organs , facilitating fluid flow and material exchange . To meet the local needs of diverse tissues , the branching patterns and tube shapes vary regionally . Diametric tapering and specialized branch targeting to the brain represent two common examples of variations with organismal benefits in the Drosophila airways and our vascular system . Several extrinsic signals instruct tube diversifications but the impact of intrinsic factors remains underexplored . Here , we show that the local , tube-intrinsic Hox code instructs the pattern and shape of the dorsal trunk ( DT ) , the main Drosophila airway . In the cephalic part ( DT1 ) , where Bithorax Complex ( BX-C ) Hox genes are not expressed , the extrinsic Hedgehog signal is epistatic to WNT/Wingless signals . Hedgehog instructs anterior DT1 cells to take a long and narrow tube fate targeting the brain . In more posterior metameres , BX-C genes make the extrinsic WNT/Wingless signals epistatic over Hedgehog . There , WNT/Wingless instruct all DT cells to take the thick and short tube fate . Moreover , BX-C genes modulate the outputs of WNT/wingless signaling , making the DT tubes thicker in more posterior metameres . We provide a model for how intrinsic factors modify extrinsic signaling to control regional tube morphologies in a network . | [
"Abstract",
"Introduction",
"Results",
"and",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | The Intersection of the Extrinsic Hedgehog and WNT/Wingless Signals with the Intrinsic Hox Code Underpins Branching Pattern and Tube Shape Diversity in the Drosophila Airways |
Replicative aging has been demonstrated in asymmetrically dividing unicellular organisms , seemingly caused by unequal damage partitioning . Although asymmetric segregation and inheritance of potential aging factors also occur in symmetrically dividing species , it nevertheless remains controversial whether this results in aging . Based on large-scale single-cell lineage data obtained by time-lapse microscopy with a microfluidic device , in this report , we demonstrate the absence of replicative aging in old-pole cell lineages of Schizosaccharomyces pombe cultured under constant favorable conditions . By monitoring more than 1 , 500 cell lineages in 7 different culture conditions , we showed that both cell division and death rates are remarkably constant for at least 50–80 generations . Our measurements revealed that the death rate per cellular generation increases with the division rate , pointing to a physiological trade-off with fast growth under balanced growth conditions . We also observed the formation and inheritance of Hsp104-associated protein aggregates , which are a potential aging factor in old-pole cell lineages , and found that these aggregates exhibited a tendency to preferentially remain at the old poles for several generations . However , the aggregates were eventually segregated from old-pole cells upon cell division and probabilistically allocated to new-pole cells . We found that cell deaths were typically preceded by sudden acceleration of protein aggregation; thus , a relatively large amount of protein aggregates existed at the very ends of the dead cell lineages . Our lineage tracking analyses , however , revealed that the quantity and inheritance of protein aggregates increased neither cellular generation time nor cell death initiation rates . Furthermore , our results demonstrated that unusually large amounts of protein aggregates induced by oxidative stress exposure did not result in aging; old-pole cells resumed normal growth upon stress removal , despite the fact that most of them inherited significant quantities of aggregates . These results collectively indicate that protein aggregates are not a major determinant of triggering cell death in S . pombe and thus cannot be an appropriate molecular marker or index for replicative aging under both favorable and stressful environmental conditions .
Replicative aging in unicellular organisms is defined by a gradual increase in generation time and probability of death as cell divisions increase . In cases of asymmetrically dividing unicellular organisms such as Caulobacter crescentus , Saccharomyces cerevisiae , and Candida albicans , aging is manifested and linked to morphological asymmetry [1–3] . The situation , however , is less clear for symmetrically dividing organisms . While some evidence suggests replicative aging in old-pole cell lineages of Escherichia coli [4–6] , Wang et al . reported that growth rates of E . coli old-pole cells did not significantly alter over 200 generations , despite the gradual increases in filamentation and death rates [7] . For the symmetrically dividing fission yeast S . pombe , earlier studies suggested replicative aging by observation of asymmetry in cell volume at divisions followed by the deaths of the larger cells and asymmetric segregation of carbonylated proteins ( one of the biomarkers of oxidative stress ) . Additionally , it was suggested that inheritance of carbonylated proteins and a birth scar might inversely correlate with survival probability [8–10] . In a more recent study , however , Coelho et al . showed that potential aging factors such as an old pole , a new spindle pole body , and protein aggregates did not correlate with generation time , suggesting that S . pombe does not age , at least under favorable conditions [11] . The key mechanism to generate aging lineages ( and their rejuvenated counterparts ) is thought to be asymmetric segregation of “aging factors , ” regardless of the mode of cell division . Among the potential aging factors are aggregates of misfolded proteins [12 , 13] . In E . coli , naturally occurring protein aggregates reside exclusively at old-pole ends , probably because of nucleoid occlusion , and a negative correlation between the aggregate burden and growth rate is observed [5 , 6] . Likewise , in S . cerevisiae , protein aggregates are preferentially found in aging mother cells . Active mechanisms that have been suggested to contribute to asymmetric segregation include organelle-associated confinement and actin cable-dependent retrograde flow [14–17] . Asymmetric damage segregation and its association with aging were also suggested in S . pombe cultured under heat or oxidative stress conditions [11] . Although it was previously shown how aggregates are formed , move , and segregate during cell division , it is not clearly established whether they contribute to an increased death rate [11 , 18] . To study the aging process of yeasts , lineage tracking on an agar plate is conventionally performed [19] . This requires micromanipulation to remove daughter cells and is relatively labor intensive , thus precluding high-throughput and long-term analyses . An increasing number of studies at the single-cell level for various model organisms utilize microfluidic devices made of polydimethylsiloxane ( PDMS ) , a chemically stable and biocompatible silicone , in combination with automated microscopic imaging techniques [20–25] . One such device , termed the “Mother Machine , ” was originally developed to track E . coli old-pole cell lineages with considerably higher throughput ( 105 individual old-pole cells ) and longer duration ( 200 generations ) than previous studies [7] . Mother Machine-like microfluidic devices for S . pombe have been reported recently , and they demonstrated the absence of replicative aging in rich medium [26 , 27] . In this work , we measured more than 1 , 500 fission yeast old-pole cell lineages up to 80 generations using a custom-built Mother Machine-like microfluidic device . By measuring cell division and death rates in 7 different balanced growth conditions , we confirmed the absence of replicative aging in old-pole cell lineages in all of the tested environments and found a positive correlation between the division and death rates . We observed formation , growth , inheritance , and asymmetric segregation of Hsp104-associated protein aggregate in the old-pole lineages and demonstrated that inheritance and quantity of protein aggregate affected neither generation time nor triggering of cell death . In addition , a large amount of protein aggregate induced by transient stress treatment could also be tolerated without affecting cellular growth rates . Collectively , our results suggest that protein aggregate does not serve as an aging marker under both favorable and stressful conditions .
We designed and developed a microfluidic device for long-term tracking of old-pole cell lineages of S . pombe ( Fig 1A and 1B and S1 Fig ) . Our device has essentially the same architecture as the “Mother Machine , ” which was originally developed by Wang et al . for studying aging and growth in E . coli [7] , except that the dimensions of the internal channels were scaled up for fission yeast , which are physically larger . During time-lapse experiments , the device was constantly supplied with fresh medium to keep the environmental conditions around the cells unchanged . We experimentally confirmed that the medium reached the ends of the observation channels within 5 min , both in the absence and presence of cells ( S2 Fig and S1 Movie ) . Cells grew and divided aligned in the observation channels , and cells that spilled out from the observation channels to the trench were washed out by the flow of medium ( Fig 1B and S2 Movie ) . These settings allowed us to follow the division dynamics of cells located at the ends of the observation channels , referred to as old-pole cells ( or mother cells ) , typically for 50–80 generations . Time-series data on cell size ( determined by visualized cell area ) for every cell lineage were extracted from a set of time-lapse images ( Fig 1C ) . We analyzed more than 1 , 500 single-cell lineages in each experiment , which is comparable to , or larger than , the numbers of cell lineages analyzed in similar microfluidic experiments [7 , 27] . We performed time-lapse experiments employing 7 different culture conditions with different media ( yeast extract medium [YE] or Edinburgh minimal medium [EMM] ) and temperatures ( see S1 Table for the summary of all measurements ) . Plotting the cumulative division probability against time confirmed that division rates were strikingly stable except during the initial measuring ( Fig 1D and S3 Fig ) . This early instability in division rates reflected a lag in cell recovery from the slow-growing state that follows the loading of the cells into the microfluidic device ( see Materials and methods ) . Population doubling times calculated from the distributions of generation times ( see [28–30] for reference ) were close to those determined in batch culture experiments in the same media and temperature conditions ( Fig 1E and S2 Table ) . This indicates that the medium exchange rate in the device is sufficiently high . The stability in division rates for 50–80 generations , in turn , suggests an absence of deterioration in the reproductive ability of old-pole cells , under favorable culture conditions . Despite these favorable growth conditions , we observed the deaths of individual cells at low frequencies over the entire observation period ( Fig 2A , S2 Movie and S1 Table ) . Because the death events were observed throughout the time-lapse experiments and in every imaged position , they were not caused by temporal and/or local alterations in culture environments . The behaviors of cells destined for death were heterogeneous but could be broadly categorized into 3 types: Type I ( swollen ) , Type II ( hyperelongated ) , or Type III ( shrunken ) . Approximately 80% of the death events were categorized as Type I , and in almost all of these cases , siblings in the same observation channel synchronously died ( S4D and S4E Fig and S2 Movie ) . These observations are consistent with a recent report using a similar microfluidics system [27] . The synchronous deaths were also observed in another PDMS microfluidic device , where the observation channels accommodate greater numbers of cells than the Mother Machine . Importantly , we observed the synchronous deaths even when the dying siblings were spatially separated , whereas the other surrounding cells continued dividing normally ( S4F Fig and S3 Movie ) . These findings suggest that the synchronous deaths are not induced by local environmental changes in channels but are triggered in their common ancestor cells . We did not detect any preceding progressive signatures in the growth and division histories of dead cells . For example , the transitions in generation times of the extinct cell lineages were indistinguishable from those of the surviving lineages; no obvious or discernible increase in generation times was observed prior to cell deaths ( Fig 2B ) . The stability of generation times in the extinct cell lineages was further confirmed by comparing the means and standard deviations of generation times for each generation between the surviving and extinct lineages ( Fig 2C and S5 Fig ) . In addition , the number of surviving cell lineages decayed exponentially with time and generation count ( Fig 2D and 2E and S6 Fig ) , indicating that cell deaths occurred randomly with fixed probabilities and that every lineage exhibited an equal chance of abrupt death . The death rates estimated from the decay curves were small , in the order of 10−5 per min ( or 10−2 per generation ) . We noticed that our standard fluorescence imaging conditions induced weak photodamage [31] . Consequently , the estimated death rates were slightly higher than the death rates obtained by bright field imaging alone , but the constancy of the death rates was unaltered ( S4B and S4C Fig ) . We next investigated how cellular division and death rates might be interrelated . As presented in Fig 3A , we found that the death rate increased linearly with the division rate . The values of each data point ( including error estimates ) and the corresponding culture conditions are summarized in S3 Table . For example , in YE at 34°C , the division rate was r = ( 8 . 94 ± 0 . 01 ) × 10−3 min-1 ( mean generation time τb = 1/r = 112 min ) , and the death rate was k = ( 1 . 1 ± 0 . 1 ) × 10−4 min-1 ( characteristic lifetime τd = 1/k = 9 . 1 × 103 min = 6 . 3 days ) . Additionally , in EMM at 28°C , the division rate was r = ( 4 . 29 ± 0 . 01 ) × 10−3 min-1 ( τb = 233 min ) , and the death rate was k = ( 2 . 0 ± 0 . 4 ) × 10−5 min-1 ( τd = 5 . 0 × 104 min = 35 days ) . Thus , fast growth significantly shortens the lifetime of single cells . Reformatting the plot reveals that the expected life span of single-cell lineages in units of generation ( τd/τb ) also decreases with division rate , asymptotically approaching the minimum bound ( the shortest expected life span ) of approximately 50 generations ( Fig 3B , gray broken line ) . The decrease in expected life span can be attributed to the fact that the death rate reaches 0 with a positive division rate value ( rmin = [3 . 6 ± 0 . 2] × 10−3 min-1 , equivalently τbmax = 280 min ) , as shown in Fig 3A . It should be noted here that Spivey et al . has recently reported that the death probability per generation of h- 972 ( the same strain used in our study ) grown in rich medium was about 1 . 8% where the mean generation time was about 130 min [27] . These values correspond to the division rate r = 7 . 7 × 10−3 min-1 and the death rate k = 1 . 4 × 10−4 min-1 . The linear relation in Fig 3A predicts that k = 8 . 6 × 10−5 min-1 when r = 7 . 7 × 10−3 min-1 , which is 1 . 6-fold lower than their experimental value but reasonably close despite the different configurations of the microfluidic devices . Our observation that fission yeast old-pole cell lineages are unlikely to undergo replicative senescence motivated us to proceed to monitor long-term dynamics of protein aggregation to gain insight into how these lineages avoid aging . In general , protein aggregates are associated with molecular chaperones and heat shock proteins that can extricate protein monomers from the aggregates and refold them into their native structures [12 , 13] . One of the most well-studied heat shock proteins in yeasts is Hsp104 , an ATP-dependent disaggregase that is often used as a molecular marker of protein aggregation in both S . cerevisiae and S . pombe . A strain that expresses Hsp104–green fluorescent protein ( GFP ) from the native chromosomal locus was observed in the microfluidic device for approximately 50 generations ( Fig 4A and S4 Movie ) . The time-lapse imaging revealed that most healthy growing cells had 0 or 1 major GFP focus . We defined aggregates as a set of connected pixels whose fluorescence ( i . e . , GFP ) intensity exceeded a defined threshold value and quantified aggregate amounts by integrating fluorescent intensity within the connected area ( S7 Fig ) . The distribution of aggregate amounts was roughly exponential ( Fig 4B ) , which is consistent with previously reported results [18] . Fig 4C presents a representative dynamic of formation , growth , and segregation of protein aggregates in an old-pole cell . Once formed at an old-pole end , the ( major ) aggregate grew and tended to remain at the pole for many generations , but it occasionally migrated toward the new-pole end and was subsequently segregated to the new-pole cell ( S4 Movie ) , which is qualitatively consistent with an earlier report [18] . The distribution of aggregate inheritance duration , which is defined as the time interval between 2 successive “born-clean events” in units of generation , had a peak at 4 generations with an extended tail to the right and spreading over more than 40 generations ( Fig 4D ) . The tail can be approximately fitted by an exponential curve with a decay rate of λ = 0 . 13 ( generation-1 ) , suggesting that the segregation of protein aggregate to a new-pole cell is a random process that occurs once in every 1/λ = 7 . 8 generations on average . These results revealed that fission yeast old-pole cell lineages could escape from the burden of protein aggregate . Although the mean generation time of the old-pole cell lineages was stable ( Fig 2B and 2C ) , heterogeneity in each cell cycle length might be related to protein aggregation . We quantified the load of protein aggregation using 2 metrics: ( 1 ) aggregate amount and ( 2 ) aggregate age , the latter being defined as elapsed time ( in units of generation ) since the last birth without aggregate inheritance ( indicated by “Born clean” bars in Fig 4C ) . The former evaluates the current load of aggregation , whereas the latter evaluates the burden of possessing the aggregate for prolonged periods . We first simply plotted generation time against ( cell cycle-averaged ) aggregate amount ( Fig 5A ) and aggregation age ( Fig 5B ) , detecting no correlations . To analyze such relations in greater detail , we partitioned the data points in Fig 5A and 5B into 3 classes ( low , middle , and high ) according to the aggregation metrics ( aggregate amount or aggregation age ) and compared the generation time distributions among the classes ( Fig 5C and 5D ) . The distributions were essentially identical among the classes for both aggregation indices , which strongly indicates that cell cycle length is unaffected by protein aggregation . Next , we examined if the amount and inheritance of protein aggregation trigger cell death . Fig 6A illustrates protein aggregation dynamics ( Hsp104-GFP aggregate amount ) along with the level of constitutively expressed protein ( mCherry mean fluorescence intensity corresponding to its cellular concentration ) for both survived and extinct lineages . Typically , cells destined for death exhibited accelerated accumulation of protein aggregates immediately prior to death ( Fig 6A [top panel] , after 4 , 500 min ) ; we detected accelerated accumulation in 79% ( 427 out of 541 ) of extinct lineages . The commencement of accelerated accumulation was detectable by clear kinks in the transitions of aggregate amounts . These kinks seem to identify the time of initiation of cell death because other apparent functional deteriorations , such as radical increases in mCherry expression and changes of cellular morphology , also initiated concurrently ( Fig 6A and 6B and S4 Movie ) . Interestingly , even after the onset of these abnormalities , cell division occurred a few times before death ( Fig 6C ) , which might underlie the observed synchronized deaths ( S4D , S4F and S4G Fig ) . Due to the occurrence of accelerated accumulation in many extinct lineages before deaths , the distribution of aggregates at the death points was shifted toward greater values ( Fig 6D ) . However , the distribution of the amount at the kink points was very close to that for the total population ( Fig 6D ) , which suggests that large aggregate amounts are not required for initiating the process of dying . To reinforce this observation , we counted the numbers of cells that exhibited deaths and kinks for given ranges of aggregate amount ( Fig 6E ) and evaluated the probability of death ( Fig 6F ) and of starting accelerated accumulation ( Fig 6G ) . The results showed that the probability of commencing accelerated accumulation did not increase with the aggregate amount , although that of observation of cell deaths was elevated , which again suggests that the aggregate amount is not causative of initiation of the dying processes . We also investigated whether retention of protein aggregates increased the probabilities of death and of starting accelerated accumulation by counting the numbers of cells that showed deaths and kinks for given aggregation age ( Fig 6H ) . We found that both probabilities were nearly constant , irrespective of aggregation age ( Fig 6I and 6J ) . The death probabilities for aggregation age < 3 generations were slightly lower than the total death probability ( 1 . 15 × 10−2 per generation ) ( Fig 6I ) , possibly because of an identified lag of a few generations before death , after the onset of accelerated accumulation ( Fig 6C ) . Indeed , the probabilities of commencing accelerated accumulation were equally high for these small aggregation-age generations ( Fig 6J ) . Overall , our data suggest that Hsp104-associated protein aggregation is unlikely to play a major role in initiating the dying process . Our results , however , do not exclude the possibility that rapid accumulation of protein aggregate might accelerate completion of the dying processes post onset . It has been suggested that fission yeast ages upon stress treatment , and inheritance of large protein aggregate results in increased death probability [11] . To see if these aging phenotypes are also observed in our system , we transiently treated cells with hydrogen peroxide ( H2O2 ) , a commonly used oxidative stressor , and monitored cell division/death kinetics along with protein aggregation dynamics . As expected , cells immediately ceased to divide upon stress treatment ( Fig 7A , around t = 6 , 000 min ) . After removal of the stress , there was a lag ( around t = 6 , 000–6 , 500 min ) before cells resumed dividing . Strikingly , once cells started to grow again , the division rate was almost the same as that previously observed under unstressed conditions ( Fig 7A ) . The survival curve in Fig 7B revealed an increase in death rate upon stress treatment ( approximately 10% cells died during 1 h of exposure to hydrogen peroxide ) . Although the recovery was slower than that of division rate , the death rate also returned to the normal level seen in the unstressed condition . We did not observe the progressive increase of generation time after stress removal , one of the hallmarks of replicative aging; marked increase of generation time was seen only in the first generation after stress removal ( Fig 7C ) . These results suggest that the apparent deterioration in cellular growth/death is a transient response to the stress and not a manifestation of aging . We next asked how protein aggregation dynamics are related to the stress response . As reported earlier , we observed that oxidative stress enhanced protein aggregation , and many lineages accumulated aggregate to high levels not attainable in normal conditions ( Fig 7D ) . When the cells re-entered division cycles after stress removal , the large amount of aggregate persisted ( and even continued to grow in some cases ) . Strikingly , even such significant amounts of aggregate did not affect generation time ( Fig 7E ) . In nonstressed conditions , the amount of aggregate did not exceed 400 ( × 103 a . u . ) in 90% of extinct lineages , whereas approximately 40% of the lineages that survived the stress treatment until the end of measurement experienced more aggregation than 400 ( × 103 a . u . ) ( Fig 7F ) . These results further support that the absolute amount of protein aggregate does not determine growth kinetics or cell fates . This contrasts with the previous report [11] , in which the authors concluded that S . pombe exhibits aging under stressful conditions . To further examine if protein aggregation can result in cellular aging and/or cell death in S . pombe , we ectopically expressed a truncated version of the orthoreovirus aggregation-prone protein , μNS , which was N-terminally tagged with mCherry or mNeonGreen for visualization by fluorescent microscopy [32–34] . We confirmed that μNS formed aggregates in the cytoplasm of S . pombe , which are detectable as bright foci in many cells ( Fig 8A ) . mCherry-μNS and Hsp104-GFP foci did not colocalize , which indicates that not all protein aggregates were associated with Hsp104 ( Fig 8A and S5 Movie ) . Formation and segregation dynamics of mNeonGreen-μNS aggregate were similar to those of Hsp104-GFP , other than that accelerated accumulation before cell death was not observed ( Fig 8B , S8A Fig and S5 Movie ) . Generation time was not correlated with either aggregate amount or aggregation age ( Fig 8C ) , and distribution of μNS aggregate amount at death points was almost identical to that at the termination of the measurements for the survived lineages ( Fig 8D ) . These results suggest that , as noted for endogenous Hsp104-associated protein aggregation , induction of ectopic protein aggregation causes no functional loads on S . pombe cells in terms of the rates of cell division and initiation of death . Finally , we examined if hsp104+ gene disruption results in replicative aging in the old-pole cell lineages . As expected , hsp104Δ strains were sensitive to heat shock ( Fig 9A ) . The deletion , however , did not affect the status of μNS aggregate in terms of both inheritance duration and distribution of the amount ( S8A and S8B Fig ) . Likewise , generation time remained noncorrelated to aggregation ( S8C Fig ) , and distribution of aggregate amount at death points was similar to that at the end points of the survived lineages ( S8D Fig ) . Division and death rates for the deletion mutant were essentially identical to those in the wild-type strain ( Fig 9B and 9C ) . Taken together , our data suggest that regulation of protein aggregation by Hsp104 is not critical for avoiding replicative aging in the old-pole lineages of S . pombe .
The observation that both division and death rates were constant over tens of generations strongly suggested that the fission yeast old-pole cell lineages were free from replicative aging ( Figs 1 and 2 ) . This endorses the conclusion of a recent report on the absence of replicative aging in fission yeast in favorable conditions [11 , 27] . However , we cannot formally exclude the possibility that aging of fission yeast occurs over a longer timescale than the observed durations in our experiments and in other recent work [11 , 27] . Regarding the discrepancy with the earlier reports , in which replicative aging was suggested [8 , 10] , it is of note that in the cited mortality assays , the populations became extinct within 20 generations , while about 80% of cells survived at that generation in our experiments ( Fig 2E ) . This implies that the physiological states of the observed cells in the experiments reported elsewhere were significantly different from those in ours . Recently , Spivey et al . have also shown the lack of replicative aging in old-pole cell lineages of S . pombe under favorable conditions using a similar microfluidics device [27] . Importantly , they observed that the death rates were variable among different strains of S . pombe even under the same culture conditions . For example , the death rate of NCYC132 strain was 5-fold lower than that of h- 972 . On the other hand , we have shown that the death rates of one S . pombe strain ( h- 972 ) are variable under different growth conditions and characterized the trade-off between fast growth and death ( Fig 3 ) . It is then natural to ask how strongly the division-death trade-off depends on the genetic backgrounds of different strains . This important subject of future studies might provide insights into the generality and the origin of the trade-off relation . Our experiments involving oxidative stress exposure in the microfluidic device revealed that generation time and death rate reverted to normal after removal of the stressor , although most old-pole cells continued to inherit and accumulate even more Hsp104-associated protein aggregates ( Fig 7 ) . This result is inconsistent with the previous report [11] , and the widely accepted notion that S . pombe ages under stressful conditions . Because oxidative stress conditions used in our study ( 1-h exposure to 2 mM H2O2 ) are more severe than those in [11] ( 1-h exposure to 1 mM H2O2 ) , the absence of noticeable aging phenotypes cannot be explained by an insufficient dose of stress . One of the possible causes of the discrepancy might be the difference in the culture conditions after stress treatment: cells were grown in microcolony on an agarose pad in [11] , whereas they were grown under a microfluidic device with a continuous liquid medium supply in our study . The lack of progressive increases in generation time suggests that the observed increases of generation time , death rate , and the amounts of Hsp104-associated protein aggregation are responses provoked by the stress exposure , not the signature of aging . Therefore , aging under stressful conditions might not be a general trait in S . pombe . The lack of noticeable aging in S . pombe old-pole lineages contrasts with that in E . coli cultured under favorable growth conditions in the Mother Machine , in which division rates of old-pole cells were stable , but death rates increased over generations [7] . It is interesting to note that the modes of cell wall syntheses at poles are quite different between the 2 symmetrically dividing microorganisms . S . pombe employs polar growth , and newly synthesized cell wall materials are exclusively incorporated at the ends of the cells [35–37] , whereas in E . coli , the cylindrical part of the cell grows and cell walls at poles are thought to be metabolically inert and unable to avoid deterioration [38–40] . Thus , for fission yeast we do not have an a priori reason to believe that old-pole lineages should undergo senescence . Rather , new-pole lineages that inherit a larger proportion of old lateral cell walls and a birth scar ( due to the delay for the new-pole end to initiate growth after division [41] ) might be subject to aging . It should be stressed that our results do not rule out the existence of any forms of lineage-specific aging in fission yeast , e . g . , lineages that inherit damaged ( carbonylated ) proteins and/or birth scars [10] . Since the Mother Machine allows tracking of only old-pole cell lineages , new methods are required to track such cell lineages within proliferating populations . Our results might also imply that symmetrically dividing unicellular organisms could escape aging if they possess efficient damage repair mechanisms . On the other hand , systematic asymmetric partitioning of most cellular components should occur in nonsymmetrically dividing unicellular organisms , resulting in consistent unequal partitioning of the damaged materials to specific lineages , associated with cellular morphological features such as cell size . This might underlie the empirical fact that aging is more prominent in asymmetrically dividing unicellular organisms . A common perception is that protein aggregate accumulates during the aging process or the stress response , and cells die catastrophically when the aggregation load exceeds the cellular capacity . Our data , however , indicated that Hsp104-associated protein aggregate is also formed in aging-free cell lineages ( Fig 4 and S4 Movie ) . We showed that neither the aggregate amount nor the retention time affected the generation time ( Fig 5 ) . In addition , we demonstrated that cells transiently exposed to oxidative stress could promptly resume normal growth , even in the presence of unusually large amounts of protein aggregate induced by such stress ( Fig 7 ) . The commonly observed correlation between protein aggregation and cell death is most likely explained by the accelerated accumulation of aggregate a few generations before cell death . The commencement points of accelerated accumulation appear to specify the initiation points of the dying processes because the other abnormalities of mCherry expression levels and cellular morphology started around the same time . The initiation of the dying process occurred irrespectively of aggregate quantity , which argues against the concept that there is an absolute threshold in protein aggregation burden for triggering cell death . The results also suggest that retention of the aggregates did not elevate the cell death probability ( Fig 6H–6J ) . Overall , our data indicate that Hsp104 foci do not reflect gradual deterioration of proteostasis in the cells and thus cannot be used as the sole molecular marker for cellular senescence in fission yeast . Our results do not exclude the possibility , however , that rapid accumulation of protein aggregates might accelerate completion of the dying processes post onset . For example , it has been suggested that large protein aggregates tend to overlap division plane and increase the frequency of death [11] . Characterizing in more detail the intracellular events that occur concurrently with the accelerated accumulation might unravel the new roles of the protein aggregates in S . pombe . Together with the results regarding Hsp104-associated protein aggregation , our experimental results involving the ectopic aggregation-prone protein , μNS , suggest that S . pombe is , in fact , highly tolerant of protein aggregation loads ( Fig 8 ) and that their adverse effects on growth and death and their relation to aging might have been overestimated . More insights would be gained by examining the correlations between the other types of foreign protein aggregation and cellular division and death . It should be noted that there can be multiple types of protein aggregation with different constituents and formation/degradation dynamics , such as stress foci/Q-bodies/CytoQ , immobile protein deposit ( IPOD ) , juxtanuclear quality control compartment ( JUNQ ) /intranuclear quality control compartment ( INQ ) , and age-associated deposits , as reported for budding yeast [17 , 42–44] . Hsp104 is enriched in stress foci , IPOD , and age-associated deposits , but typically not in JUNQ/INQ [17] . Although Hsp104 is assumed to represent total protein aggregates in fission yeast [18] , detailed classification and characterization of protein aggregates in this organism are still lacking . Therefore , future studies should clarify to what extent Hsp104-associated protein aggregates capture the behaviors of total protein aggregates in cells . We found that as the cell division rate elevated , the death rate increased in a linear fashion ( Fig 3A ) . Although a simple extrapolation of the linear trend predicts immortality of single cells when the division rate is below rmin , we have not been able to experimentally achieve stable growth with such a low division rate under our current measurement setup . The slow growth of cells with a division rate close to , or even smaller than , rmin could be achieved by applying stressors , such as high/low temperatures , nutrient limitation , drug exposure , or harsh chemical conditions ( e . g . , extreme redox environments and high/low osmolarity ) . However , stress responses would render internal cellular states different from those in nonstressed conditions . We speculate that this linear trend is a hallmark of a balanced growth state , rather than a universal constraint on cell division and death rates in any environment . Why might death rate be positively correlated with division rate ? In line with the historical free radical theory of aging , it could be contemplated that faster growth with higher metabolic rates generates greater oxidative stress and/or toxic metabolic wastes that result in cell death [45] . Although a substantial body of work supports the notion that reactive oxygen species ( ROS ) underlie aging and/or the reproduction-survival trade-off , skeptical views have also been proposed in recent studies [46–48] . In addition , caution should be exercised when applying these theories to aging and/or life history , given the lack of aging before death in our observations . Another , and not mutually exclusive , possibility is that cells might allocate energy and resources to growth and division at the expense of maintenance mechanisms such as DNA repair , protein quality control , and stress responses . Indeed , the expression of stress-induced genes is negatively correlated with growth rate in budding yeast [49] , and enhanced stress resistance in slower growing cells has been reported [50 , 51] . Demetrius proposed that it is neither metabolic rate nor specific metabolites , but rather the stability of the entire metabolic system that determines lifespan [52] . However , how external environments ( i . e . , temperature and/or nutrition status ) affect the robustness of steady-state levels of metabolites to stochastic fluctuations in metabolic processes is still poorly understood and would be an interesting topic for future research . In general , death can be triggered by both external and internal cues [53] . Because the microfluidic system ensures stable culture conditions by continuous supply of fresh medium , cellular deaths observed in this study are most likely caused by internal signals , although we cannot exclude the possibility that subtle fluctuations of local environments around cells had some impacts on death . DNA lesions are unlikely to be a major cause of death because the extensive elongation in cell length , a characteristic phenotype of DNA damage response [54] , was observed in only approximately 20% of cases ( S4E Fig ) . ROS are involved in many examples of programmed cell death from yeasts to humans [55–57] and thus are at least one of the significant candidates to be examined . Accordingly , it would be of great interest to visualize the dynamics of mitochondria and/or peroxisomes in our experimental system . A very recent study has reported that Sir2p ( silent information regulator 2: a highly conserved NAD+-dependent deacetylase ) overexpression or inhibition of the TOR pathway ( target of rapamycin: a highly conserved phosphatidylinositol kinase-related protein kinase that is responsible for nutrient responsive growth regulation ) by rapamycin decreases death rates in S . pombe [27] . Involvement of a variety of biological processes implicated by such observations , including heterochromatin regulation [58 , 59] , asymmetric partition of damaged proteins [10] , and nitrogen-starvation responses [60–62] , should be investigated in future research . At the present time , however , we do not understand if sporadic death is dependent on any specific molecular mechanism . For example , intrinsic fluctuations in gene expression and/or biochemical reaction rates may well differentially result in catastrophic alterations to cellular physiology . A variety of fluorescent biosensors for more global physiological states , such as intracellular pH , ATP , NAD+/NADH , molecular crowding , and second messengers—cAMP , diacylglycerol ( DAG ) , Ca++ , etc . —would be valuable and informative to infer the causes of cell death [63–68] . We hope that our simple but powerful microfluidics approach can contribute to the detailed study of the mechanisms of cell mortality and a deeper understanding of limitations in cellular homeostasis , one of the significant questions in biology .
The microfluidic device was fabricated by standard photolithography techniques ( S1B Fig ) [69–71] . The CAD designs , created using ZunoRAPID software ( Photron ) , were printed on photoresist-coated chrome-on-glass masks ( CBL4006Du-AZP , Clean Surface Technology ) using a laser-drawing system ( DDB-201-TW , Neoark ) . The UV-exposed regions of the photoresist ( AZP1350 ) were removed by NMD-3 ( Tokyo Ohka Kogyo ) , and the exposed chromium was etched by MPM-E350 ( DNP Fine Chemicals ) . After removing the remaining photoresist layer using acetone ( Wako ) , the masks were rinsed with MilliQ water and air-dried . An SU-8 mold for the PDMS device was made on a silicon wafer ( ϕ = 76 mm , P<100> , resistance 1 to 10 Ω·cm , thickness 380 μm; Furuuchi Chemical ) in 2 steps . First , to make observation channels , the wafer was coated with SU-8 3005 ( Nippon Kayaku ) using a spin-coater ( MS-A150 , Mikasa ) at 500 rpm for 10 s and then at 4 , 000 rpm for 30 s . After soft baking for 2 min at 95°C , the photoresist was exposed to UV using the mercury lamp of a mask-aligner ( MA-20 , Mikasa ) at 22 . 4 mW/cm2 for 12 s . Postexposure baking was performed at 95°C for 3 min , followed by exposure to SU-8 developer ( Nippon Kayaku ) and a 2-propanol ( Wako ) rinse . The same procedure was repeated to fabricate trenches using SU-8 3025 ( Nippon Kayaku ) at 500 rpm for 10 s and then at 2 , 000 rpm for 30 s . Soft baking was performed at 95°C for 10 min , followed by UV exposure at 22 . 4 mW/cm2 for 16 s and postexposure baking at 95°C for 10 min . The PDMS base and curing agent ( Sylgard 184 ) were mixed at a ratio of 10:1 , poured onto the SU-8 mold in a container , and degassed using a vacuum desiccator . Curing was performed at 65°C overnight . The device was peeled from the mold , washed briefly in ethanol with sonication , and then air-dried . After punching 2 holes in the device to connect the inlet and outlet tubes , the surfaces of the device and a coverslip ( 24 × 60 mm , thickness 0 . 12–0 . 17 mm , Matsumani ) were activated using a plasma cleaner ( PDC-32G , Harrick Plasma ) and bonded together . Finally , the inlet and outlet tubes were inserted into the holes ( see also S1 Fig ) . HN0025 , which expresses mVenus under the control of a constitutive adh1 promoter ( h- leu1-32::leu1+-Padh1-mVenus ) , was constructed by transforming HN0003 ( h- leu1-32 ) with the Not I-digested fragment of pDUAL-mVenus , generated by replacing the Sph I-Cla I fragment of pDUAL 13G10 with the Padh1-mVenus-nmt1 terminator cassette . HN0041 was constructed in the same manner as HN0025 , except that the mVenus fragment was replaced with mCherry , and used as a parental strain to establish HN0045 . To generate HN0045 , the GFP tag was fused at the C-terminus of Hsp104 by PCR-based gene targeting [72]: the 3′ end of the ORF and 3′ UTR region of hsp104+ were assembled with the GFP ( S65T ) -kanR fragment on the pFA6a-GFP ( S65T ) -KanMX6 to generate a targeting module . For the strain depicted in S1D Fig ( HN0034 ) , the Padh1-mCherry cassette was integrated at the ade6+ locus . To generate HN0060 , Ptef-mNeonGreen-μNS cassette was first cloned into pDUAL vector , and then a Not I-digested fragment was integrated to the leu1-32 locus . Note that the μNS is a truncated version ( 1 , 411–2 , 163 ) that corresponds to amino acids 471–721 . Genomic DNA of bacterial strain CJW4617 [34] ( a kind gift from Dr . M . Nibert at Harvard Medical School and Dr . C . J . Wagner at Yale University ) was used as a template for PCR cloning of the μNS cDNA . To disrupt hsp104+ , the host strains were transformed by a PCR-amplified deletion cassette , and G418-resistant clones were selected . A complete list of the strains and PCR primers used in this study can be found in S4 Table . The pDUAL 13G10 strain was provided by Riken BRC , a member of the National Bio-Resources Project of the MEXT , Japan [73 , 74] . Confocal fluorescence microscopic images of yeast cells in the microfluidic device ( S1D Fig ) were acquired using a Nikon Ti-E microscope equipped with a laser scanning confocal system , equipped with a 60× objective lens ( Plan Apo λ N . A . 1 . 4 , Nikon ) and under oil immersion . The resolution along the Z-axis was 0 . 15 μm , and 335 images ( corresponding to approximately 50 μm in height ) were taken . Three-dimensional reconstructions of the images were achieved using an ImageJ 3D viewer plug-in . For long-term time-lapse measurements , 10 mL of a log-phase culture of yeast cells at 28–34°C in YE containing 3% glucose or EMM containing 2% glucose was concentrated 50-fold by centrifugation and injected into the microfluidic device using a 1 mL syringe ( Terumo ) . Cells were loaded into the observation channels by gravity , simply slanting the device . The loading procedure typically took a couple of hours , during which time the cells often entered into an early stationary phase in response to the highly crowded environment , resulting in a time lag before stable growth was achieved . The device was supplied with appropriate medium supplemented with a low concentration ( 10 μg/mL ) of ampicillin sodium ( Wako ) to minimize the risk of bacterial contamination . Note that ampicillin has no effect on fission yeast growth . The flow rate was 10–15 mL/h . For transient oxidative stress treatment , the medium was changed to YE containing 2 mM hydrogen peroxide for 1 h and then switched back to YE . We used a Nikon Ti-E microscope with a thermostat chamber ( TIZHB , Tokai Hit ) , 40× objective ( Plan Apo λ N . A . 0 . 95 , Nikon ) , cooled CCD camera ( ORCA-R2 , Hamamatsu Photonics ) , and an LED excitation light source ( DC2100 , Thorlabs ) . Several cell divisions were allowed before initiating measurements . Micromanager software ( https://micro-manager . org/ ) was used for fluorescence and/or bright field image acquisition . The time-lapse interval was 3 min ( for the experiments described in Figs 1–3 ) or 5 min ( for the experiments described in Figs 4–9 ) . Exposure times were 400 ms ( for mVenus ) , 200 ms ( for GFP ) , 100 ms ( for mCherry ) , and 10 ms ( for bright field ) . The acquired fluorescence images were converted into binary images using a custom-written OpenCV program . The binary images were used to identify cellular regions or ROIs , and lineage tracking ( relating ROIs along lineages ) was performed using a customized ImageJ macro . The transition between ROIs along each lineage was analyzed to mark cell division points where an ROI area suddenly decreased more than 1 . 5-fold . To mark cell death points , 2 criteria were employed: ( 1 ) if there was no division during a 360-min window , then the beginning of the window was defined as a death point , and ( 2 ) if there was a profound ( more than 1 . 75-fold ) decrease in fluorescence during a 30-min time window , then the beginning of the window was defined as a death point . We confirmed that the decay curve of surviving cell lineages obtained using these death criteria quantitatively concurred with that obtained by manual image inspection ( S4A Fig ) . In the data set used in Figs 6–9 , we examined all of the cell-size trajectories by eye and manually marked death points so as to ensure confidence in the data . The death onset points ( = kinks on the aggregate amount trajectories ) were identified by manually inspecting the aggregate amount trajectory plots for all of the 541 extinct lineages . Aggregate amount/age at the kinks and generations to die after the onset of the dying process were subsequently recorded using a custom-developed ImageJ-plugin . Lineage tracking data for all the environments tested are deposited in the Dryad repository [75] . Generation time distributions obtained by following old-pole cell lineages in the Mother Machine represent intrinsic cellular division properties when 2 sister cells are physiologically indistinguishable . In standard batch cultures , however , selection occurs because of heterogeneities in cellular generation times , which cause the population doubling time ( Td ) to be smaller than the mean generation time 〈τ〉g=∫0∞τg ( τ ) dτ . Here g ( τ ) is the probability density function of an intrinsic generation time distribution in a given culture condition . When cells randomly and independently determine their generation times according to g ( τ ) , the population growth rate , Λ = ln2/Td , must satisfy the Euler-Lotka equation [28–30] , 1=∫0∞2e−Λτg ( τ ) dτ . ( 1 ) Equivalently , ( Eq 1 ) can be rewritten as 1=2Λ∫0∞e−ΛτB ( τ ) dτ , ( 2 ) where B ( τ ) ≡∫τ∞g ( τ′ ) dτ′ is the survival function ( complementary cumulative distribution ) of g ( τ ) , which represents the probability of a newly divided cell remaining undivided until age τ . One can show that 〈τ〉g=∫0∞B ( τ ) dτ , and then ψl ( τ ) =B ( τ ) /〈 τ 〉g , , becomes a probability density function and can be interpreted as “age distribution” along cell lineages . Therefore , Eq 2 can be also expressed as 1=2Λ〈τ〉g∫0∞e−Λτψl ( τ ) dτ . ( 3 ) We numerically estimate Λ based on a discretized version of Eq 3 , i . e . , 1=2Λ〈τ〉g∑ie−Λ ( iΔτ+Δτ2 ) ψi ( iΔτ ) Δτ ( i=0 , 1 , 2 ⋯ ) , ( 4 ) where Δτ is the time-lapse interval . 〈τ〉g was calculated as 〈τ〉g=1n∑i=1nτi , where τi is the generation time and n is the number of samples , and the probability distribution of age is calculated as ψl ( a ) Δτ=# of cells with age=a# of total cells . We calculated division rate r as r=〈1τ〉=1n∑i=1n1τi , where τi is the generation time and n is the number of samples . We estimated death rate k from the decay curve by the least squares fitting ( t versus ln ( N ( t ) ) , where t is time and N ( t ) is the number of surviving lineages at t ) . The expected value of “time to death” is thus 1k . We calculated “expected life span” in units of generation as rk . The standard errors for the division rates were calculated as S . E . division=σdivisionn=1n∑i=1n ( 1τi ) 2− ( 1n∑i=1n1τi ) 2n , where τi is the generation time and n is the number of samples . ±2S . E . division ranges were shown as error bars in Fig 3A and 3B . To evaluate errors in the death rate estimations , we produced simulated decay curves of the surviving fraction using parameters ( death rate , initial cell number , and observation period ) specific to each experiment . The simulation was repeated 5 , 000 times for each environment , and the death rate was obtained from a simulated survival curve in each run . The standard deviation of the determined death rates was calculated as σdeath , and the ±2σdeath ranges were shown as error bars in Fig 3A . Errors in expected life span ( σlifespan ) were calculated using the error propagation rule . We first estimated death probability per generation p0 to be 1 . 15 × 10−2 from the survival curve . In Fig 6F and 6I , the death probability p for each aggregate amount or aggregation age was then tested using a binomial test for the two-tailed null hypothesis H0: p = p0 at the significance level = 0 . 05 . For the onset probability of accelerated accumulation , we set the null hypothesis to be H0: q = 0 . 79 p0 = 9 . 09 × 10−3 based on our observation that a clear kink in the protein aggregation dynamics was detected in 79% of the extinct lineages and implemented binomial testing at the significance level = 0 . 05 ( Fig 6G and 6J ) . | Multicellular organisms universally senesce and must produce rejuvenated progenies in order to transmit life . Although similar age-related deterioration in physiological functions and reproduction is also found in unicellular organisms that divide asymmetrically to produce morphologically distinct aged and younger cells , it has been unclear whether symmetrically dividing microbes—such as fission yeast—exhibit the same traits . Using long-term live-cell microscopy combined with a microfluidic device , we monitor the growth and death of a large number of fission yeast cells and demonstrate the existence of aging-free lineages . These lineages are , however , not immortal , and the probability of death increases as the cells grow more rapidly; thus , the “live fast , die fast” trade-off exists in fission yeast . We further characterize the segregation and inheritance of protein aggregates , which are commonly thought of as “aging factors . ” The aging-free lineages bear the aggregate load for some generations with no apparent adverse effects on growth . We also show that there is no threshold amount of protein aggregate above which cells are destined to death in both normal and stressed conditions: protein aggregate is thus not a direct initiation signal for cell death . Our data reveal that protein aggregation might not be an appropriate index for aging and that we should revisit its role in cell physiology . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"death",
"rates",
"cell",
"death",
"fluorescence",
"imaging",
"medicine",
"and",
"health",
"sciences",
"engineering",
"and",
"technology",
"demography",
"cell",
"cycle",
"and",
"cell",
"division",
"protein",
"aggregation",
"cell",
"processes",
"physiological",
"proces... | 2017 | Aging, mortality, and the fast growth trade-off of Schizosaccharomyces pombe |
We present the genome sequences of a new clinical isolate of the important human pathogen , Aspergillus fumigatus , A1163 , and two closely related but rarely pathogenic species , Neosartorya fischeri NRRL181 and Aspergillus clavatus NRRL1 . Comparative genomic analysis of A1163 with the recently sequenced A . fumigatus isolate Af293 has identified core , variable and up to 2% unique genes in each genome . While the core genes are 99 . 8% identical at the nucleotide level , identity for variable genes can be as low 40% . The most divergent loci appear to contain heterokaryon incompatibility ( het ) genes associated with fungal programmed cell death such as developmental regulator rosA . Cross-species comparison has revealed that 8 . 5% , 13 . 5% and 12 . 6% , respectively , of A . fumigatus , N . fischeri and A . clavatus genes are species-specific . These genes are significantly smaller in size than core genes , contain fewer exons and exhibit a subtelomeric bias . Most of them cluster together in 13 chromosomal islands , which are enriched for pseudogenes , transposons and other repetitive elements . At least 20% of A . fumigatus-specific genes appear to be functional and involved in carbohydrate and chitin catabolism , transport , detoxification , secondary metabolism and other functions that may facilitate the adaptation to heterogeneous environments such as soil or a mammalian host . Contrary to what was suggested previously , their origin cannot be attributed to horizontal gene transfer ( HGT ) , but instead is likely to involve duplication , diversification and differential gene loss ( DDL ) . The role of duplication in the origin of lineage-specific genes is further underlined by the discovery of genomic islands that seem to function as designated “gene dumps” and , perhaps , simultaneously , as “gene factories” .
Aspergillus fumigatus is exceptional amongst the aspergilli in being both a primary and opportunistic pathogen as well as a major allergen associated with severe asthma and sinusitis [1]–[3] . It was first reported to cause opportunistic invasive infection about 50 years ago [4] . In immunocompromised patients , mycelial growth can proliferate throughout pulmonary or other tissues causing invasive aspergillosis . For these patients , the incidence of invasive aspergillosis can be as high as 50% and the mortality rate is often 50% , even with antifungal treatment . Since the late 1800's [2] , A . fumigatus has been demonstrated to be a primary pathogen of the airways , sinuses , lungs , damaged skin and subcutaneous tissues . For example , it can cause post-operative infection in all human organs [5] . In most cases diagnosis remains problematic and can compromise effective medical treatment . A . fumigatus is thought to possess particular metabolic capabilities and genetic determinants that allow it to initiate and establish an in vivo infection . This conclusion is supported by the observation that the majority of invasive aspergillosis disease is caused by A . fumigatus , even though its conidia comprise only a small percentage of the total conidia found in air-sampling studies [6] . While the interaction of A . fumigatus spores with the human respiratory mucosa is understood to an extent , the basic biology of the organism has until recently received little attention . Recently we presented the genomic sequence of A . fumigatus strain Af293 ( FGSC A1100 ) [7] isolated from a neutropenic patient , who died from invasive aspergillosis [8] . Its comparison with the genomes of two distantly related species , Aspergillus nidulans and Aspergillus oryzae , has led to many unexpected discoveries , including the possibility of a hidden sexual cycle in A . fumigatus and A . oryzae , and the detection of remarkable genetic variability of this genus [9] , [10] . Although members of the same genus , these three species are approximately as evolutionarily distant from each other at the molecular level as humans and fish ( Figures 1 and 2 ) [11] . This significant phylogenetic distance has hindered some aspects of comparative genomic analysis of the aspergilli such as identification of the genetic traits responsible for differences in virulence as well as in sexual and physiological properties . To maximize the resolving power of whole-genome comparative analysis , we selected the environmental type strains of a very closely related sexual species , Neosartorya fischeri NRRL181 ( A . fischerianus ) , and a more distantly related asexual species , A . clavatus NRRL1 , for complete sequencing . These three species are referred to here as the Affc lineage for A . fumigatus , N . fischeri , and A . clavatus ( Figure 2 ) . In contrast to A . fumigatus , N . fischeri is only rarely identified as a human pathogen [12]–[15]; while A . clavatus is probably an important allergen and the causative agent of extrinsic allergic alveolitis known as malt worker's lung [16] . A . clavatus also produces a number of mycotoxins and has been associated with neurotoxicosis in sheep and cattle fed infected grain worldwide ( e . g . [17] ) . Our phenotypic characterization ( Table S1 ) has shown that both A . fumigatus and N . fischeri can grow at 42°C , which indicates that A . fumigatus may possess other genetic determinants besides thermotolerance that allow it to establish a successful in vivo infection . As determined by multilocus sequence comparison , most A . fumigatus isolates , including Af293 and A1163 , lie within the main A . fumigatus clade and persist as a single , global phylogenetic population , presumably due to its small spore size [18] . Natural A . fumigatus isolates were described previously as having low genetic diversity in comparison to N . fischeri isolates [19] . However recent studies identified a number of strain-specific [7] and polymorphic [20] , [21] genes . To further explore the extent of genetic variation within the A . fumigatus species , we included in this analysis the genome sequence of a second strain , A1163 , made available through Merck & Co . , Inc . , Whitehouse Station , NJ . Our preliminary analysis has shown that Af293 and A1163 isolates vary greatly in their resistance to antifungals ( Table S2 ) .
The genome of A . fumigatus strain A1163 was sequenced by the whole genome random sequencing method [22] . Its genome ( 29 . 2 Mb ) is 1 . 4% larger than the genome of the first sequenced strain Af293 ( 28 . 8 Mb ) ( Table 1 ) . About 98% of each genome can be aligned with high confidence . Alignment of the A1163 genome against the eight Af293 chromosomes has revealed 17 large syntenic blocks , which correspond roughly to the 16 Af293 chromosomal arms ( Figure 3 ) . The syntenic blocks were defined as regions containing at least five syntenic orthologs separated by no more than 20 genes without orthologs . Most translocation events involving A . fumigatus chromosomes appear to have taken place within 300 Kb from the telomeres . The largest exchange involved a ∼500 Kb segment between Af293 chromosomes 1 and 6 and A1163 , which contain regions aligning with A1163 assembly 1 ( syntenic blocks 1 . 1 and 1 . 2 in Figure 3 ) . This appears to be a recent event that happened in A293 . In addition , Af293 chromosome 1 harbours a 400 Kb subtelomeric region that does not align well with A1163 assemblies . There is evidence of gene conversion between distal subtelomeric sequences encoding RecQ family helicases in A . fumigatus chromosomes 2 , 4 , and 7 . Consistent with previous reports [19] , the identity over the shared regions is very high ( 99 . 8% at the nucleotide level ) . This is higher than 99 . 3% and 99 . 5% identity between the two sequenced A . niger isolates ( ATCC 1015 and CBS 513 . 88 ) [23] and between A . oryzae [10] and A . flavus [8] , respectively . Unique regions represent 1 . 2% and 2 . 3% ( and harbour 143 and 218 genes ) in the Af293 and A1163 genomes , respectively . More than half of the Af293-specific genes are also absent in A . fumigatus isolates Af294 and Af71 , according to the array-based comparative genome hybridization ( aCGH ) data [7] . The vast majority of Af293- and A1163-specific genes are clustered together in blocks ranging in size from 10 to 400 Kb , which seem to be the most variable segment of the species genome . A manual examination of these isolate-specific islands revealed that they contain numerous pseudogenes and repeat elements . One of the regions contains a putative secondary metabolism cluster ( AFUA_3G02530-AFUA3G02670 ) . The origin of 20% of Af293-specific genes can be attributed to two segmental duplication events . One of the duplicated regions ( AFUA_1G16010- AFUA_1G16170 ) contains an arsenic detoxification cluster . The other ( AFUA_1G00420-AFUA_1G00580 ) contains genes that may be involved in metabolism of betaine , which is often synthesized under osmotic and heavy metal stress . Interestingly the duplicated regions are also absent in Af294 and Af71 isolates , which suggests that the duplication event took place very recently . Segmental duplication events are thought to contribute to rapid adaptation of the species by increasing their expression . Since Af293 is a clinical isolate it is possible that these chromosomal aberrations were created due to selective pressures in the host . Although most Af293 proteins are 100% identical to their A1163 orthologs , we have identified 41 orthologous pairs that share only 37% to 95% identity . To find out if these genes are also divergent in other A . fumigatus isolates , we identified Af293 genes that do not hybridize with DNA extracted from the Af294 and Af71 strains in aCGH experiments [7] . The comparison revealed that 27 out of 41 genes were possibly polymorphic ( marked as absent or divergent ) with respect to at least one other isolate ( Table S3 ) . Further analysis of three polymorphic loci in other A . fumigatus isolates has demonstrated that each of them harbours two or three alleles ( Table S4 ) . A PCR survey followed by Southern blot analysis and partial DNA sequencing has shown the presence of at least two alleles at each locus containing nearly identical sequences within each group of alleles ( data not shown ) . In filamentous fungi , this high level of variability has been previously associated with heterokaryon incompatibility ( het ) genes involved in a programmed cell death ( PCD ) pathway triggered by hyphal fusion between two genetically incompatible individuals [24] , [25] . So far several het loci have been described in A . nidulans [26] , although none have been characterized at the molecular level . Incidentally , our results are consistent with previously identified vegetative incompatibility groups suggesting that some of these polymorphic genes may function in heterokaryon incompatibility in A . fumigatus . Thus , four clinical isolates from the same multi-member incompatibility group ( WSA-270 , WSA-1195 , WSA-449 , and WSA-172 ) contained the same alleles of the polymorphic genes ( Table S4 ) . Furthermore , at least five putative A . fumigatus het genes exhibit a pattern of trans-species ( or trans-specific ) polymorphism ( Table S5 ) , which has been previously associated with somatic and sexual incompatibility in fungi , self-incompatibility in plants , and the major histocompatibility complex ( MHC ) in vertebrates . These genes are more similar to their orthologs from other Aspergillus species than to those from A1163 . We chose one putative het gene , rosA ( AFUA_1G15910 ) , and its close relative , nosA ( AFUA_4G09710 ) , whose orthologs encode two Zn2C6 transcriptional regulators of sexual development in A . nidulans [27] , [28] for phylogenetic analysis ( Figure 4 ) . Unexpectedly , Af293 RosA clusters with its A . clavatus ortholog , while A1163 RosA clusters with N . fischeri . This is in contrast with the NosA tree , which perfectly mirrors the species tree ( Figure 2 ) , suggesting that these allelic classes may transcend species boundaries in the aspergilli . This is the first study that shows the diversity of het genes in aspergilli at the molecular level as well as patterns of trans-species polymorphism . These putative het genes are distinct from those identified in Neurospora crassa or Podospora anserina [24] , [25] , although many of them share the same domains such as the NACHT and NB-ARC domains of the STAND superfamily [29] . Coincidentally four of the A . fumigatus variable genes encoding STAND domain proteins have previously been predicted to function in heterokaryon incompatibility [30] . The discovery of putative het loci in the aspergilli may facilitate identification of downstream components of fungal PCD pathways or other drug targets . These loci may be also used as a basis for classification of natural and clinical isolates into different compatibility groups . The genomes of N . fischeri and A . clavatus were sequenced by the whole genome sequencing method [22] . The N . fischeri genome ( 32 . 6 Mb ) is 10–15% larger than the A . clavatus and A . fumigatus genomes ( Table 1 ) . There are 10 , 407 protein-coding genes and a large number of transposable elements , which may have contributed to its genome size expansion . The A . clavatus genome ( 27 . 9 Mb ) is the smallest seen to date among the sequenced aspergilli ( Table 1 ) . There are currently 9 , 125 predicted protein-coding genes . This is consistent with past comparative studies that identified notable ( up to 30% ) genome size differences between distantly related aspergilli [7] , [9] , [10] . Despite this significant genome size variability , gene-level comparisons confirmed phylogenetic proximity of A . fumigatus , N . fischeri and A . clavatus ( Figures 1 and 2 ) . The three genomes also appear to be largely syntenic . Alignment of the N . fischeri and A . clavatus genomes against the eight Af293 chromosomes has revealed 20 and 55 syntenic blocks , respectively ( Table 2 ) . There is only one large-scale reciprocal translocation between chromosomes 2 and 5 in N . fischeri ( blocks 8927 . 1 , 8927 . 2 , 9292 . 1 and 9292 . 2 , in Figure 3 ) . The A . clavatus supercontigs align with A . fumigatus chromosomes 2 and 5 , suggesting that this was the ancestral topology . Previous studies however have shown a high level of evolutionary conservation and phyletic retention among known A . fumigatus virulence-associated genes [7] . Our analysis confirmed the low rate of protein evolution among these genes in four Aspergillus species ( Table S12 ) . Interestingly , four of the virulence-associated genes , pabaA ( AFUA_6G04820 ) , fos-1 ( AFUA_6G10240 ) , pes1 ( AFUA_1G10380 ) and pksP ( AFUA_2G17600 ) , reveal evidence of accelerated evolution in the branch leading to the two A . fumigatus isolates . This pattern can affect only a few amino acid residues ( e . g . PksP ) or a significant proportion of the protein ( e . g . Pes1 ) . Such a pattern can be due to either relaxation of selection or selection for rapid diversification ( positive selection ) . In the latter case specific amino acid substitutions may decrease susceptibility to specific environmental challenges and thus enhance A . fumigatus virulence . These four genes are involved in oxidative stress or nutrient availability , which is consistent with the positive selection scenario . Indeed , PabaA is involved in biosynthesis of folate , an essential co-factor for DNA synthesis . Since PABA is apparently limited in the mammalian lung , a functional pabaA gene is required for virulence [48] . Fos1 , a putative two-component histidine kinase , may play a role in the regulation of cell-wall assembly [49] . Finally , PksP and Pes1 are enzymes , which catalyze the first steps in biosynthesis of the spore pigment and an unknown non-ribosomal peptide , have been shown to mediate resistance to oxidative stress in addition to their role in A . fumigatus virulence [50] , [51] . The inclusion of additional taxa in the analyses might clarify the significance of the observed differences . This overall lack of variability among known virulence-associated factors suggests that yet unknown A . fumigatus-specific genes may contribute to its ability to survive in the human host . A recent microarray study demonstrated that the Affc-specific genes are over-represented among genes that are up-expressed in the neutropenic murine lung ( Elaine Bignell submitted for publication ) . Many of them are found in chromosomal gene clusters associated with macromolecule catabolism and secondary metabolite biosynthesis . Similarly , clustered lineage-specific genes simultaneously induced in infected tissue have been observed in the ubiquitous maize pathogen Ustilago maydis [52] and some other species ( for a recent review see [53] ) . Alternatively A . fumigatus virulence may be a combinatorial process , dependent on a pool of genes , which interact in various combinations in different genetic backgrounds as suggested previously [7] . Similar ‘ready-made’ virulence features have been described in other environmental pathogens such as Pseudomonas aeruginosa [54] and Cryptococcus neoformans [55] , [56] . In addition to virulence factors , the A . fumigatus genome encodes 20 allergens ( Table S13 ) and 25 proteins displaying significant sequence similarity to known fungal allergens ( Table S14 ) , some of which appear to contribute to its pathogenicity [57] . For example , A . fumigatus Asp f6 ( AFUA_1G14550 ) , also known as Mn2+-dependent superoxide dismutase ( MnSOD ) , is specifically recognized by IgE from patients with allergic bronchopulmonary aspergillosis ( ABPA ) and is differentially expressed during germination [58] . The broad distribution of allergens among fungal taxa ( Text S1 ) suggests that A . fumigatus possesses the same allergen complement as most other aspergilli and that its effect on hypersensitive individuals can be explained mostly by its ubiquity in the environment . Our analysis has demonstrated that , similar to known virulence-associated genes , most sexual development genes appear to be under negative ( purifying ) selection in both sexual and asexual Aspergillus species ( Text S1 and Table S15 ) . More detailed analysis has revealed four genes in the N . fisheri lineage that may be under positive selection . This suggests that a few amino acid changes may enable sexuality in N . fischeri . The conservation of sex genes in asexual species is due to a latent sexuality , a recent loss of sexuality , pleiotropy , or parasexual recombination following heterokaryon formation as suggested previously [59] , [60] . Lineage-specific ( LS ) genes ( i . e . genes with limited phylogenetic distribution of orthologs in related species ) have been the focal point of many comparative genomic studies , because of the assumption that they may be responsible for phenotypic differences among species and niche adaptation . Our analyses of the genomes of A . fumigatus and the two closely related species , N . fischeri and A . clavatus , demonstrates that A . fumigatus may possess genetic determinants that allow it to establish a successful in vivo infection . LS genes that have no orthologs in the other two species comprise 8 , 5% of the A . fumigatus genome and often have accessory functions such as carbohydrate and amino acid metabolism , transport , detoxification , or secondary metabolite biosynthesis . Further analysis showed that these genes have distinct features ( e . g . the small gene length and number of introns ) and tend to cluster in subtelomeric genomic islands , which may function as “gene dumps/factories” . The phylogenies of LS genes , their subtelomeric bias and size differences are consistent with the DDL hypothesis stating that duplication being the primary genetic mechanism responsible for the origin of species-specific genes . The presence of genomic islands indicates that A . fumigatus and may possess sophisticated genetic mechanisms that facilitate its adaptation to heterogeneous environments such as soil or a living host .
A . fumigatus Af293 ( FGSC A1100 ) was isolated from patients with invasive aspergillosis [61] . A . fumigatus A1163 ( FGSC A1163 ) is a derivative of A . fumigatus CEA17 converted to pyrG+ via the ectopic insertion of the A . niger pyrG gene [62] , [63] . CEA17 is a uracil auxotroph of A . fumigatus clinical isolate CEA10 ( CBS144 . 89 ) . The type strains of A . clavatus ( NRRL 1 ) and N . fischeri ( NRRL 181 ) were used for sequencing and phenotypic characterization . The genome sequences of A . clavatus , N . fischeri and A . fumigatus A1163 were deposited to the GenBank under the following accession numbers: AAKD00000000 , AAKE00000000 and ABDB00000000 , respectively . A1163 , A . clavatus and N . fischeri were sequenced using the whole genome shotgun method as previously described [22] . Random shotgun libraries of 2–3 Kb , 8–12 Kb and 50 Kb were constructed from genomic DNA from each strain , and DNA template was prepared for high-throughput sequencing using Big Dye Terminator chemistry ( Applied Biosystems ) . Sequence data was assembled using Celera Assembler . For A . fumigatus A1163 , scaffolds were compared to those of the first sequenced isolate , Af293 [7] . A1163 assemblies larger than 5 Kb were aligned to the Af293 chromosomes using the MUMmer package ( http://mummer . sourceforge . net/ ) [64] . Alignments longer than 100 Kb were used to determine average sequence identity to avoid highly repetitive and duplicated regions . The same approach was used to estimate sequence identity between A . flavus and A . oryzae and between the two sequenced A . niger strains . The JCVI eukaryotic annotation pipeline was applied to the A1163 , A . clavatus and N . fischeri assemblies ( supercontigs ) larger than 2 Kb as described earlier [7] . We used PASA [65] and EvidenceModeler [66] to generate consensus gene models based on predictions from several types of genefinders including GlimmerHMM , Genezilla , SNAP , Genewise and Twinscan . Putative pseudogenes , small species-specific genes ( less than 50 amino acids ) , and gene models overlapping with transposable elements ( TE ) shown in Table S16 were excluded from the final gene lists . Identification of repeat elements was performed using RepeatMasker ( http://www . repeatmasker . org/ ) , RepeatScout ( http://repeatscout . bioprojects . org/ ) , and Tandem Repeats Finder ( http://tandem . bu . edu/trf/trf . html ) . Putative TEs ( Table S16 ) were identified by Transposon-PSI ( http://transposonpsi . sourceforge . net ) , a program that performs tBLASTn searches using a set of position specific scoring matrices ( PSSMs ) specific for different TE families . TE and repeat densities were calculated as the percentage of nucleotide bases in the regions of interest ( i . e . , syntenic or non-syntenic blocks ) that overlap with a feature of the appropriate type ( repeat or TE ) . We leveraged the comparative genomic data to significantly improve annotation quality of the Af293 genome , which was previously annotated with relatively little supporting evidence [7] . The refinement of initial annotation was performed using the Sybil software package ( http://sybil . sourceforge . net/ ) , which allows for rapid identification of discrepancies in gene structure among orthologs . The comparison with orthologous N . fischeri and A . clavatus genes resulted in significant changes to the Af293 gene catalogue . Over 1100 gene models were updated and 130 new genes were identified . Initial A . fumigatus A1163 gene models were also improved using the PASA pipeline , initially developed to align expressed sequence tag ( EST ) data onto genomic sequences [65] . The pipeline was adapted to automatically update A1163 gene models by aligning them against Af293 coding sequences ( CDSs ) . We have performed transitive functional annotation from Af293 proteins to their A1163 , N . fischeri and A . clavatus orthologs . Previously GO terms [32] were assigned to Af293 proteins based on sequence similarity to PFAM domains or experimentally characterized S . cerevisiae proteins [7] . Secondary metabolism gene clusters were identified using Secondary Metabolism Region Finder ( SMURF ) available at http://www . jcvi . org/smurf ( Nora Khaldi , unpublished ) . The complete list of gene clusters can be downloaded at ftp://ftp . jcvi . org/pub/software/smurf/ . Gene Ontology ( GO ) terms [32] were assigned as described in [7] After extensive computational and manual refinement , the improved protein datasets were used to generate the final set of orthologs . Orthologous groups in Aspergillus genomes were identified using a reciprocal-best-BLAST-hit ( RBH ) approach with a cut-off of 1e-05 . In addition to the A1163 , A . clavatus and N . fischeri genomes , the previously sequenced genomes of Af293 [7] , A . terreus NIH2624 ( http://www . broad . mit . edu ) , A . oryzae RIB40 [10] , A . nidulans FGSC A4 [9] and A . niger CBS 513 . 55 [23] were included in the comparative analysis . The results of this analysis , as well as synteny visualisation and comparative analysis tools can be also found in the Aspergillus Comparative database at http://www . tigr . org/sybil/asp . Orthologous , unique and divergent genes in Af293 were identified based on alignments of Af293 CDSs against A1163 assemblies using gmap as implemented in PASA [65] using default parameters . Syntenic blocks for each pair of genomes ( Af293 vs . A . clavatus and Af293 vs . N . fischeri ) were defined as areas containing a minimum of five matching ( orthologous ) genes with a maximum of 20 adjacent non-matching genes ( having no orthologs ) in the reference and target genomes . Since most syntenic regions slightly overlapped , the original blocks were merged to calculate repeat and TE density . Af293 non-syntenic blocks were defined as areas excluded from the syntenic blocks and containing at least ten Af239 non-matching genes . Genes in four lineage-specificity groups were analyzed by the EASE module [67] in MEV within TM4 ( http://TM4 . org ) [68] to identify overrepresented Gene Ontology ( GO ) terms , Pfam domains and Chromosomal Regions ( telomere-proximal and central ) . Only categories with Fisher's exact test probabilities above with P>0 . 05 from the EASE analyses were reported for each gene set . Selective constraints were estimated for sets of orthologous genes from the Af293 , A1163 , A . clavatus , N . fischeri and A . terreus genomes . The rate of substitution in synonymous ( dS ) and in non-synonymous ( dN ) sites , and their ratio ( dN/dS ) was calculated using the PAML package [69] . If a gene is very well conserved , dN/dS<0 . 1; if a gene is under weak purifying selection , 0 . 1<dN/dS<1; if a gene is evolving neutrally ( e . g . pseudogenes ) , dN/dS∼ = 1; and if a gene is evolving under diversifying selection , dN/dS>1 . The results are reported only for orthologous genes sets having unsaturated dS values , the same number of exons , and sequence alignment coverage >95% . For each gene , the average dN/dS ratio for five pairwise species comparisons was calculated . We assembled a local database of protein sequences from the 28 publicly available fungal genome projects ( Table S17 ) . All phylogenetic analyses in this paper were carried out on protein sequences . The A . niger ATCC 1015 , Nectria haematococca , Phanerochaete chrysosporium and Trichoderma reesei genomes projects was completed under the auspices of the US Department of Energy's Office of Science , Biological and Environmental Research Program and the by the University of California , Lawrence Livermore National Laboratory ( Contract No . W-7405-Eng-48 ) , Lawrence Berkeley National Laboratory ( contract No . DE-AC03-76SF00098 ) and Los Alamos National Laboratory ( contract No . W-7405-ENG-36 ) . To produce a reference tree of species phylogeny we used the protein sequences of 90 likely orthologs from A . niger , A . nidulans , A . terreus , A . oryzae , A . clavatus , N . fischeri , A . fumigatus and Fusarium graminearum ( teleomorph of Gibberella zeae ) as an outgroup . To minimize the effect of incorrect or incongruent gene models , these proteins were chosen on the basis of having identical numbers of introns in each species and similar lengths . Sequences were aligned using MUSCLE [70] and columns of low conservation were removed manually . Maximum-likelihood trees were constructed using the PHYLIP package , applying the JTT substitution model with a gamma distribution ( alpha = 0 . 5 ) of rates over four categories of variable sites . Phylogenetic analyses of individual Af293 , A1163 , and N . fischeri proteins were carried out on sets of homologs identified in BLASTP searches against our fungal database . The top 20 hits with E<10−4 were retained for analysis . Sequences were aligned using ClustalW [71] . Poorly aligned regions were removed using Gblocks [72] . Finally , a maximum likelihood tree was drawn using PHYML [73] . To detect polymorphisms in the rosA ( AFUA_6G07010 ) gene , several hybridizations were performed using rosA gene as the probe and genomic DNA cleaved with EcoRI , ClaI , BamHI or EcoRV . For comparison , an invariable gene for all species ( apg5; AFUA_6G07040 ) was used as the hybridization probe on genomic DNA digested with HpaI . Colony radial growth rate measurements were performed as described [74] . For each isolate , four ( 90 mm diameter ) Petri dishes containing 25 ml agar medium were inoculated centrally with 2 . 5 µl of 1×106 spores/ml suspension in PBS/Tween 80 . Plates were then incubated at temperatures ranging from 25°C to 50°C and colony edges were marked using a plate microscope . Colonies were marked twice daily for 4–5 days . For each colony , two diameters perpendicular to each other were measured . Eight replicates were measured for each isolate . The results reported here are the mean of two experiments . At least five time points during the log phase were used to calculate growth rate . The radius of the colonies was plotted against time using least-square regression analysis , and the slope of the regression line , which represents the growth rate , was calculated . Each replicate was analysed separately and the mean of the growth rate was then calculated . | Aspergillus is an extremely diverse genus of filamentous ascomycetous fungi ( molds ) found ubiquitously in soil and decomposing vegetation . Being supreme opportunists , aspergilli have adapted to overcome various chemical , physical , and biological stresses found in heterogeneous environments . While most species in the genus are saprophytes , a surprising number are able to infect wounded plants and animals . Remarkably , the allergic human host also responds abnormally to the aspergilli with lung and sinus disease . The advent of immunosuppressive agents and other medical advances have created a large worldwide pool of human hosts susceptible to some Aspergillus species , including the world's most harmful mold and the causative agent of invasive aspergillosis , Aspergillus fumigatus . In this study , we have used the power of comparative genomics to gain insight into genetic mechanisms that may contribute to the metabolic versatility and pathogenicity of this important human pathogen . Comparison of the genomes of two A . fumigatus clinical isolates and two closely related , but rarely pathogenic species showed that their genomes contain several large isolate- and species-specific chromosomal islands . The metabolic capabilities encoded by these highly labile regions are likely to contribute to their rapid adaptation to heterogeneous environments such as soil or a living host . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/genomics",
"computational",
"biology/population",
"genetics",
"genetics",
"and",
"genomics/microbial",
"evolution",
"and",
"genomics",
"genetics",
"and",
"genomics/comparative",
"genomics",
"computational",
"biology/comparative",
"sequence",
"analys... | 2008 | Genomic Islands in the Pathogenic Filamentous Fungus Aspergillus fumigatus |
We have generated and made publicly available two very large networks of molecular interactions: 49 , 493 mouse-specific and 52 , 518 human-specific interactions . These networks were generated through automated analysis of 368 , 331 full-text research articles and 8 , 039 , 972 article abstracts from the PubMed database , using the GeneWays system . Our networks cover a wide spectrum of molecular interactions , such as bind , phosphorylate , glycosylate , and activate; 207 of these interaction types occur more than 1 , 000 times in our unfiltered , multi-species data set . Because mouse and human genes are linked through an orthological relationship , human and mouse networks are amenable to straightforward , joint computational analysis . Using our newly generated networks and known associations between mouse genes and cerebellar malformation phenotypes , we predicted a number of new associations between genes and five cerebellar phenotypes ( small cerebellum , absent cerebellum , cerebellar degeneration , abnormal foliation , and abnormal vermis ) . Using a battery of statistical tests , we showed that genes that are associated with cerebellar phenotypes tend to form compact network clusters . Further , we observed that cerebellar malformation phenotypes tend to be associated with highly connected genes . This tendency was stronger for developmental phenotypes and weaker for cerebellar degeneration .
A quarter of century ago a ( former ) Hewlett-Packard executive famously complained: “If only HP knew what HP knows” [1] . This inability to access invaluable “collective wisdom” is by no means specific to a single community . It is felt acutely in every present-day endeavor involving multi-human exploration of complex phenomena . The problem is especially dramatic in the case of the explosively expanding molecular biology literature . There are thousands of existing biological periodicals and millions of potentially useful publications . New journals are emerging on a weekly basis and new articles accumulate as if deposited by an avalanche . Understandably , no omniscient repository exists that lists all known ( published ) molecular events ( such as protein–protein interactions ) detected in human or murine cells . Although current text-mining tools are imperfect in their extraction accuracy and recall , they do help us to process huge amounts of unstructured text in nearly real time ( which humans cannot do ) , moving us a bit closer to total awareness about the current state of knowledge [2] . Here we describe and make available two large new data sets derived through mining one-third of a million full-text research articles and a complete and up-to-date PubMed collection of journal abstracts . These data sets comprise mouse- and human-specific molecular interactions between genes and/or their products . We present here only the subset of text-mined interaction assertions that involve gene or protein names that we can link to unique identifiers in the standard sequence databases . This choice is determined by the goal of making our data immediately useful for applications that would have difficulty handling ambiguity in gene identity . The complete data are available through the Columbia University ( http://wiki . c2b2 . columbia . edu/workbench ) and the University of Chicago ( http://anya . igsb . anl . gov/genewaysApp ) . We use our newly generated data to analyze genetic variation related to abnormal cerebellum phenotypes in mouse and human . Our analysis results in a compact set of statistically significant predictions that can be tested experimentally .
Text mining with the GeneWays system [3] , [4] allows us to capture multiple classes of relationships among biological entities , such as “A phosphorylates B , ” “C activates D , ” and “E is a part of F . ” Table S1 displays the full list of relations that we can extract currently . The system also can recognize multiple classes of biological entities ( terms ) mentioned in the text: genes , proteins , mRNAs , small molecules , processes ( such as cell death and proliferation ) , tissues , cell types , and phenotypes ( such as diabetes and hypertension ) . While one can immediately think of a wide spectrum of applications where the full diversity of entities must be used , most of the current experimental methods are either gene-centric or genetic loci-centric ( e . g . , gene expression arrays , ChIP-on-chip , yeast two-hybrid , and genetic linkage or association data ) . For this reason , the molecular networks we present here are gene-centric . This means that a given node in the network represents the union of the gene and its products ( mRNA ( s ) and protein ( s ) , if any ) ; we exclude all other types of nodes ( such as small molecules and phenotypes ) . Our practice of collapsing multiple nodes to a single node ( gene plus mRNA plus protein ) does not lead to a loss of information , because most of the physical interactions are defined for specific types of molecules . For example , in our restricted network relationship , “phosphorylate” can link only a pair of proteins , one acting as a protein kinase and another as the kinase's substrate , but not a gene and an mRNA . Furthermore , each original sentence used to extract the relation is preserved in the data set , along with the extracted fact and the reference to the appropriate paper , so that additional disambiguation can be conducted later , if required . We refer to each pair of extracted relationships and the original snippet of text as an action mention , as opposed to action , which is a relation disconnected from the source text and potentially mapped to multiple distinct action mentions . A single pair of nodes in our text-mined network can be connected with multiple edges . These edges ( interactions ) can be undirected ( we treat “A binds B” and “B binds A” as identical ) or directed ( “C activates D” is not the same as “D activates C” ) . We also subdivide edge types into two groups: logical and physical . Logical interactions include a family of regulatory relations that can be either direct ( physical contact between two molecules ) or indirect ( mediated by one or more other molecules ) , such as activate , inhibit , and regulate ( see Table S1 ) . Physical interactions are by definition direct , such as methylate , bind , glycosylate , and cleave ( see Table S1 ) . The distinction between physical and logical interactions is important in understanding the data sets that we describe here . GeneWays ontology [5] includes a number of relationships between molecules that are neither physical nor logical interactions ( for example , A is an ortholog of B , or C is part of D ) . We call this class of relations other . In typical free text , gene names are dissociated from any references to gene-annotation databases . Furthermore , the “raw” text-mined molecular-interaction data are vast ( GeneWays 7 . 0 comprises more than 8 million action mentions ) but rather noisy: the error rate is close to 35% [6] . To get to smaller , cleaner , species-specific networks , we performed the following steps . First , out of the complete network we retained only those gene names that can be linked to either human or mouse sequence database entries ( normalization step ) ( see “Mapping names to genes” in the Text S1 ) . Second , we filtered out relationships that are not molecular interactions and collapsed multiple edges between two nodes into a single edge . Third , we weeded out “raw” text-mined statements that did not meet our precision threshold ( precision is defined as the proportion of correctly extracted statements among all those automatically extracted by a system ) . The third step was conducted automatically , using our automated curator engine [6] , which has near-human curation precision ( see the “Quality-of-extraction assessment” section in the Text S1 . ) . The first step resulted in the H70 and M70 networks ( human- and mouse-specific GeneWays 7 . 0 ) , in which nodes can be connected by multiple directed or undirected edges . The second step led to generation of the H70-PL and M70-PL networks ( PL stands for physical and logical ) , where direction of edges was abandoned . The third step , assigned a precision threshold of 0 . 9 ( 90% of action mentions are correct ) , produced even smaller data sets , H70-PL0 . 9 and M70-PL0 . 9 . Table 1 provides an overview of these networks at different levels of granularity . All intermediate data sets in this pipeline of data filtering are available for third-party computational analyses ( see Datasets S2 to S5 ) In addition , we produced networks with non-redundant edges and solely physical interactions , H70-P0 . 9 and M70-P0 . 9 . As in the previous data sets , to filter these networks we used a precision threshold of 0 . 9 . To evaluate the quality of the H70-PL0 . 9 network , we chose two random sets of logical and physical action mentions , a hundred mentions each , and asked an expert to evaluate their correctness . The expert commented on two steps of the process: whether the action mention is correctly extracted by the GeneWays system and , if the answer was “yes , ” whether the corresponding gene names were correctly mapped to sequence identifiers . This allowed us to measure the absolute precision of the H70-PL0 . 9 network , the precision of term mapping , and the overall precision over the information extraction and term mapping stages . The physical action mentions set indicated a precision of 0 . 8 , with a confidence interval ( CI ) of [0 . 71 , 0 . 87] . ( We use CI at the 95% level of significance consistently throughout this paper . ) The logical action mentions set showed a higher precision of 0 . 91 , CI: [0 . 84 , 0 . 95] . Because in our data set the number of logical interactions exceeds the number of physical interactions by more than two-to-one ( 2 . 49∶1 ) , the overall precision of the HL70-PL0 . 9 data set is close to the target value of 0 . 9 ( 0 . 88 ) . Term-to-sequence mapping precision was 0 . 89 ( CI: [0 . 84 , 0 . 93] ) and 0 . 87 ( CI: [0 . 81 , 0 . 91] ) for physical and logical action mentions , respectively ( see Table 2 ) . Despite the favorable precision of the GeneWays extraction and the per-term mapping , the precision over both steps is less impressive: 0 . 66 ( CI: [0 . 56 , 0 . 74] ) and 0 . 69 ( CI: [0 . 59 , 0 . 77] ) for the physical and logical datasets , respectively . The reason for the lower overall result is the multiplicative calculus of the probability of not making an error: The overall precision of a term-mapped logical action is a product of the information extraction precision and the precision of two independent term mappings: 0 . 91×0 . 87×0 . 87 = 0 . 69 . Thus far we have evaluated the quality of extraction and mapping of action mentions . Recall that the same relation ( action ) between a pair of genes can be independently extracted from multiple sentences , generating distinct action mentions . Intuitively , the precision of an action ( because an action is correctly extracted if at least one of its associated action mentions is correctly extracted ) should be at least as high ( or higher ) than precision of the corresponding action mentions . To evaluate this precision , we sampled a hundred random actions from the H70-PL0 . 9 dataset , asked an expert to evaluate them at the levels of extraction and term mapping , and obtained an estimate of action-level two-stage precision of 0 . 74 , CI: [0 . 65 , 0 . 82] . This estimate is higher than the estimate of two-stage action mention precision ( 0 . 66 or 0 . 69 ) . We believe that the action-level precision is more relevant to real-life applications in which scientists tend to care primarily about the precision of actions ( statements distilled from multiple sources ) rather than about their individual instances linked to text . Note that the precision discussed in this section reflects only properties of our information extraction system and not the verity of published data . Several publicly accessible databases generated by manual analysis of research literature are available , including the Human Protein Reference Database ( HPRD ) [7] , [8] , Reactome [9] , the Biomolecular Interaction Network Database [10] , and the Database of Interacting Proteins [11] . These four data sets , along with a few others , were carefully compared in a recent study [12] . HPRD is by far the largest of the four . As another quality control measure for our study , we compared our data with HPRD 7 . The HPRD 7 network [7] , [8] comprises 9 , 460 nodes ( unique gene identifiers ) and 37 , 081 edges , compared to 7 , 793 nodes and 52 , 518 edges in the H70-PL0 . 9 network . The H70-P0 . 9 network comprises 5 , 453 nodes and 16 , 707 edges; the node-wise and the edge-wise overlaps of H70-P0 . 9 with the HPRD networks are 4 , 543 and 4 , 877 , respectively . The HPRD and H70-PL0 . 9 networks share 5 , 945 unique gene-specific nodes . Out of the possible maximum of 17 , 668 , 540 interaction pairs between these nodes , the HPRD network has 23 , 662 and the H70-PL0 . 9 covers 43 , 496 . We would expect a random overlap of about 58 interactions , while in reality we observe 7 , 577 . The expected and the observed values are so far apart that the p-value ( obtained with a hyper-geometric overlap test ) is effectively zero—that is , the apparent overlap between the two sets of data is extremely non-random . Because human-curated databases may still harbor errors [13] , we also compared our literature-mined dataset to a small set of high quality interactions produced by careful manual verification of a set of interactions shared by several human-curated databases [13] . In a recent study , Cusick et al . sought to evaluate the ultimate ( truth ) quality of the molecular interaction datasets generated via manual curation of the literature [13] . The authors selected two sets of curated interactions: one consisted of interactions that were curated in multiple databases and that were supported by multiple manuscripts and the other consisted of interactions supported by a single publication . They then carefully recurated the selected interactions and were able to estimate the corresponding error rates . As a byproduct of the evaluation , the authors produced two relatively small datasets , LC-multiple and LC-single , with 110 and 92 interacting pairs respectively , of exceptionally high-quality curated interactions . The LC-multiple set contained the interactions that were supported by multiple manuscripts even after the recuration and the LC-single set contained the interactions with one supporting manuscript that was confirmed during the recuration . The LC-multiple set subsequently was used as a gold standard for the evaluation of high-throughput yeast two-hybrid assays in a second manuscript by the same group , in which an additional random set of 188 supposedly non-interacting pairs ( the Negative set ) was selected [14] . We used the LC-multiple , LC-single , and Negative sets as comparison standards for our own literature-mined networks ( see Table 3 ) . It is reassuring that our H70-PL network covers nearly 70% ( 75 out of 110 pairs ) and that our most filtered human network , H70-P0 . 9 , covers more than 55% of the well-supported interactions in the LC-multiple set . The more obscure interactions from the LC-single set are not covered as well ( i . e . , H70-P0 . 9 contains about 20% of the LC-single set ) . However , given that we have processed only a small portion of all of the scientific literature with a system that highly favors precision over recall , being able to recover 20% of the interactions supported by a single article is surprisingly high . Finally , our networks do not contain any of the interactions listed in the Negative set . For comparison , the last two lines in Table 3 give the results for the two high-throughput assays MAPPIT and Y2H-CCSB evaluated in Figure 2 of [14] . The performance of text-mining methods is commonly evaluated using two metrics: precision and recall . For information-extraction systems , precision is defined as the proportion of correctly extracted statements among all those automatically extracted by a system . The recall is the ratio between the number of statements correctly extracted by the system and the total number of statements that can be extracted from the original text by a hypothetically perfect system . In a less than perfect system , recall and precision are antagonistic: one is increased at the expense of the other . In this study we favored precision at the expense of recall: We explicitly used a statement precision threshold as a filtering criterion . We also excluded actions with ambiguous gene names and disqualified some 105 potentially useful instances of text-mined intramolecular relations that fit neither physical nor logical categories ( such as contain and is a homolog of ) , thus worsening recall and improving precision . In addition , we used only those actions that involve either genes or their products ( and no other entity classes ) . While our human-specific network , which unifies physical and logical interactions ( H70-PL0 . 9 ) , is larger than HPRD 7 , the relationship is reversed for the physical-interaction ( H70-P0 . 9 ) data set and HPRD 7 . This is because we filtered out from our data numerous physical action mentions that did not pass our precision threshold . ( Note that HPRD 7 incorporates high-throughput interaction data that is probably distinct from the small-scale experimental data published in research papers , in terms of error patterns . ) Nevertheless , the HPRD 7 data sets and our data sets are very different . The joint interaction coverage of HPRD 7 , H70-P0 . 9 , and M70-P0 . 9 ortholog data sets is more than twice as large as the coverage of HPRD 7 alone ( Figure 1 ) ; this is enough to merit the use of a union of these networks in biological applications . Because we are making the “raw” ( unfiltered ) statements publicly available , anyone interested in using our data can apply his/her own custom-made filtering process to achieve the desired balance between recall and precision in the output . Two genes residing in genomes of distinct species can either share a common origin ( homology ) or be unrelated . Homologs come in at least in two flavors [15]: orthologs and paralogs . Two genes in , say , human and mouse , are orthologs if they were separated by a speciation event . If , in addition to speciation , an intragenomic gene duplication occurred , separating two genes from a common ancestral gene , they are paralogs . For example , human and mouse embryonic β-globins are orthologs , but mouse α-globin is a paralog of human β-globin . Physical interactions between molecules are not immutable over long evolutionary intervals [16] . Nevertheless , an interaction between two proteins discovered in one species has a reasonable chance of existing between orthologs of these proteins in another species if the two species are closely related . Therefore , if we know of interacting molecules in one species and can identify orthologous molecules in another species , we can formulate hypotheses about the existence of orthologous interactions in the latter species . All such computationally formed hypotheses are subject to experimental validation . Mouse and human genomes are separated by more than 100 million years of independent evolution [17] , but mouse genetics and molecular biology are commonly used to understand human phenotypes in health and disease . Therefore , we decided to compile a molecular-interaction network summarizing the wealth of knowledge for humans and mice . We used orthology-mapping of human and mouse genes to connect the two networks . ( Reactome's developers [9] used a similar strategy with their manual compilation of data . ) Such a network could potentially have a multitude of practical applications . We assembled our network by combining mouse- and human-specific networks extracted from the biomedical literature using text-mining tools . We used human-to-mouse gene orthology mapping provided by the Mouse Genome Database [18] , [19] . Some of the mouse interactions could not be mapped to corresponding human interactions because at least one of the involved genes lacked known human orthologs . We transferred by mouse-to-human orthology-mapping 49 , 493 and 16 , 317 interactions for physical-logical and physical networks , respectively . These orthology-mapped interactions are subsets of the 57 , 786 and 18 , 252 interactions in the physical-logical and physical networks , respectively . Although a large number of interactions occur both in humans and mice individually ( see Figure 1 ) , the double-confirmed overlap constitutes only about a third of all interactions in the union network ( see Figure 1 A and B ) . Figure 1 C shows a three-way Venn diagram for our text-mined interactomes ( human and mouse ) and the HPRD dataset . Clearly , all three networks contain a substantial number of unique interactions that merit their joint consideration in biological applications ( see Dataset S6 ) . To illustrate an application of our data to the analysis of mammalian phenotypes , we performed mapping of mouse cerebellar phenotypes ( related to ataxia ) to the three-data set network . The word ataxia ( αταξια—“lack of order” ) , in its English usage , refers to a lack of muscular coordination in an animal body . Humans with ataxia often have difficulty walking , talking , maintaining posture and balance , controlling eye movements , holding and manipulating objects , gesturing , and even swallowing food . In a mammalian brain , the cerebellum is predominantly responsible for spatial and temporal coordination of complex neuromuscular processes . Cerebellar function is also essential for cognition sensory discrimination [20] . Most cases of ataxia are associated with either environmental or genetic damage to this brain region . The typical environmental triggers of ataxia include head trauma , viral infections , and exposure to recreational or medicinal poisons , such as alcohol , lithium carbonate , tranquilizers , antipsychotics , and the anticonvulsant carbamazepine . Genetic factors include a diverse spectrum of genomic aberrations that cause abnormal development and/or premature degeneration of the cerebellum . Ataxia can be severely debilitating and , unfortunately , the phenotype is reversible in only a minority of cases ( such as those caused by short-term alcohol intake ) . Mouse and human geneticists who study brain phenotypes typically group developmental malformations by the anatomical structures that are affected . As brain topology in three-dimensional space does not lend itself readily to verbal description , we provide three projections of a typical mouse brain in Figure 2 ( see also the interactive model in Figure S1 ) . Moving front-to-back in the external view of the mouse brain , there are two olfactory bulbs followed by hemispheres of cerebral cortex that are immediately adjacent to the cerebellum and brainstem ( see Figure 2 A–C ) . We focus here on the cerebellum ( literally , “little brain” ) that contains involuted cortex with narrow leaf-like gyri ( “folia , ” see Figures 3 A and C ) . Like the brain itself , the cerebellum has two hemispheres with a worm-like medial structure , the vermis , between them ( Figure 3 A and B ) . In both humans and mice , a collection of genetic aberrations exist that are known to predispose the bearer to specific cerebellar abnormalities . For computational implementation it is convenient to represent phenotypic variations of cerebellar structure with hierarchically ordered categories in a mammalian phenotype ( MP ) ontology [21] . We focused on five broad anatomical/cerebellar causes of ataxia which can be observed as structural abnormalities in brain imaging studies ( such as MRI scans ) or histological analysis . These phenotypes are represented with MP concepts: degeneration ( MP:0000876 ) , abnormal foliation ( MP:0000857 and MP:0000853 ) , abnormal vermis ( MP:0000864 ) , small cerebellum/cerebellar hypoplasia ( MP:0000852 and MP:0000851 ) , and absent cerebellum ( MP:0000850 ) . Cerebellar degeneration is abnormal death of cerebellar neurons—the cerebellar folia become narrower over time and are separated by irregular and wider spaces compared with those in a healthy brain ( see Figure 3 A ) . As with other major insults to the cerebellum , degeneration reveals itself in abnormalities in body balance , jerky movements of limbs , unsteady ( wide-legged ) gait , and irregularities of speech ( slurred or slow ) and eye movement ( nystagmus , or rapid involuntary eye movements ) . Most defined degenerative ataxias affect the fully mature cerebellum , but a small subset of degenerative ataxias have a developmental onset [22] . Abnormal foliation typically involves the absence of some of the cerebellar folia and irregular shape of those that are present . In normal individuals , cerebellar foliation is stereotypical , with the basic folial pattern conserved between mice and humans . Disruption of foliation disrupts the topographical map of incoming and outgoing neuronal connections [23] . An abnormal vermis is typically reduced ( compressed and distorted ) compared with a normal one , or it can even be completely missing ( see Figure 3 A ) . Clinical outcome is variable [24] . Dandy-Walker malformation is one of the well-known birth defects in humans and mice that are defined by an abnormal vermis . In addition to an aberrant vermis , Dandy-Walker malformation frequently involves enlargement of the fourth brain ventricle and an increase in fluid-filled space around the brain [25] . It is not uncommon in clinical reports to find an abnormal vermis coupled with other cerebellar malformations [26] , [27] . Small cerebellum , or cerebellar hypoplasia , refers to phenotypes in which the cerebellum , while present , does not develop to normal size ( see Figure 3 A ) . In humans , cerebellar hypoplasia can lead to delayed or undeveloped speech , difficulties with walking and maintaining balance , mental retardation , floppy muscle tone , nystagmus , and seizures . In its worst forms , cerebellar hypoplasia can be completely debilitating and even lethal [28] . Absent cerebellum is infrequent in adult humans and mice , possibly because in most cases it causes early death . Rarely , individuals are only mildly affected . For example , a documented brain autopsy of a 38-year-old individual who accidentally electrocuted himself revealed a virtually absent cerebellum [29] . The individual was apparently functional and capable of conducting all common human activities , including gesturing , talking , performing complex manual work , and participating in social interactions . Some have proposed that an absent cerebellum is therefore less disabling than a present , but abnormal cerebellum [30] . Fortunately , the Mouse Genome Database ( MGD , [18] ) uses the MP ontology to link genetic variation in mouse genes to phenotypic aberrations that are causally related to known genomic changes . We were able to use the MGD to associate 244 human genes ( with the help of the human and mouse orthology ) with the five ataxia phenotypes described above and with ataxia ( MP:0001393 ) ( see Table S7 ) . By integrating mouse ( M70-PL0 . 9 ) , human ( H70-PL0 . 9 ) , and HPRD ( Release 7 ) networks through human–mouse gene orthology mapping , we obtained a larger network of interacting human genes with annotation of ataxia phenotypes generated in mouse studies . The largest connected component of the ataxia graph includes 88 human genes . These 88 genes are connected with 145 , 147 , and 72 interactions derived from human GeneWays , mouse GeneWays , and HPRD , respectively ( see Figure 3 D ) . Our analysis of ataxia-related phenotypes in the context of a molecular network generated rather curious and statistically significant results , as described in the following section . We have provided two very large molecular-interaction sets for human and mouse genes ( Datasets S2 , S3 , S4 , S5 ) . The sets were integrated through gene orthology and are immediately applicable to a spectrum of experimental data analysis tasks ( Dataset S6 ) . Our analysis of mouse mutant cerebellar phenotypes , with the aid of our text-mined networks , lead to a number of intuitively reasonable and biologically testable predictions . Our present study shares its spirit , goals , and some methods with other efforts in the field . For example , one of the most recent studies succeeded in integrating a diverse array of approaches to design a tool generating new disease-related hypotheses [57] . This group was able to combine information extraction [58] , biomedical ontology mining , statistical analysis of sequences of natural language tokens , probabilistic analysis of error patterns across data types , computational reasoning , understanding of large-scale experimental datasets , and exploratory visualization in one application . Because we are releasing our complete set of annotated data to public domain , these data might be instrumental for direct comparison for analogous text-mining results produced elsewhere [59]–[67] . Automated reasoning over text-mined , experimental , and machine-deduced data ( Reading , Reasoning , and Reporting , as [57] put it ) , is likely to become a dominant mode of science in the near future , as size of experimental datasets and complexity of natural system under scrutiny grows .
GeneWays is an information extraction ( text-mining ) system: It ingests computer-encoded full-text research articles or journal abstracts and distills from them a collection of biological relations . The architecture and implementation of the system are described in great detail elsewhere [3] , [4] , [6] , [68]–[71] , so here we provide just a brief outline of the system . In a simplified view , the processing pipeline inside the GeneWays system is a sequence of just two steps . The first step deals with recognition of words or phrases representing important biological entities ( such as p53 , Alzheimer's disease , or endoplasmic reticulum; computer scientists call this step named entity recognition , NER ) . The second step deals with detecting relations among the entities ( such as p53 activates JAK ) and is called information extraction ( IE ) . Our NER module ( MarkIt , [72] ) works by following a hierarchy of rules defined by human experts . Our IE module ( GENIES , [3] , [68] ) also is based on a collection of expert-encoded rules , but the underlying mathematical model is a bit more sophisticated ( a deterministic context-free grammar ) . We use MarkIt to identify a spectrum of biological entities , such as disease , process , gene , protein , RNA , small molecule , tissue , and species . We apply GENIES to each individual sentence , trying to reconstruct the most probable steps that led to generation of the sentence . This reconstruction process is called parsing; besides satisfying our academic curiosity , parsing is useful for capturing complex relationships between entities . The results of parsing depend strongly on the formal grammar implemented in the parser . Most of the relations that we can extract from biomedical texts are directional ( A activates B is not the same as B activates A ) and binary ( only two entities are involved , which we call upstream and downstream , according to their position within the relation ) . A sentence can contain any number between none and dozens of relations . We can think of a typical binary relation as a quadruplet of values: [negation , upstream entity , action , downstream entity] ( see Figure 5 ) . Negation allows us to capture negative statements ( “In our experiment , AKT failed to phosphorylate BAD”→[not , AKT , phosphorylate , BAD] ) as well as positive statements . Relation type ( action ) indicates semantic connection between the two entities ( bind , activate , methylate , transport , is part of , is homolog of , etc . ) . To facilitate automatic reasoning over semantic groups of relations , we arrange them into an acyclic directed graph , where graph arcs represent the relation “is a . ” The GeneWays system currently recognizes 391 different action types , 207 of which are frequent ( see Table S1 ) . To generate the molecular networks described in this study , we analyzed 368 , 331 full-text articles and 8 , 039 , 972 article abstracts from PubMed ( see Table S8 ) . The system identified 5 , 890 , 150 relations in the full text articles and 2 , 534 , 299 relations in the abstracts: in total , 5 , 934 , 024 unique relationships . The action types with the largest number of relationships are induce ( 695 , 615 ) , bind ( 532 , 385 ) , inhibit ( 386 , 523 ) , associate ( 370 , 133 ) , contain ( 366 , 654 ) , and activate ( 332 , 336 ) ; the numbers in parentheses indicate the abundance of relations of each type in the GeneWays 7 . 0 database . | We described and made publicly available the largest existing set of text-mined statements; we also presented its application to an important biological problem . We have extracted and purified two large molecular networks , one for humans and one for mouse . We characterized the data sets , described the methods we used to generate them , and presented a novel biological application of the networks to study the etiology of five cerebellum phenotypes . We demonstrated quantitatively that the development-related malformations differ in their system-level properties from degeneration-related genes . We showed that there is a high degree of overlap among the genes implicated in the developmental malformations , that these genes have a strong tendency to be highly connected within the molecular network , and that they also tend to be clustered together , forming a compact molecular network neighborhood . In contrast , the genes involved in malformations due to degeneration do not have a high degree of connectivity , are not strongly clustered in the network , and do not overlap significantly with the development related genes . In addition , taking into account the above-mentioned system-level properties and the gene-specific network interactions , we made highly confident predictions about novel genes that are likely also involved in the etiology of the analyzed phenotypes . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Methods"
] | [
"computational",
"biology/literature",
"analysis",
"neuroscience/neurodevelopment",
"computational",
"biology/systems",
"biology"
] | 2009 | Looking at Cerebellar Malformations through Text-Mined Interactomes of Mice and Humans |
Heterosis describes the phenotypic superiority of hybrids over their parents in traits related to agronomic performance and fitness . Understanding and predicting nonadditive inheritance such as heterosis is crucial for evolutionary biology as well as for plant and animal breeding . However , the physiological bases of heterosis remain debated . Moreover , empirical data in various species have shown that diverse genetic and molecular mechanisms are likely to explain heterosis , making it difficult to predict its emergence and amplitude from parental genotypes alone . In this study , we examined a model of physiological dominance initially proposed by Sewall Wright to explain the nonadditive inheritance of traits like metabolic fluxes at the cellular level . We evaluated Wright’s model for two fitness-related traits at the whole-plant level , growth rate and fruit number , using 450 hybrids derived from crosses among natural accessions of A . thaliana . We found that allometric relationships between traits constrain phenotypic variation in a nonlinear and similar manner in hybrids and accessions . These allometric relationships behave predictably , explaining up to 75% of heterosis amplitude , while genetic distance among parents at best explains 7% . Thus , our findings are consistent with Wright’s model of physiological dominance and suggest that the emergence of heterosis on plant performance is an intrinsic property of nonlinear relationships between traits . Furthermore , our study highlights the potential of a geometric approach of phenotypic relationships for predicting heterosis of major components of crop productivity and yield .
If the inheritance of phenotypic traits was additive , progenies would always exhibit intermediate trait values compared to their respective parents . Genetic nonadditivity is , however , a common result of intraspecific crosses . It has been exploited for decades in agronomy [1–5] , although the underlying mechanisms remain a major question for evolutionary genetics and crop science [6 , 7] . In plants , superior vigour in hybrids compared to the parents , or heterosis , is frequent [8–12] , and its molecular bases have been investigated in numerous studies [13–17] . Empirical observations have often led to contrasting conclusions , and none of the proposed genetic mechanisms can fully explain the emergence of heterosis of different traits across all study systems [13 , 18–23] . Thus , we are still lacking a unifying theoretical corpus that enables us to explain and predict heterosis of fitness-related traits , including biomass , growth rate , and fecundity . Several genetic hypotheses have been proposed to explain heterosis [7 , 14 , 17 , 24] . According to the ‘dominance hypothesis’ , each parent contributes favourable , dominant alleles ( generally at many loci ) that together complement the deleterious effects of recessive alleles originating from the other parent . The ‘overdominance hypothesis’ postulates the existence of loci in which the heterozygous state ( Aa ) contributes to the superiority of the hybrid compared to both homozygotes ( AA or aa ) . Pseudo-overdominance corresponds to cases in which dominant , favourable alleles are in linkage with recessive , unfavourable alleles so that the heterozygous combinations appear to behave as overdominant loci . These different mechanisms are not mutually exclusive , and they can all operate at the same time in a given species , which can explain some of the contradictory results observed in different studies . The picture is further complicated by the contributions of epistasis [19 , 25] and epigenetics [16 , 26] to heterosis in plants . To tease apart the different hypotheses , quantitative trait locus ( QTL ) mapping has been carried out in many species [9 , 13 , 19 , 21 , 25 , 27] . In general , individual studies differ strongly in their conclusions regarding the underlying genetic mechanisms , apparently depending on the investigated traits , the genetic material used , the mating system , and the experimental constraints related to the number of crosses necessary for robust statistical inference ( e . g . , diallel mating design ) [17 , 23 , 24] . In addition , the QTL approach has inherent limitations when it comes to the detection of small-effect loci [28] . It has been shown that fitness-related traits such as growth rate , size , and fruit production are often controlled by a large number of genes , which individually may have very weak effects [29–31] , reducing the usefulness of the QTL approach . Because of the polygenic nature of fitness-related traits , heterosis is expected to be associated with molecular dominance and should positively correlate with genetic distance between parents , at least up to a certain extent [32–35] . Some findings are consistent with this hypothesis [36–38] , suggesting that parental genetic distance could be used to quantitatively predict heterosis in plants . However , experimental studies generally employ for practical reasons only a relatively small number of crosses and parental lines [11 , 35] , which makes it difficult to generalize the findings of individual studies . In a recent work , Seymour and colleagues [17] investigated heterosis in a half-diallel cross between 30 genome-sequenced accessions of A . thaliana collected from diverse Eurasian populations [39] . As expected under the dominance model [32–34] , they found a positive correlation between parental genetic distance and heterosis . However , genetic distance between parents only accounted for less than 3% of heterosis among A . thaliana hybrids , making predictions based on genetic distances alone strongly uncertain . Despite or perhaps because of all of these previous efforts , many continue to consider the physiological bases of heterosis a mystery [40 , 41] . An alternative to genetics-first studies to understand and predict heterosis is to consider the physiological constraints that determine phenotypic variation across organisms . As early as 1934 , Sewall Wright proposed a model of physiological dominance to explain the maintenance of recessive alleles in natural populations [42] . Wright began with the universal relationship that connects the concentration of enzymes to the metabolic flux that results from their activity ( Fig 1 ) . Since the relationship between these two traits is concave with a horizontal asymptote ( e . g . , Michaelis–Menten kinetics [43] ) , dominance of metabolic flux is expected even if enzyme concentrations are additive and hybrids intermediate between their parents ( Fig 1 ) . Modern knowledge of metabolic networks and fluxes was later incorporated into this model by Kacser and Burns in 1981 [44] . However , Wright’s model of dominance has otherwise received little attention in the heterosis literature , presumably because in the 1930s , physiology was synonymous to metabolism , whereas this is not the case in modern scientific language [45] . Recently , Fiévet and colleagues [46 , 47] evaluated Wright’s model with a focus on the relationship between metabolic flux and enzyme activity in the chain of glycolysis in yeast . They modelled the strength of the curvature of the relationship between enzyme concentration and metabolic flux and validated its role in the emergence of heterosis on glycolysis [46] . This approach has considerable promise for predicting the phenotype of hybrids by considering the nonlinear relationships that often link phenotypic traits with each other . However , these theoretical expectations remain to be tested using more complex traits as well as on multicellular organisms , such as plants , with potentially major perspectives for cultivated species . Many phenotypic relationships exhibit nonlinear geometries at different organizational levels . At the cellular level , examples include the relationships between RNA transcripts and protein level [48–50] or the relationship between mitochondrial respiration and cell growth [51] . At the organism level , macroecology studies have demonstrated the existence of nonlinear allometric relationships between the biomass of an organism and its morphology , physiology , and metabolism [52–55] or between reproductive and fitness-related traits [31 , 53 , 56] . The metabolic scaling theory ( MST ) [55] posits that the geometry of the resource distribution network , like vein branching in plants [57 , 58] , constrains allometric relationships to be predictably nonlinear across taxonomic scales [54 , 59] ( Box 1 , Fig 2 ) . Consistent with this idea , recent findings demonstrated that the allometric relationship of growth rate is not only similar across species but also across natural accessions of A . thaliana as well as recombinant inbred lines ( RILs ) of this species artificially created in laboratory [31 , 60] . These relationships are analogous to those between metabolic and enzymatic activities described by Wright [42] , Kacser and Burns [44] , and Fiévet and colleagues [46 , 47] . For instance , if an additive trait x is linked to a trait y by an asymptotic relationship , like enzyme concentration is to metabolic flux in Fig 1 , hybrids should exhibit mid-parent heterosis ( but not best-parent heterosis ) for trait y . However , Fig 1 represents a simple and unrealistic case in which the metabolic flux depends on a single enzyme . The recent study performed by Fiévet and colleagues [46] demonstrated that in a more realistic situation , the multidimensional relationship between several enzymes and metabolic flux is concave , hump shaped . In such a case , hybrids can exhibit best-parent heterosis , as has been shown for the chain of glycolysis [46] . Inspired by Wright’s model of physiological dominance for enzyme concentration and metabolic flux [42] , we have tested whether allometric relationships can explain the emergence of heterosis in macroscopic traits related to performance , such as growth rate and fruit production . We used the plant A . thaliana , which has been used for many genetic studies of heterosis [11 , 17 , 70–73] . This species is characterized by a high rate of inbreeding , which leads to a high rate of homozygosity in natural accessions , considered as inbreds [74 , 75] . The complete sequencing of 1 , 135 natural accessions [76] has provided abundant information on the extent of genetic variation between populations , the genotype–phenotype map of the species , as well as the causes of phenotypic variation in intraspecific hybrids [77–81] . Several studies reported strong diversification of primary and secondary metabolism between accessions [82–86] , which points to large genetic variation in enzyme properties between accessions , a key observation for Wright’s model of physiological dominance . Furthermore , allometric relationships that link growth rate and fruit production to plant biomass have been evaluated in this species [31 , 60 , 87] . Here , we specifically addressed the following questions: ( i ) Are there nonlinear allometric relationships that similarly constrain trait variation in natural accessions and hybrids ? And ( ii ) do these nonlinear relationships explain the emergence and extent of heterosis in growth rate and fruit production ? To this end , we compared several traits between 451 accessions , representing a wide range of native environments , and 450 crosses derived from crosses among 415 of these accessions . Our results are consistent with many aspects of Wright’s model , suggesting that heterosis emerges intrinsically from nonlinear relationships between traits .
We generated 450 hybrids by manual crosses between 415 accessions ( Fig 3A ) . We then measured four traits , vegetative dry mass , plant age at reproduction , growth rate , and total fruit number in the 450 hybrids , the 415 inbred parents , plus a further 35 accessions . The parental combinations for the hybrids had been chosen to represent a wide range of genetic , geographic , and phenotypic distances ( Fig 3B and 3C ) . Traits were strongly variable between genotypes: vegetative dry mass M varied between 1 . 3 and 2 , 218 mg , plant age at reproduction between 24 and 183 days , growth rate between 0 . 04 and 40 . 4 mg/d−1 , and fruit number between 5 and 400 ( S1 Data ) . Across the entire data set of hybrids and accessions , a high proportion of the phenotypic variance was explained by genotypic differences ( broad-sense heritability ) : from 80% for growth rate to 94% for vegetative dry mass ( S1 Table ) . We found similar ranges of variation for both accessions and hybrids ( Fig 4 ) . For instance , standard deviation of plant age at reproduction was 19 . 8 days across accessions and 20 . 1 days across hybrids ( S1 Table ) . Thus , hybrids represent a similar phenotypic space as natural accessions , and they were on average not significantly different from accessions ( P > 0 . 01 for all traits , S1 Table ) . Despite the similar distribution of trait values for accessions and hybrids , we found significant heterosis for all traits . The emergence of heterosis in our data did not simply reflect a few exceptional hybrids that exhibited trait values outside of the range observed across accessions but rather hybrids with specific combinations of traits within a similar range of trait values . The proportion of hybrids exhibiting significant differences compared to their parents was measured by the discrete categorization of heterosis , based on the comparison of hybrid trait distribution with both mid-parent values and highest ( ‘best’ ) or lowest ( ‘worst’ ) parent values . For all traits , there was both positive and negative heterosis ( Fig 4 ) . However , the extent and direction of heterosis was variable between traits . For instance , 28% of hybrids were ( positively or negatively ) heterotic for age at reproduction . By contrast , 62% of hybrids were heterotic for vegetative dry mass . In general , there was more negative than positive heterosis , specifically for vegetative dry mass ( 49% of hybrids ) . We focused on two traits with important evolutionary and agronomic outcomes: growth rate and fruit number . As predicted by the MST [55 , 58] , these two traits exhibited nonlinear allometric relationships with vegetative dry mass ( Fig 5 ) . Moreover , in both accessions and hybrids , the allometric relationship of growth rate had an exponent close to ¾ ( the slope of the log10-transformed relationship was 0 . 74 for accessions and 0 . 78 for hybrids , S1 Fig ) . Nonetheless , standard major axis ( SMA ) regressions revealed that the slope was significantly different between accessions and hybrids ( P < 0 . 01 ) . The slope was not significantly different from ¾ for the accessions ( P > 0 . 05 ) but differed significantly from ¾ for the hybrids ( P < 0 . 01 ) . However , and consistent with recent studies of plant allometry [31 , 60 , 88] , growth rate was more accurately modelled by a power-law function with mass-dependent exponent ( g[M] , Table 1 ) rather than a fixed exponent ( Akaike information criterion [ΔAIC] = 53 for the accessions and ΔAIC = 55 for the hybrids ) . This generated a nonlinear relationship with a concave curvature ( Fig 5A ) , whose coefficients were not significantly different between accessions and hybrids ( Table 1 ) . Fruit number exhibited a right-skewed hump-shaped relationship ( Fig 5B ) . We modelled this allometric relationship with an inverse quadratic function ( f[M] , Table 1 ) . As for growth rate , accessions and hybrids exhibited similar nonlinear relationships , characterized by coefficients that were not significantly different from each over ( Table 1 ) . We first compared what fraction of heterosis can be explained by genetic and phenotypic distances between inbred parents . For all hybrids , we quantified heterosis as the observed phenotypic deviation relative to mid-parent heterosis ( MPH ) value and best parental heterosis ( BPH ) value . Pairwise genetic distances were calculated either with all SNPs in the genome or with SNPs in the 1% top-genes associated with the corresponding trait [31] . For the phenotypic distance , we used the absolute difference in vegetative dry mass between parents . We fitted quadratic relationships to test for potential nonlinearity . The second-order term was not significant for genetic distance but significant for the relationships between phenotypic distance and MPH of growth rate ( Fig 6B ) and BPH of fruit number ( Fig 6D ) . This suggests an optimal phenotypic distance for maximization of heterosis , although it was not a good predictor on its own ( MPH of growth rate , r2 = 0 . 05 , Fig 6B; and BPH of fruit number , r2 = 0 . 09 , Fig 6D ) . For both traits , the relationships with phenotypic distance did not differ significantly between MPH and BPH ( 95% confidence intervals [CIs] of all regression coefficients overlapped between MPH and BPH ) . Genetic distance between parents was positively , and linearly , correlated with heterosis of growth rate ( both MPH and BPH P < 0 . 001; Fig 6A ) but accounted for less than 10% of heterosis ( 7% and 6% for MPH and BPH , respectively ) . By contrast , genetic distance did not correlate with heterosis of fruit number ( both P > 0 . 01; Fig 6C ) . Using only the 1% top-genes associated with growth rate and fruit number , which are more likely to have a causal role in the measured traits , did not improve correlations ( r2 < 0 . 10 ) . Even when combined in a multivariate model , genetic and phenotypic distances together explained less than 10% of heterosis of growth rate and fruit number ( S2 Table ) . For comparison , geographic distance between parents explained less than 2% of heterosis of fruit number and less than 0 . 2% of heterosis of growth rate . In summary , none of the parental distances—be they genetic , phenotypic , or geographic—had much power to explain and predict a significant portion of heterosis . In a second approach , we used the fitted equations ( Table 1 ) to take into account the nonlinearity of allometric relationships and to predict heterosis . Our first goal was to predict growth rate and fruit number with ( i ) the allometric relationship fitted on parents and ( ii ) the measurement of vegetative dry mass in hybrids ( Fig 7 ) . Predicted growth rate in hybrids strongly correlated with observed growth rate ( r2 = 0 . 95 , S2A Fig ) . By contrast , predicted fruit number was poorly correlated with observed trait values ( r2 = 0 . 08 , S2B Fig ) . This is consistent with the larger dispersion of trait values around the fitted curve for fruit number ( Fig 5B ) compared to growth rate ( Fig 5A ) . We then compared nonlinear deviation of traits predicted in hybrids—relative to the predicted mid-parent value ( nonlinear deviation [NLDevMP] ) and to the predicted best-parent value ( NLDevBP ) —with the observed MPH and BPH ( Fig 7 ) . Observed heterosis and predicted NLDev of growth rate were strongly correlated ( r2 = 0 . 75 and 0 . 66 for NLDevMP versus MPH and NLDevBP versus BPH , respectively; Fig 7B ) . Observed heterosis and predicted NLDev of fruit number were also positively correlated , although weaker than for growth rate ( r2 = 0 . 14 and 0 . 10 for NLDevMP versus MPH and NLDevBP versus BPH , respectively; Fig 7D ) . This suggests that allometric relationships allow the prediction of heterosis amplitude and that prediction accuracy depends on the strength of the underlying nonlinear covariation between traits .
Already in 1934 , Wright wrote in his seminal paper that ‘dominance has to do with the physiology of the organism and has nothing to do with the mechanism of transmission’ [42] . Eighty years later , the emergence of heterosis is still considered as an enigma , and its physiological bases remain debated . Despite the many dominant , overdominant , and epistatic QTL identified in a plethora of species [23 , 24 , 40] , none of the genetic models have been formally validated in more than a few cases [13] . Here , we approached the question of heterosis from a physiological angle based on a geometric analysis of trait–trait relationships . Combining the ideas put forth by Wright [42] with principles from metabolic scaling theory ( MST ) [52 , 65] , we demonstrate that a significant part of heterosis can be explained by the nonlinear relationships that link phenotypic traits with each other . Wright’s model of physiological dominance [42] was based on the nonlinear relationship that connects two traits at different levels of integration , enzyme concentration , and metabolic fluxes . Consistent with Wright’s model , evolutionary theory suggests that the intensity of inbreeding depression , and hence the potential for heterosis , should increase with the level of phenotypic integration [89] . The questions around the hierarchy of trait integration have been a key objective and enduring challenge for the field of phenotypic integration [90–93] . Following Arnold’s definition [94] , a trait is said to be integrated when it results from the action of several underlying traits ( and presumably , genes ) . This leads to a pyramidal view of phenotypic integration , in which gene products ( RNA , transcription factors , enzymes ) are at the bottom and fitness-related traits ( growth rate , fecundity ) at the top . In agreement with Arnold’s definition of phenotypic integration , several studies have shown that RNAs and transcription factors can regulate plant growth through their effects on intermediate traits such as plant development , vegetative meristem activities , cell elongation , photosynthetic efficiency , and secondary wall biosynthesis [95–97] . Under these assumptions , biomass is expected to be highly integrated but at a lower level than growth rate and fruit number [98] . Consistent with this view , survival and fertility are more sensitive to inbreeding depression than size , biomass , and gross morphology in animals [99] . In addition , fitness traits show lower heritabilities than morphological traits [100] . Our model supports the idea that integrated traits in plants like growth rate and fruit number exhibit strong heterosis because they result from the multiplicative effects of nonlinear relationships at different organizational levels [101 , 102] , such as abundances of RNA transcriptions and proteins [48–50] , mitochondrial respiration and cell growth [51] , and allometric laws of biomass allocation [63] . In this study , we focused on modelling the allometric variations of growth rate and fruit number . However , nonlinear relationships at successive organizational levels can have different curvatures and directions , which might either amplify or cancel each other . Moreover , different components of integrated traits could exhibit variable degrees of heterosis depending on how these components are connected to traits at lower levels of integration . For instance , there is an expected trade-off between seed size and seed number [103 , 104] , even though a recent work suggests that seed size can strongly vary without changes in seed number [105] . Thus , since the metabolic pathways of fruit size can differ from the one of fruit number , we could expect that the prediction of heterosis of total seed yield would be complex [106] and that it must rely on the joint analysis of several trait–trait relationships . The high fraction of negative heterosis observed in our study for plant biomass is consistent with previous findings in A . thaliana [17 , 71] , although it contrasts with some others [11 , 73] . It could potentially be explained by a convex relationship with a trait at a lower integration level [107] , for instance , a trait related to organism development and phenology . In addition , another explanation to the emergence of negative heterosis is that we used crosses from strongly divergent accessions , which might be incompatible because of deleterious interactions between defence-related genes [108] . Autoimmunity is associated with a reduction of growth and fecundity , and it has been repeatedly observed when crossing distant accessions of A . thaliana [81] . Moreover , a general repression of the immune system has been observed in hybrids with positive heterosis on biomass [73] . Consistent with a recent suggestion that inferior hybrids are as common within regions as between regions [81] , we did not find evidence of a connection between negative heterosis and parental geographic distance ( geographic distance was not correlated with MPH and BPH of plant dry mass , r2 < 0 . 01 ) . The genetic bases of integrated traits such as growth rate and fruit production are complex by nature because these traits result from the effects of numerous genes acting on different components of performance [30 , 101] . Inexpensive high-density genotype information coupled with very detailed phenotyping provides a series of promising avenues for the genomic prediction of heterosis [109] . In this context , the dominance hypothesis implicates , within certain limits , a positive relationship between parental genetic distance and heterosis [32 , 33] . Results from a range of species , however , do not conform with these expectations [11 , 17 , 35 , 110 , 111] . One of the reasons could be the often small number of crosses and the relatively small range of genetic distances analysed , with the latter holding true especially in cultivated species [11] . Our results with 450 hybrids representing crosses between diverse A . thaliana populations pointed to a positive but weak correlation between heterosis and parental genetic distance for growth rate but no such correlation for fruit number . This suggests that a genetic approach alone may not be sufficient to accurately predict heterosis . By contrast , our results indicate that phenotypic nonlinearity determines the emergence of heterosis . We have shown that even if vegetative dry mass exhibited important deviation from additivity , it could still be used to predict heterosis of growth rate and fruit number based on parental allometric relationships . Our model notably predicts that strong heterosis of biomass ( e . g . , BPH ) is always associated with strong heterosis of growth rate . However , strong heterosis of biomass can be associated with negative heterosis of fruit number because fruit number is linked to biomass with a hump-shaped curve . For instance , consider a case in which two parents are to the right of the fitness peak: it is expected that BPH of biomass results in fewer fruits than for either parent . As observed in yeast [112 , 113] , this example illustrates how phenotypic nonlinearity can generate overall positive heterosis despite many single loci that are inherited as underdominant . The accuracy of heterosis prediction with phenotypic nonlinearity depends on how strongly traits correlate with each other . MST [52 , 55 , 57] postulates that body size constrains trait variation in the form of universal mathematical laws ( Box 1 ) . According to the mechanistic assumption of MST , biomass integrates the number of metabolically active units ( cells , mitochondria ) . Because the number of metabolically active units is the main driver of physiological and metabolic variation between individuals , biomass is thus expected to predictably determine physiological fluxes , growth rate , and metabolic activity at the organismal level [65 , 66] . We have found that hybrids exhibit the same allometric relationships as natural inbred accessions . Moreover , the relationship for growth rate is similar to the one observed within and across species as well as in recombinant inbred lines ( RILs ) that never experienced natural selection [31 , 60 , 114] ( Fig 2 ) . This reinforces the idea of strong , evolutionary stable and predictable biophysical constraints on the variation of growth rate with biomass . For instance , our method explained up to 75% of heterosis amplitude for growth rate ( Fig 7B ) while genetic distance at best explained only 7% ( Fig 6A ) . Consistent with the present study , Flint-Garcia and colleagues [10] with maize , Seymour and colleagues [17] as well as Palacio-Lopez and colleagues [35] with A . thaliana , all found very weak correlation between heterosis and genetic distance between parents ( r2 < 0 . 10 ) . The prediction of heterosis using the nonlinearity of the allometric relationship was lower for fruit number ( 14% for MPH ) , which can be attributed to the higher variation of this trait around the fitted curve . Fruit number is difficult to measure accurately , so a possible explanation to this low prediction compared to growth rate is that our protocol to estimate fruit number was prone to a certain degree of uncertainty ( e . g . , measurement errors due to time of sampling , image analysis ) . Alternatively , we can suppose that fruit number depends on much more traits than just biomass and that it is more plastic than growth rate to small environmental variations . For instance , source/sink competition between vegetative biomass and reproductive output , visible here by the decrease of fruit number after a certain size threshold ( Fig 5B ) , is expected to vary in response to the abundance of resources . Previous studies have shown that the nonlinearity of relationships between traits and genetic markers that explain heterosis in maize [115] and jack pine [107] was more pronounced under stressful conditions . This suggests that environmental factors could play an important role in the manifestation of heterosis by increasing the curvature of trait relationships [116–118] . Since allometric relationships are expected to vary with the environment [87 , 119]—in particular the relationship between plant biomass and fruit production—an open question that needs to be addressed in the future is how genotype-by-environment interactions impact the prediction of heterosis through their effects on trait covariation . Predictions of heterosis can reach relatively high levels in specific biparental populations in which dominant or overdominant QTL have been mapped ( e . g . , [13 , 27] ) , but such QTL cannot be generalized to other populations with different allelic composition and genetic determinism of traits . Several authors have proposed powerful methods based on genomic predictions to predict hybrid performance , such as de Abreu e Lima and colleagues [120] , who predicted up to 41% of hybrid performance in maize; Zhao and colleagues [121] , who predicted up to 89% of hybrid performance in wheat; and Werner and colleagues [122] , who predicted up to 82% of hybrid performance in oilseed rape . However , these studies predicted hybrid performance , i . e . , hybrid trait value , and not heterosis per se ( i . e . , the deviation of hybrid trait value to the parental mean value or parental best value ) . For crop breeding , it is hybrid performance that is the fundamental unit of financial care , but it is possible that a good prediction of hybrid performance fails to accurately predict the emergence and amplitude of heterosis . For instance , our study suggests that high-performance hybrids are not necessarily those that exhibit the strongest amplitude of heterosis ( 0 . 31 < r2 < 0 . 47; S3 Fig ) . Inversely , if a hybrid has high heterosis but is not in the top for hybrid performance , then it is not useful in a crop setting . Our study thus calls for an assessment of whether predictive models of hybrid performance are good predictive models of heterosis . From a theoretical point of view , the predominance of outcrossing and the maintenance of recessive alleles among organisms have been suggested to be directly linked to nonadditive inheritance and superior performance of hybrids [123] . Wright’s initial model of physiological dominance was proposed as a response to Fisher’s idea that ‘gene modifiers of dominance’ must exist and be selected to maximize the fitness of the heterozygote [124] . Wright [42] , and later Kacser and Burns [44] , claimed that modifiers are not necessary because nonlinearity is an intrinsic characteristic of metabolic networks . The same argument holds true for complex traits such as growth rate and fruit number . The assumption of modifiers of dominance is based on the unrealistic expectation of an intermediate phenotype in the heterozygote , while phenotypic relationships are essentially nonlinear [46 , 63 , 125] . A next step in getting to the physiological root of heterosis will be the integration of other , more low-level traits , especially gene expression and metabolite levels . We expect that this will improve the characterization of nonlinear relationships in multidimensional phenotypic space and ultimately shed light on the physiological mechanisms at the origin of nonadditive inheritance .
The development of a predictive approach for heterosis is a long-term goal of modern biology , especially in the applied framework of varietal selection in crops [126] . Our study has shown that trait variation is similarly constrained in accessions and hybrids of A . thaliana , which highlights the power of a geometric approach of trait–trait relationships for predicting heterosis related to two major components of plant productivity and yield . It opens promising avenues for cultivated species with a perspective of targeting optimal crosses based on allometric relationships in parental lines . However , several differences between crops and A . thaliana might hamper the prediction of heterosis based on allometric relationships . For instance , the genetic architecture of metabolism and performance-related traits could differ in crops compared to A . thaliana . Strong selection for high-yielding lines in crops may have reduced genetic variability in modern varieties and may have modified to a certain degree the relationship between genetic and phenotypic distance . Deleterious recessive alleles may have been purged in cultivated species as a result of domestication and human selection . Unfortunately , most plant species that have been evaluated for inbreeding depression and heterosis are cultivated species , and we lack a proper comparison of wild and cultivated species regarding these processes . In addition , traits other than biomass ( harvesting date , plant height ) could be relevant for prediction of hybrid performance in crops , especially in field conditions . For instance , plant height might be more related to grain yield than biomass , as suggested by the quadratic relationship between these traits in wheat [127] . With respect to understanding the emergence of heterosis of complex traits , the analysis of the relationships between traits and metabolomic profiles is promising . For instance , correlations between performance-related traits and the metabolome have been reported in A . thaliana [85] , rice hybrids [128] , and maize hybrids [6] . It is now time to test the phenotypic approach to heterosis in crops , for which the study of allometric relationships is a nascent research front [129–131] .
For inbred genotypes , 451 natural accessions of A . thaliana were phenotyped in 2014 at the Max Planck Institute for Developmental Biology in Tübingen ( MPI-Tübingen ) , Germany ( n = 2 , Exp 1 , S1 Data ) . They are the same individuals as those used for the analysis of the natural variation in growth rate and plant allometry [31 , 132] . These accessions were chosen to represent a wide geographic area: a majority of them were derived from Russia and European countries , with two additional accessions from Japan and a few from North Africa . To generate the 450 hybrids phenotyped in a second experiment in 2015 at MPI-Tübingen ( n = 4 , Exp 2 , S1 Data ) , 415 accessions were used as parental lines and randomly crossed . Among the parental accessions , 342 were used as female parent and 318 as male parent . Genetic data were available for 407 accessions among the 451 total and 369 accessions among the 415 parental , as they had been genome sequenced as part of the 1001 Genomes project ( http://1001genomes . org/ ) [76] . Among the 415 parental accessions , 134 ( 32% ) were used in a single cross , 166 ( 40% ) were used in two crosses , 63 ( 15% ) in three crosses , and 52 ( 13% ) in at least four crosses . To overcome potential maternal effects , the same mother plants grown in the greenhouse in 2013 provided the seeds for both the inbred accessions ( by self-fertilization ) and the hybrids ( by manual cross ) ( S4A Fig ) . Data for the 334 plant species shown in Fig 2A were obtained from the study by Niklas and Enquist [66] , with unicellular algae removed . The data consist of annual growth rate estimates across aquatic and terrestrial metaphytes , including herbaceous dicots as well as arborescent monocots , dicots , and conifers . For terrestrial metaphytes , traits were measured on even-age monospecific stands grown under horticulturally controlled field [66] . Most of the data were originally compiled by Cannell [133] based on the primary literature published up to mid-1981 for approximately 1 , 200 forest stands from 46 countries . We designed a hydroponic system in which plants were cultivated on inorganic solid media ( rockwool ) , and all nutrients were provided through the watering solution . Circular pots ( Pöppelmann , Lohne , Germany ) of 4 . 6 cm ( diameter ) x 5 cm ( depth ) were filled with 3 . 6 cm x 3 . 6 cm x 4 cm depth rockwool cubes ( Grodan cubes , Rockwool International , Denmark ) . Pots were covered with a black foam disk with a 5–10 mm central circular opening . Seeds were sown in individual pots , randomly distributed in trays of 30 pots each ( S4B Fig ) . Before sowing , all seeds were surface-sterilized with 100% ethanol and frozen overnight at −80°C to kill any insect eggs . The rockwool cubes were placed in 75% strength nutrient solution , as described in ref . [134] in order to achieve full humidification and fertilization . After sowing on the surface of the rockwool cubes , trays with 30 pots each were incubated for 2 days in the dark at 4°C for stratification . Trays were transferred for 6 days to 23°C ( 8-h day length ) for germination . After 6 days , when most seedlings had two cotyledons , trays were transferred to 4°C ( 8 h light ) for 41 days of vernalization in order to reduce the range of flowering times among accessions . During germination and vernalization , all trays were watered once a week with a 75% strength nutrient solution . After vernalization , when true leaves had emerged on most individuals , plants were thinned to one individual per pot , and trays were moved to the Raspberry Pi Automated Plant Analysis ( RAPA ) facility [132] ( S4B Fig ) , set to 16°C , air humidity at 65% , and 12-h day length , with a PPFD of 125 to 175 μmol m−2 s−1 provided by a 1:1 mixture of Cool White and Gro-Lux Wide Spectrum fluorescent lights ( Luxline plus F36W/840 , Sylvania , Germany ) . Trays were randomly positioned in the room and watered every 1 to 3 days with 100% strength nutrient solution . Plants were grown and phenotyped using rigorously the same protocol , following methodologies previously published for accessions [31 , 132] . Plants were harvested at the end of the life cycle when the first fruits were senescing . Flower production ceased at this stage , so only a few fruits were expected to be newly produced when the first fruits were drying [132] . Consistently , when we left the plants under well-watered conditions after the senescence of the first fruits , we could not detect the formation of new stems , and fruit yield was only marginally increased . At harvesting , rosettes were separated from roots and reproductive parts , dried at 65°C for 3 days , and weighed . Plant age at reproduction ( d ) was measured as the duration between the appearance of the two first leaves after vernalization and the end of the life cycle [132] . Growth rate ( mg/d−1 ) was calculated as the ratio of final rosette dry mass over plant age at reproduction . Inflorescences were photographed with a high-resolution , 16 . 6 megapixel SLR camera ( Canon EOS-1 , Canon Inc . , Japan ) and analysed with ImageJ [135] to estimate the number of fruits through 2D image skeletonization , following published protocols [31 , 132] . The RAPA system was used for daily imaging using 192 microcameras ( OmniVision OV5647 ) , which simultaneously acquired 6 daily top-view 5-megapixel images for each tray during the first 25 days after vernalization . We used a published method to estimate plant dry mass during ontogeny from top-view rosette pictures [31 , 132] . From fitted sigmoid growth curves on all individuals , we calculated inflection point ( d ) at which daily growth was maximal and started to decrease . We used rosette dry mass ( mg ) at the inflection point as measurement of vegetative dry mass M ( mg ) . In total , trait values for plant age at reproduction , growth rate , and vegetative dry mass were available on 451 accessions and 447 hybrids and fruit number on 441 accessions and 449 hybrids ( S1 Data ) [136] . To correct for potential biases between the two experiments performed , 16 accessions phenotyped in Exp 1 were also included in Exp 2 . Among all traits measured , only plant age at reproduction exhibited a significant difference between the two experiments ( P = 0 . 03 ) . We thus corrected trait values in Exp 2 with the following equation: agecorrected at reproduction = −37 + 1 . 8 × ageobserved at reproduction . First , hybrid phenotypic classes were categorized by comparing trait distribution in hybrids to mean and best ( or worst ) parental values . Trait distributions for hybrids were obtained from a bootstrap approach with 1 , 000 permutations . Each F1 trait distribution was compared to the mean parental value using a two-sided t test ( ɑ = 0 . 05 ) as well as to the worst and best parent values using a one-sided t test ( ɑ = 0 . 05 ) . To determine the significance of heterosis categories ( below or above mean/best/worst parental values ) , P values were adjusted for multiple testing by a Bonferroni correction . Secondly , two metrics commonly used in the literature to quantify heterosis [17] were calculated in the present study for all traits Y , i . e . , growth rate and fruit number: The allometric equations were fitted by the nonlinear least-squares method using the nls function in R [137] . Following MST model [31 , 52 , 60 , 88] , we chose a power-law equation for growth rate with a mass-corrected allometric exponent ( g[M] in Table 1 ) . The corrected exponent corresponds to the derivative of the quadratic function obtained after logarithmic transformation of the allometric relationship [31 , 60 , 88] . In our data , biomass correction of the exponent improved the fitting of the allometric relationship of growth rate ( ΔAIC = 53 and 55 for accessions and hybrids , respectively ) . For the allometric relationship of fruit number , we compared the fitting between a Ricker function ( y = axe−bx ) and inverse polynomial function . The inverse polynomial equation ( f[M] in Table 1 ) was retained based on AIC ( ΔAIC = 12 and 22 for accessions and hybrids , respectively ) . Confidence intervals ( 95% CIs ) of the fitted coefficients were estimated with the confint function in R . We used the allometric equations fitted on the accessions to predict the phenotype of the hybrids from vegetative dry mass M . We first estimated growth rate of both parents and hybrids ( g[M1] , g[M2] , g[M1 x 2] , with M1 and M2 corresponding to vegetative dry mass of worst and best parent , respectively , and M1 x 2 to hybrid dry mass ) . We then predicted mean parental value of growth rate as MPpred = ( g[M1] + g[M2] ) /2 and best parental value as BPpred = max ( g[M1] , g[M2] ) . Finally , we predicted the phenotypic nonlinearity as Both NLDevMP and NLDevBP were calculated for growth rate , and we performed similarly for estimating the nonlinearity of fruit number ( Fig 7C ) . Out of the 451 natural accessions phenotyped here , 407 accessions have been genome sequenced ( http://1001genomes . org/ ) [76] . Using vcftools [138] , the 12 , 883 , 854 single-nucleotide polymorphisms ( SNPs ) were first filtered to retain those in which minor allele frequency was above 5% , with a genotyping rate above 85% across all accessions . This resulted in 391 , 016 SNPs . We used PLINK v1 . 9 [139] to estimate pairwise genetic distances as the number of alleles that differed between pairs of accessions ( —distance function ) after log10-transformation . We also measured genetic distances using the 1% SNPs with the strongest ( positive and negative ) effects on growth rate or fruit number from a previously published study [31] . SNPs effects on each trait were estimated using a polygenic GWA model implemented in GEMMA ( ‘Bayesian Sparse Linear Mixed Model’ [BSLMM] ) [140] . Pairwise geographic distances between accessions were estimated from their longitude–latitude coordinates [76] and the distm function of the geosphere package in R . For the calculation of pairwise phenotypic distances , we first used Euclidean distance among all traits measured in accessions ( vegetative dry mass , age at reproduction , growth rate , fruit number; used for Fig 1 ) . We also used absolute difference in vegetative dry mass between parents for comparing the contribution of genetic and phenotypic distances to heterosis in Fig 6 . Pearson correlation and regression coefficients were calculated using R . The effect of the experiment on trait values was tested on the 16 accessions common to Exp 1 and Exp 2 using two-way ANOVA , with genotype and experiment as interacting factors . To determine heterosis categories ( Fig 4 ) , F1 trait distributions were estimated with a nonparametric bootstrap approach across 1 , 000 permutations using the boot . ci function of the simpleboot R package and compared to mean , worst , or best parental values for each F1 with a one- or two-sided t test . P values were corrected for multiple testing by a Bonferroni approach . To calculate the proportion of phenotypic variance associated with genotypic ( G ) effects ( a measure of broad-sense heritability , H2 = var[G] / [var ( G ) + residuals] ) , we fitted a mixed model ( lmer function in R ) as Y = Genotype + residuals , in which Y is trait value and Genotype is used as random factor . Statistical analyses were conducted in R v3 . 2 . 3 [137] . Allometric relationships on log10 scale ( S1 Fig ) were fitted with SMA regressions , using the package smatr [141] in R . Smooth distributions of trait values ( Fig 4 ) were obtained with kernel density estimation ( KDE ) , a nonparametric way to estimate the probability density function , using the geom_density function of the ggplot2 package in R , with a bandwidth of 150 mg for vegetative dry mass , 9 days for plant age at reproduction , 3 mg/d−1 for growth rate , and 25 for fruit number . | Hybrids often grow faster and produce more offspring than their parents . This phenomenon , called hybrid vigour or heterosis , has been extensively exploited in agriculture . Even though several hypotheses have been proposed to explain the genetic basis of hybrid superiority , there is still no unifying model able to accurately predict the extent and amplitude of heterosis . Here , we tested a model based on physiological constraints to explain the emergence of heterosis in the model plant Arabidopsis thaliana . Rooted in the mathematical modelling of phenotypic relationships , our study demonstrates that heterosis can be explained simply with the nonlinearity of trait variation . Our model notably predicts the amplitude of heterosis of two performance traits: growth rate and fruit production . We expect that our model can help to understand the physiology of nonadditive inheritance in different species and that it will open new avenues in both theoretical and applied biology . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Conclusions",
"Material",
"and",
"methods"
] | [] | 2019 | Nonlinear phenotypic variation uncovers the emergence of heterosis in Arabidopsis thaliana |
The two-component NS2B-NS3 proteases of West Nile and dengue viruses are essential for viral replication and established targets for drug development . In all crystal structures of the proteases to date , the NS2B cofactor is located far from the substrate binding site ( open conformation ) in the absence of inhibitor and lining the substrate binding site ( closed conformation ) in the presence of an inhibitor . In this work , nuclear magnetic resonance ( NMR ) spectroscopy of isotope and spin-labeled samples of the West Nile virus protease was used to investigate the occurrence of equilibria between open and closed conformations in solution . In solution , the closed form of the West Nile virus protease is the predominant conformation irrespective of the presence or absence of inhibitors . Nonetheless , dissociation of the C-terminal part of the NS2B cofactor from the NS3 protease ( open conformation ) occurs in both the presence and the absence of inhibitors . Low-molecular-weight inhibitors can shift the conformational exchange equilibria so that over 90% of the West Nile virus protease molecules assume the closed conformation . The West Nile virus protease differs from the dengue virus protease , where the open conformation is the predominant form in the absence of inhibitors . Partial dissociation of NS2B from NS3 has implications for the way in which the NS3 protease can be positioned with respect to the host cell membrane when NS2B is membrane associated via N- and C-terminal segments present in the polyprotein . In the case of the West Nile virus protease , discovery of low-molecular-weight inhibitors that act by breaking the association of the NS2B cofactor with the NS3 protease is impeded by the natural affinity of the cofactor to the NS3 protease . The same strategy can be more successful in the case of the dengue virus NS2B-NS3 protease .
West Nile virus ( WNV ) is a flavivirus related to yellow fever virus , dengue virus , and Japanese encephalitis virus all of which cause human diseases . During infection , the flavivirus RNA genome is translated into a polyprotein comprising of three structural and seven non-structural proteins [1] . The N-terminal part of nonstructural protein 3 ( NS3 ) encodes a serine protease that cleaves the polyprotein into several components . The activity of the NS3 protease ( NS3pro ) is greatly enhanced by covalent tethering of about 40 residues from the membrane-bound NS2B protein that acts as a co-factor . NS3 is essential for viral replication making it an attractive drug target [2]–[4] . The C-terminal part of NS3 contains a nucleotide triphosphatase , an RNA triphosphatase , and a helicase which have only little influence on the protease activity [5] . Crystal structures of WNV NS2B-NS3pro in the absence of inhibitor [6] and in the presence of tetra- and tripeptide inhibitors [7] , [8] or bovine pancreatic trypsin inhibitor ( BPTI ) [6] have been determined . The fold of NS2B is very different in the presence of inhibitors from that in the absence of inhibitor ( Figure 1 ) . In all structures , the N-terminal segment of NS2B ( residues 52–58 ) inserts into a β-sheet formed by NS3pro . In the presence of inhibitor , the C-terminal segment ( CTS ) of NS2B wraps around NS3pro , bringing the C-terminal β-hairpin of the NS2B cofactor in close proximity of the active site . This fold is referred to in the following as the closed conformation . In the absence of inhibitor , the NS2B CTS is located at a very different position far from the active site of the protease . We refer to this fold and any other conformation , where NS2B is disengaged from the substrate binding site , as open conformations . As the NS2B CTS is essential for full catalytic activity of WNV NS2B-NS3pro [2] , [9] the closed conformation appears to be a prerequisite for full proteolytic activity . The fold of the corresponding protease from the closely related dengue virus type 2 ( DENV ) NS2B-NS3pro construct was also determined by X-ray crystallography in the absence of an inhibitor [7] . NS2B without the CTS results in an inactive protease suggesting that this part of the cofactor forms part of the active site [7] . The fold observed in DENV is remarkably similar to the open conformation of WNV NS2B-NS3pro , suggesting that the open conformation may be the predominant species in solution ( Figure 1 ) . Alternatively , the open conformation may be a crystallographic artefact . We undertook the present research in order to address this issue and also because of the failure of high throughput drug screens to identify stable low-molecular weight compounds that bind to the WNV NS2B-NS3 protease specifically and with subnanomolar affinity [10]–[15] . Better understanding of the dynamics of the NS2B cofactor of the protease could be the key towards the design of improved inhibitors and has important implications for the conformational space accessible to the protease when bound to the host cell membrane . In the following , we present an NMR analysis of the conformational equilibria of the WNV NS2B-NS3 protease in the absence and presence of inhibitors .
1-oxyl-2 , 2 , 5 , 5-tetramethyl-Δ3-pyrroline-3-methyl ) methane thiosulfonate ( MTSL ) was purchased from Toronto Research Chemicals ( North York , Ontario , Canada ) . Compound 1 ( Figure 2 ) was obtained from Maybridge ( Tintagel , UK ) . Compound 2 was synthesized in-house . The West Nile virus protease construct used contained NS2B covalently linked to NS3pro via a Gly4-Ser-Gly4 linker [16] as used for crystallization [7] . In addition , Lys96 of NS2B was mutated to alanine to prevent self-cleavage of the protease [17] . In the following , this construct and the unmutated wild type are referred to as NS2B-NS3pro and wt NS2B-NS3pro , respectively . A second construct containing the additional mutation N89C at the C-terminus of NS2B was prepared to provide a thiol group for the attachment of a spin-label . This construct is referred to as NS2B-NS3proC . The N89C mutation was made by site-directed mutagenesis using PCR . Uniformly 15N/13C- and 15N-labeled protein samples of NS2B-NS3pro and five combinatorially 15N-labeled samples of the NS2B-NS3 protease without the K96A mutation were prepared as described previously [18] . In vivo protein yields were about 9 mg of purified protein per litre of medium . The selectively 15N-Ile labeled sample was prepared by cell-free protein synthesis as described previously [18] , [19] . The nitroxide radical tag [ ( 1-oxyl-2 , 2 , 5 , 5-tetramethyl-Δ3-pyrroline-3-methyl ) methane thiosulfonate , MTSL] was used for spin-labeling . A 0 . 3 mM solution of NS2B-NS3proC in 0 . 5 ml reaction buffer ( 50 mM Tris , pH 7 . 6 ) was treated with 5 equivalents of DTT and washed with DTT-free reaction buffer using a Millipore ultrafilter with a molecular weight cutoff of 5 kD . Reaction buffer was added to a volume of 4 ml . 30 equivalents of MTSL were dissolved in 60 µl acetone and the MTSL solution was added stepwise , mixing the solution well after each addition . The mixture was stirred at room temperature for about 12 hours and subsequently centrifuged to remove undissolved MTSL , followed by concentration to 0 . 3 ml and washing with 20 mM Tris , pH 7 . 2 . All NMR spectra were recorded at 25°C on Bruker 600 and 800 MHz NMR spectrometers equipped with cryoprobes . 15N-HSQC spectra of the combinatorially 15N-labeled samples were recorded in a 20 mM HEPES buffer ( pH 7 . 0 ) containing 1 mM TCEP . 15N-HSQC spectra of a 0 . 26 mM solution of uniformly 15N-labeled NS2B-NS3proC derivatized with MTSL were recorded in 20 mM Tris , pH 7 . 2 , using the 800 MHz NMR spectrometer with t1max = 25 ms and t2max = 73 ms . Inhibitor 2 was added to a final concentration of 0 . 6 mM by adding 3 µl of a 100 mM stock solution in DMSO-d6 . A 20 mM stock solution of inhibitor 1 in 50% H2O/50% DMSO-d6 was used for preparing samples containing inhibitor 1 . 15N-relaxation rates R2 were measured at a 1H NMR frequency of 600 MHz using the CPMG sequence of Farrow et al . [20] with relaxation delays of 8 . 8 , 17 . 6 , 26 . 4 , 35 . 2 , 44 . 0 , 52 . 8 , 61 . 6 , 70 . 4 , 79 . 2 and 88 . 0 ms and a τcp delay between subsequent 180° ( 15N ) pulses of 900 µs . The protein concentration was 0 . 9 mM in 20 mM HEPES , pH 7 . 2 , 2 mM DTT at 298 K . Experiments with inhibitor were performed with 3 mM inhibitor 2 . The data were analyzed using the program Sparky [21] . The chemical shifts were deposited in the BioMagResBank ( accession code 16359 ) .
In contrast to NMR spectra in the presence of inhibitors which allowed virtually complete resonance assignments of the 15N-HSQC spectra by conventional 3D NMR techniques [18] , many of the cross-peaks in the 15N-HSQC spectrum of NS2B-NS3pro are broadened beyond detection in the absence of an inhibitor , making the assignment of the NMR resonances challenging ( Figure S1 ) [14] , [15] , [18] . In order to assign the observable cross-peaks , we used the previously established resonance assignments of the 15N-HSQC spectrum of the protease in the presence of inhibitor 1 ( Figure 2 ) and an inhibitor closely related to inhibitor 2 ( compound 1 reported by Su et al . [18] ) as a starting point . The Kd values of the inhibitors are in the 10–100 µM range [13] , [14] . The inhibitors were in fast exchange between bound and free state , so that titration of the protein with inhibitor yielded a series of 15N-HSQC spectra with continuously shifting cross-peaks , allowing tracking of the resonance assignments for the resolved cross-peaks . In addition , combinatorial 15N-labeling established the amino acid type associated with each cross-peak . As many of the cross-peaks in the 15N-HSQC spectrum of uniformly 15N-labeled NS2B-NS3pro are overlapped , a selectively 15N-Ile labeled sample was prepared for improved spectral resolution . The 15N-HSQC spectra of the 15N-Ile labeled sample changed greatly in appearance upon addition of the inhibitor 1 ( Figure 3A ) . In particular , all isoleucine residues appeared as single peaks in the presence of 1 , whereas many peaks were missing , significantly shifted or split into several peaks in the absence of inhibitor . Extreme broadening of some but not all lines is a hallmark of chemical exchange of a protein subdomain between different conformations . Figure 3B shows the locations of the isoleucine residues in the closed conformation , identifying the residues that were strongly or only slightly affected by the presence of inhibitor . The isoleucine residues located in the NS2B CTS or in parts of NS3 in the vicinity of the NS2B CTS are the isoleucine residues most prone to conformational heterogeneity . Too many residues are affected to explain these effects by conformational flexibility of a few active-site residues . The most plausible explanation is that the NS2B CTS assumes multiple conformations in the absence of an inhibitor . We previously reported nuclear Overhauser effects ( NOEs ) that indicate that the closed conformation of Figure 3A is the prevailing conformation in the presence of inhibitor 1 [18] . In the absence of inhibitor , the closed conformation may still be present but a conformational equilibrium exists that cannot be with the open conformation of Figure 3C as the only additional species , as this would significantly alter the chemical environment of Ile60NS2B . The cross-peak of Ile60NS2B is , however , essentially unperturbed . Comparison of the 15N-HSQC spectra of uniformly 15N-labeled NS2B-NS3pro confirmed the analysis above . Due to spectral overlap , the absence of cross-peaks in the sample without inhibitor could be reliably assessed only for resolved cross-peaks . Nonetheless , we could confirm that the absence of inhibitor led to many missing cross-peaks in the NS2B CTS segment following Glu73 and for segments 73–76 , 110–116 , 124–130 and 150–153 of NS3 ( Figure 3D ) . The overall picture is that of extensive dynamics in and around the substrate binding site including , in particular , the C-terminal β-hairpin of NS2B . In contrast , cross-peaks were observable in both states with largely conserved intensities for residues near the N-terminal segment ( NTS ) of NS2B , as expected for a stable association of the NS2B NTS to NS3 . Binding of 1 caused significant chemical shift changes ( >0 . 05 ppm ) for a large part of the protein , highlighting the extent of conformational adjustments to the local binding event . Remarkably , the experiment could not identify excessive line broadening for any peaks observable in the segment between residues 53 and 72 of NS2B , whereas the data of Figure 3A–C had identified Ile68 of NS2B as a significantly affected residue . The combined results suggest that , in contrast to the crystal structure data , only the CTS of NS2B following Ser72 is prone to dissociation from NS3pro and that the line broadening of Ile68NS2B is due to a different effect . The cross-peaks of His51 and S135 of the catalytic triad of NS3 were observable both with and without inhibitor . Therefore , mobility of the side-chain of the active-site histidine , His51 , that was reported in a recent crystallographic analysis [8] or other conformational changes in the active site cannot be the cause of the excessive line broadening observed in the absence of inhibitors . Dissociation of the NS2B CTS from NS3 thus is the main cause for the observed broadening and absence of cross-peaks in the protease without inhibitor . Nonetheless , the NS2B CTS does not exclusively populate a highly mobile random coil conformation ( which would result in very narrow peaks ) but interacts at least to some extent with NS3 . The flexibility of the polypeptide chain on the subnanosecond time scale was probed by measurement of the R2 ( 15N ) relaxation rates of uniformly 15N-labeled NS2B-NS3pro . The average relaxation rate was slightly higher in the absence of inhibitor , which may reflect a greater tendency for aggregation ( Figure 4 ) . Only the residues C-terminal of Glu173NS3 displayed relaxation rates characteristic of a highly mobile random-coil peptide . In contrast , as far as the relaxation rates of NS2B could be assessed , they were similar to those of NS3 , both in the absence and presence of inhibitor , suggesting continuous association between the N-terminal segment of NS2B until Ser72NS2B and NS3 . In the presence of 1 , the NS2B CTS displayed some of the largest R2 values , suggesting that the inhibitor did not completely suppress the chemical exchange . Finally , at least ten of the linker residues tethering the C-terminus of NS2B to the N-terminus of NS3 ( not shown in Figure 4 due to missing sequence specific assignments ) displayed R2 relaxation rates characteristic of highly mobile segments ( below 10 s−1 ) , regardless of the presence or absence of inhibitor . This indicates that the peptide linker between NS2B and NS3 presents little hindrance for the dissociation of the NS2B CTS from NS3 . In order to gain more insight into possible conformational equilibria of the NS2B CTS , we used NS2B-NS3proC with an MTSL spin-label at Cys89 ( location shown in Figures 3B and C and 5D–F ) . As wild-type WNV NS3pro contains a buried cysteine residue at position 78 , we also tested the reactivity of Cys78 in a control experiment by treating wild-type NS2B-NS3pro with MTSL under the same reaction conditions . Analysis by 15N-HSQC spectra showed that the cross-peak of Cys78 did not shift or change in intensity , confirming its inaccessibility to MTSL ( Figure S1 ) . As expected for a residue without specific long-range contacts , the N89C mutation did not affect the structural integrity of the protease as evidenced by the close similarity of the 15N-HSQC spectra of NS2B-NS3pro and NS2B-NS3proC except for sequential neighbours of residue 89 ( data not shown ) . Figure 5A shows the results of the spin-labeling experiment . Different 15N-HSQC cross-peaks were attenuated differently by the paramagnetic spin-label in a way that is much more readily explained by the closed conformation of Figure 1A ( Figure 5B ) than the open conformation of Figure 1B ( Figure 5C ) . In particular , the signal attenuations near Cys89NS2B and Ser160NS3 were similarly pronounced , indicating that the closed conformation is the predominant species both in the presence and absence of the inhibitor 2 . There were , however , also significant differences between the paramagnetic attenuations observed in the presence and absence of the inhibitor 2 ( Figure 5D and E ) which must arise either from minor conformational species or from intermolecular effects . In order to assess possible intermolecular effects , we evaluated the signal attenuations at two different concentrations ( Figure S2 ) . A pronounced concentration dependence was observed for the exposed loop region with residues 29–32 and nearby residues ( e . g . Gly103 ) , indicating that these regions were significantly affected by intermolecular effects , partly or wholly explaining the apparent discrepancies between the plots of Figure 5D and E for those residues . An increased tendency for intermolecular aggregation in the absence of inhibitor as suggested by the R2 ( 15N ) data ( Figure 4 ) also explains why the PRE effects of these residues are greater in the absence of inhibitor ( Figure 5A ) . The PRE effects of at least four residues ( Thr50NS2B , Arg56NS2B , Gly63NS3 and Trp89NS3 ) are pronounced in the absence of inhibitor , depend only little on concentration and cannot be explained by the closed conformation ( Figure 5F ) . We interpret those as intramolecular effects caused by minor conformational species resulting from transient dissociation of the CTS of NS2B from NS3pro . These open conformations must be more heterogeneous than the open conformation of Figure 1B . The relaxation enhancements of all four residues were significantly less pronounced in the presence of inhibitor ( Figure 5A ) , indicating stabilization of the closed conformation by the inhibitor . Therefore , the PRE data support the notion of an equilibrium between a major conformation corresponding to the closed conformation of Figure 1A and an ensemble of transient open conformations that are generally different from the conformation of Figure 1B . Inhibitors shift the equilibrium towards the closed conformation but the equilibrium persists to some extent also in the presence of inhibitors .
The wild-type WNV polyprotein is associated with the host cell membrane . Also after proteolytic processing of the polyprotein , the NS2B-NS3 protease remains membrane associated via two transmembrane helices N-terminal of the NS2B cofactor part displayed in Figure 1A and B . In addition , the C-terminus of the NS2B cofactor is connected to NS3 by a highly hydrophobic segment of 35 residues [1] ( replaced by a Gly4-Ser-Gly4 linker in our construct ) that is also thought to insert into the host cell membrane [22] , [23] . This ties the NS2B cofactor to the membrane at either end . Auto-proteolytic cleavage occuring near Lys96NS2B and Lys15NS3 excises the segment between NS2B and NS3 in vitro [7] , [17] but does not affect the N-terminal membrane attachment . As the cleavage site near Lys96NS2B lacks the characteristic recognition sequence of two sequentially neighboring basic residues and is also not conserved in the highly homologous dengue virus NS2B-NS3 protease , cleavage at this site may be inefficient , in which case NS2B would remain tethered to the host cell membrane at either end . Independent of whether the NS2B cofactor is tied to the membrane at one or both ends , there is no reason why the association of NS3pro with the NS2B CTS should be any tighter than that observed in the model system studied here , once the covalent linkage between NS2B and NS3 has been broken . The association between NS2B and NS3pro appears to be independent of the helicase domain of NS3 . In the crystal structure of a DENV NS2B-NS3 construct comprising both the protease and helicase domains of NS3 , the interface is centered about residue 68 of the NS3 , displacing the N-terminal β-strand of NS3 observed in the structure 2IJO ( dark blue in Figure 1A ) , and the helicase domain makes no contacts with NS2B or the substrate binding site [24] . The same arguments apply to the WNV homologue [22] , [23] . It is not clear , however , whether the crystal structure of the DENV NS3 protease-helicase construct is a good model for membrane-associated NS2B-NS3 as it suggests that the helicase domain clashes with the membrane when both ends of NS2B are tied to a planar lipid bilayer . If the NS3 helicase domain is only loosely associated with the NS3 protease domain , the NS3 protease domain could dissociate from the NS2B CTS to access substrate cleavage sites further away from the membrane surface , although the separation from the NS2B CTS would simultaneously impede its proteolytic activity . The NS2B CTS could , however , follow the NS3 protease domain if its C-terminal membrane attachment is broken by cleavage at the non-canonical site near Lys96NS2B , generating a catalytically active NS3 protease that is anchored to the membrane only at the NS2B N-terminus . The present work shows that , in solution , the closed conformation ( or a closely related conformation ) is predominantly populated even in the absence of inhibitors . The strongest evidence for this conclusion comes from the similarity of the 15N-relaxation with and without inhibitor and from the close similarity between the paramagnetic relaxation enhancements of residues 147–170 with and without inhibitor ( Figure 5A ) . In addition , the PREs indicate that the open conformations detected by the spin-labeling experiment are mostly unrelated to those found in the crystal structures of the WNV and dengue NS2B-NS3 proteases . As PREs strongly depend on the distance from the spin label , they enable the detection of minor conformational species which may be populated by as little as 1% . The exchange broadening observed for many NMR resonances is also in agreement with the closed conformation as the major species . Chemical exchange of a major conformational species with one or several minor species populated by as little as 5% can lead to the disappearance of cross-peaks if the exchange rate is comparable to the difference in chemical shifts of the conformational states . The appearance of all cross-peaks in the presence of inhibitors can thus be explained , if the inhibitor shifts the conformational equilibrium towards a single conformation ( populated by at least 90% ) . The alternative explanation of a greatly accelerated exchange rate leading to recovery of the cross-peaks is unlikely , as it is difficult to imagine a mechanism by which a low-molecular weight inhibitor would accelerate a major conformational exchange process in the protein . Considering that inhibitors 1 and 2 do not form van der Waals contacts with NS2B [14] , [18] , the shift in equilibrium may result from attractive electrostatic interactions with the β-hairpin of the NS2B CTS that carries a sequence of three Asp residues ( Asp80NS2B-Asp82NS2B ) . Available structures show that inhibitors 1 and 2 [18] , BPTI [6] and peptide inhibitors [7] , [8] all project positively charged groups towards the β-hairpin of the NS2B CTS . The importance of electrostatic interactions for enzymatic activity of the WNV NS2B-NS3 protease is supported by the observation that increasing concentrations of NaCl decrease enzymatic activity [16] . All available observations indicate that the structural association of the NS2B CTS with NS3pro is not an artifact of the covalent linker connecting the C-terminus of NS2B with the N-terminus of NS3 . ( i ) R2 relaxation rates show that at least ten of the linker residues are highly mobile regardless of the presence or absence of inhibitor . ( ii ) NMR spectra of wt NS2B-NS3 without this covalent link showed no evidence of increased flexibility of the NS2B CTS regardless of the presence of inhibitor ( data not shown ) . ( iii ) The association of the NS2B CTS with NS3pro is structurally conserved between all three crystal structures of the WNV NS2B-NS3 protease in complexes with different inhibitors , independent of the presence or absence of the linker and for different linker lengths [6]–[8] . All three structures shown in Figure 1 were crystallized with uncleaved linker peptides present , illustrating the freedom of the NS2B CTS to assume the different conformations consistently despite the linker . ( iv ) Shortening of the linker sequence between NS2B and NS3 by one Gly residue has been shown not to affect the proteolytic activity of the WNV protease [5] . ( v ) The construct of the DENV NS2B-NS3 homologue used previously for crystallization [7] contained the same number of residues and the Gly4-Ser-Gly4 segment as the linker peptide in the WNV NS2B-NS3pro construct without preventing enhanced flexibility of the NS2B CTS ( see below ) . Comparison of the present results with previous NMR data of a 15N/13C-Ile labeled sample of the dengue virus type 2 NS2B-NS3 protease [25] reveals an unexpected important difference . In the case of the DENV protease , in the absence of inhibitor , the 15N-HSQC cross-peaks of all five isoleucine residues spanning NS2B from Ile67 to Ile86 appeared at random coil chemical shifts and displayed much narrower line widths than any isoleucine residue in the structured core of NS3pro [25] . The absence of any line broadening in this segment indicates that it is more than 90% of the time dissociated from NS3pro . Nonetheless , the NS2B CTS seems to affect the NMR line widths in the rest of the protein , as much more uniform NMR peak intensities have been reported following cleavage of recombinant NS2B-NS3 protease at Asp81NS2B by endoprotease Asp-N [26] . The association of NS2B to NS3 has important implications for structure-based drug design . First , if the NS2B CTS remains largely dissociated also in the presence of inhibitors , this may explain the persistent difficulties to crystallize the dengue virus enzyme with a bound inhibitor . Second , the relatively close association of the NS2B CTS with NS3 in the case of the WNV NS2B-NS3 protease may make it harder to find inhibitors that suppress the protease activity by preventing the association of the NS2B CTS with NS3pro . In contrast , the closed conformation is an appropriate target for rational drug design and has already been used successfully to identify hits by virtual screening [14] , [15] . | Dengue and West Nile virus infections put an estimated 2 . 5 billion people at risk . Neither drugs nor vaccines are currently available against these diseases . The non-structural protein NS3 is a protease that , together with the cofactor NS2B , is essential for viral maturation . The NS2B-NS3 proteases of dengue and West Nile viruses are highly homologous and present promising drug targets . Crystal structures of the West Nile virus protease with and without bound inhibitor revealed large structural differences in NS2B , while no crystal structure of the dengue virus protease could be determined with a bound inhibitor . We investigated the structural change in solution and found that the C-terminal segment ( CTS ) of the NS2B cofactor is prone to dissociation from NS3 . In the case of the West Nile virus protease , the CTS of NS2B is mostly associated with NS3 , especially in the presence of inhibitors . In the case of the dengue virus protease and in the absence of inhibitors , the CTS of NS2B is mostly dissociated from NS3 . Finding drug candidates to inhibit the association of the NS2B cofactor may thus be easier for the dengue virus protease . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"biophysics/experimental",
"biophysical",
"methods",
"virology/viral",
"replication",
"and",
"gene",
"regulation"
] | 2009 | NMR Analysis of the Dynamic Exchange of the NS2B Cofactor between Open and Closed Conformations of the West Nile Virus NS2B-NS3 Protease |
In standard attractor neural network models , specific patterns of activity are stored in the synaptic matrix , so that they become fixed point attractors of the network dynamics . The storage capacity of such networks has been quantified in two ways: the maximal number of patterns that can be stored , and the stored information measured in bits per synapse . In this paper , we compute both quantities in fully connected networks of N binary neurons with binary synapses , storing patterns with coding level , in the large and sparse coding limits ( ) . We also derive finite-size corrections that accurately reproduce the results of simulations in networks of tens of thousands of neurons . These methods are applied to three different scenarios: ( 1 ) the classic Willshaw model , ( 2 ) networks with stochastic learning in which patterns are shown only once ( one shot learning ) , ( 3 ) networks with stochastic learning in which patterns are shown multiple times . The storage capacities are optimized over network parameters , which allows us to compare the performance of the different models . We show that finite-size effects strongly reduce the capacity , even for networks of realistic sizes . We discuss the implications of these results for memory storage in the hippocampus and cerebral cortex .
Attractor neural networks have been proposed as long-term memory storage devices [1] , [2] , [3] . In such networks , a pattern of activity ( the set of firing rates of all neurons in the network ) is said to be memorized if it is one of the stable states of the network dynamics . Specific patterns of activity become stable states thanks to synaptic plasticity mechanisms , including both long term potentiation and depression of synapses , that create positive feed-back loops through the network connectivity . Attractor states are consistent with the phenomenon of selective persistent activity during delay periods of delayed response tasks , which has been documented in numerous cortical areas in behaving monkeys [4] , [5] , [6] , [7] . A long standing question in the field has been the question of the storage capacity of such networks . Much effort has been devoted to compute the number of attractor states that can be imprinted in the synaptic matrix , in networks of binary neurons [8] , [9] , [10] , [11] . Models storing patterns with a covariance rule [12] , [1] , [8] , [11] were shown to be able to store a number of patterns that scale linearly with the number of synapses per neuron . In the sparse coding limit ( in which the average fraction of selective neurons per pattern goes to zero in the large limit ) , the capacity was shown to diverge as . These scalings lead to a network storing on the order of 1 bit per synapse , in the large limit , for any value of the coding level . Elizabeth Gardner [10] computed the maximal capacity , in the space of all possible coupling matrices , and demonstrated a similar scaling for capacity and information stored per synapse . These initial studies , performed on the simplest possible networks ( binary neurons , full connectivity , unrestricted synaptic weights ) were followed by a second wave of studies that examined the effect of adding more neurobiological realism: random diluted connectivity [9] , neurons characterized by analog firing rates [13] , learning rules in which new patterns progressively erase the old ones [14] , [15] . The above mentioned modifications were shown not to affect the scaling laws described above . One particular modification however was shown to have a drastic effect on capacity . A network with binary synapses and stochastic on-line learning was shown to have a drastically impaired performance , compared to networks with continuous synapses [16] , [17] . For finite coding levels , the storage capacity was shown to be on the order of , not stored patterns , while the information stored per synapse goes to zero in the large limit . In the sparse coding limit however ( ) , the capacity was shown to scale as , and therefore a similar scaling as the Gardner bound , while the information stored per synapse remains finite in this limit . These scaling laws are similar to the Willshaw model [18] , which can be seen as a particular case of the Amit-Fusi [17] rule . The model was then subsequently studied in greater detail by Huang and Amit [19] , [20] who computed the storage capacity for finite values of , using numerical simulations and several approximations for the distributions of the ‘local fields’ of the neurons . However , computing the precise storage capacity of this model in the large limit remains an open problem . In this article we focus on a model of binary neurons where binary synapses are potentiated or depressed stochastically depending on the states of pre and post synaptic neurons [17] . We first introduce analytical methods that allow us to compute the storage capacity in the large limit , based on a binomial approximation for the synaptic inputs to the neurons . We first illustrate it on the Willshaw model and to recover the well-known result on the capacity of this model [18] , [21] , [22] . We then move to a stochastic learning rule , in which we study two different scenarios: ( i ) in which patterns are presented only once - we will refer to this model as the SP ( Single Presentation ) model [17]; ( ii ) in which noisy versions of the patterns are presented multiple-times - the MP ( Multiple presentations ) model [23] . For both models we compute the storage capacity and the information stored per synapse in the large limit , and investigate how they depend on the various parameters of the model . We then study finite size effects , and show that they have a huge effect even in networks of tens of thousands of neurons . Finally we show how capacity in finite size networks can be enhanced by introducing inhibition , as proposed in [19] , [20] . In the discussion we summarize our results and discuss the relevance of the SP and MP networks to memory maintenance in the hippocampus and cortex .
The capacity of the Willshaw model has already been studied by a number of authors [18] , [21] , [22] . Here , we present the application of the analysis described in the previous section to the Willshaw model , for completeness and comparison with the models described in the next sections . In this model , after presenting patterns to the network , the synaptic matrix is described as follows: if at least one of the presented patterns had neuron and co-activated , otherwise . Thus , after the learning phase , we have , ( 21 ) Saturating the inequalities ( 19 , 20 ) with fixed , one obtains the information stored per synapse , ( 22 ) The information stored per synapse is shown as a function of in Figure 1a . A maximum is reached for at , but goes to zero in both the and limits . The model has a storage capacity comparable to its maximal value , in a large range of values of ( between and ) . We can also optimize capacity for a given value of , as shown in Figure 1b . It reaches its maximum at , and goes to zero in the small and large limits . Again , the model has a large storage capacity for a broad range of , for between and . Previous studies [18] , [21] have found an optimal capacity of . Those studies focused on a feed-forward network with a single output neuron , with no fluctuations in the number of selective neurons per pattern , and required that the number of errors on silent outputs is of the same order as the number of selective outputs in the whole set of patterns . In the calculations presented here , we have used a different criteria , namely that a given pattern ( not all patterns ) is exactly a fixed point of the dynamics of the network with a probability that goes to one in the large limit . Another possible definition would be to require that all the patterns are exact fixed points with probability one . In this case , for patterns with fixed numbers of selective neurons , the capacity drops by a factor of , , as already computed by Knoblauch et al [22] . A drawback of the Willshaw learning rule is that it only allows for synaptic potentiation . Thus , if patterns are continuously presented to the network , all synapses will eventually be potentiated and no memories can be retrieved . In [17] Amit and Fusi introduced a new learning rule that maintains the simplicity of the Willshaw model , but allows for continuous on-line learning . The proposed learning rule includes synaptic depression . At each learning time step , a new pattern with coding level is presented to the network , and synapses are updated stochastically: for synapses such that : if , then is potentiated to 1 with probability ; and if it stays at . for synapses such that : if , then stays at ; and if it is depressed to with probability . The evolution of a synapse during learning can be described by the following Markov process: ( 23 ) where is the probability that a silent synapse is potentiated upon the presentation of pattern and is the probability that a potentiated synapse is depressed . After a sufficient number of patterns has been presented the distribution of synaptic weights in the network reaches a stationary state . We study the network in this stationary regime . For the information capacity to be of order 1 , the coding level has to scale as , as in the Willshaw model , and the effects of potentiation and depression have to be of the same order [17] . Thus we define the depression-potentiation ratio as , ( 24 ) We can again use Eq . ( 9 ) and the saturated inequalities ( 19 , 20 ) to compute the maximal information capacity in the limit . This requires computing and , defined in the previous section , as a function of the different parameters characterizing the network . We track a pattern that has been presented time steps in the past . In the following we refer to as the age of the pattern . In the sparse coding limit , corresponds to the probability that a synapse is potentiated . It is determined by the depression-potentiation ratio , ( 25 ) and ( 26 ) where . Our goal is to determine the age of the oldest pattern that is still a fixed point of the network dynamics , with probability one . Note that in this network , contrary to the Willshaw model in which all patterns are equivalent , here younger patterns , of age , are more strongly imprinted in the synaptic matrix , , and thus also stored with probability one . Choosing an activation threshold and a coding level that saturate inequalities ( 19 ) and ( 20 ) , information capacity can be expressed as: ( 27 ) The optimal information is reached for which gives . The dependence of on the different parameters is shown in Figure 2 . Panel a shows the dependence on the fraction of activated synapses in the asymptotic learning regime . Panels b , c and d show the dependence on , and . Note from panel c that there is a broad range of values of that give information capacities similar to the optimal one . One can also observe that the optimal information capacity is about times lower in the SP model than in the Willshaw model . This is the price one pays to have a network that is able to continuously learn new patterns . However , it should be noted that at maximal capacity , in the Willshaw model , every pattern has a vanishing basin of attraction while in the SP model , only the oldest stable patterns have vanishing basins of attraction . This feature is not captured by our measure of storage capacity . In the SP model , patterns are presented only once . Brunel et al [23] studied the same network of binary neurons with stochastic binary synapses but in a different learning context , where patterns are presented multiple times . More precisely , at each learning time step , a noisy version of one of the prototypes is presented to the network , ( 28 ) Here is a noise level: if , presented patterns are identical to the prototypes , while if , the presented patterns are uncorrelated with the prototypes . As for the SP model this model achieves a finite non-zero information capacity in the large limit if the depression-potentiation ratio is of order one , and if the coding level scales with network size as . If learning is slow , , and the number of presentations of patterns of each class becomes large the probabilities and are [23]: ( 29 ) and ( 30 ) We inserted those expressions in Eqs . ( 19 , 20 ) to study the maximal information capacity of the network under this learning protocol . The optimal information bits/synapse is reached at for which gives . In this limit , the network becomes equivalent to the Willshaw model . The maximal capacity is about times larger than for a network that has to learn in one shot . On Figure 3a we plot the optimal capacity as a function of . The capacity of the slow learning network with multiple presentations is bounded by the capacity of the Willshaw model for all values of , and it is reached when the depression-potentiation ratio . For this value , no depression occurs during learning: the network loses palimpsest properties , i . e . the ability to erase older patterns to store new ones , and it is not able to learn if the presented patterns are noisy . The optimal capacity decreases with , for instance at ( as many potentiation events as depression events at each pattern presentation ) , . Figure 3c shows the dependence as a function of . In Figure 3d , we show the optimized capacity for different values of the noise in the presented patterns . This quantifies the trade-off between the storage capacity and the generalization ability of the network [23] . The results we have presented so far are valid for infinite size networks . Finite-size effects can be computed for the three models we have discussed so far ( see Methods ) . The main result of this section is that the capacity of networks of realistic sizes is very far from the large N limit . We compute capacities for finite networks in the SP and MP settings , and we validate our finite size calculations by presenting the results of simulations of large networks of sizes , . We summarize the finite size calculations for the SP model ( a more general and detailed analysis is given in Methods ) . In the finite network setting , conditional on the tested pattern having selective neurons , the probability of no error is given bywith ( 31 ) where and is given by Eq . ( 13 ) . In the calculations for discussed in the previous sections we kept only the dominant term in , which yields Eqs . ( 19 ) and ( 20 ) . In the above equations , the first order corrections scale as , which has a dramatic effect on the storage capacity of finite networks . In Figure 4a , b , we plot ( where the bar denotes an average over the distribution of ) as a function of the age of the pattern , and compare this with numerical simulations . It is plotted for and for learning and network parameters chosen to optimize the storage capacity of the infinite-size network ( see Section ‘Amit-Fusi model’ ) . We show the result for two different approximations of the field distribution: a binomial distribution ( magenta ) , as used in the previous calculations for infinite size networks; and a gaussian ( red ) approximation ( see Methods for calculations ) as used by previous authors [19] , [20] , [24] . For these parameters the binomial approximation gives an accurate estimation of , while the gaussian calculation overestimates it . The curves we get are far from the step functions predicted for by Eq . ( 45 ) . To understand why , compare Eqs . ( 15 ) , and ( 31 ) : finite size effects can be neglected when and . Because the finite size effects are of order , it is only for huge values of that the asymptotic capacity can be recovered . For instance if we choose an activation threshold slightly above the optimal threshold given in Section ‘Amit-Fusi model’ ( ) , then , and for we only have . In Figure 4c we plot as a function of where is the value of that optimizes capacity in the large limit , and the other parameters are the one that optimizes capacity . We see that we are still far from the large limit for . Networks of sizes have capacities which are only between 20% and 40% of the predicted capacity in the large limit . Neglecting fluctuations in the number of selective neurons , we can derive an expression for the number of stored patterns that includes the leading finite size correction for the SP model , ( 32 ) where and are two constants ( see Methods ) . If we take fluctuations in the number of selective neurons into account , it introduces other finite-size effects as can be seen from Eqs . ( 43 ) and ( 44 ) in the Methods section . These fluctuations can be discarded if and . In Figure 4d we plot for different values of N . We see that finite size effects are even stronger in this case . To plot the curves of Figure 4 , we chose parameters to be those that optimize storage capacity for infinite network sizes . When is finite , those parameters are no longer optimal . To optimize parameters at finite , since the probability of error as a function of age is no longer a step function , it is not possible to find the last pattern stored with probability one . Instead we define the capacity as the pattern age for which . Using Eqs . ( 31 ) and performing an average over the distribution of , we find parameters optimizing pattern capacity for fixed values of . Results are shown on Figure 5a , b for and . We show the results for the different approximations used to model the neural fields: the blue line is the binomial approximation , the cyan line the gaussian approximation and the magenta one is a gaussian approximation with a covariance term that takes into account correlations between synapses ( see Methods and [19] , [20] ) . For the storage capacity of simulated networks ( black crosses ) is well predicted by the binomial approximation while the gaussian approximations over-estimates capacity . For , the correlations between synapses can no longer be neglected [17] . The gaussian approximation with covariance captures the drop in capacity at large . For , the SP model can store a maximum of patterns at a coding level ( see blue curve in figure 5c ) . As suggested in Figures 4c , d , the capacity of finite networks is strongly reduced compare to the capacity predicted for infinite size networks . More precisely , if the network of size had the same information capacity as the infinite size network ( 27 ) , it would store up to patterns at coding level . Part of this decrease in capacity is avoided if we consider patterns that have a fixed number of selective neurons . This corresponds to the red curve in figure 4c . For fixed sizes the capacity is approximately twice as large . Note that finite-size effects tend to decrease as the coding level increases . In Figure 5c , , and the capacity is of the value predicted by the large limit calculation . The ratio of actual to asymptotic capacities increases to at and at . In Figure 5d , we do the same analysis for the MP model with . Here we have also optimized all the parameters , except for the depression-potentiation ratio which is set to , ensuring that the network has the palimpsest property and the ability to deal with noisy patterns . For , the MP model with can store up to patterns , at ( versus at for the SP model ) . One can also compute the optimized capacity for a given noise level . At , for and or at , for and . So far , we have defined the storage capacity as the number of patterns that can be perfectly retrieved . However , it is quite common for attractor neural networks to have stable fixed point attractors that are close to , but not exactly equal to , patterns that are stored in the connectivity matrix . It is difficult to estimate analytically the stability of patterns that are retrieved with errors as it requires analysis of the dynamics at multiple time steps . We therefore used numerical simulations to check whether a tested pattern is retrieved as a fixed point of the dynamics at a sufficiently low error level . To quantify the degree of error , we introduce the overlap between the network fixed point and the tested pattern , with selective neurons ( 33 ) In Figure 6a we show , the number of fixed-point attractors that have an overlap larger than with the corresponding stored pattern , for , and . Note that only a negligible number of tested patterns lead to fixed points with smaller than , for neurons . Considering fixed points with errors leads to a substantial increase in capacity , e . g . for the capacity increases from to . In Figure 6b , we quantify the information capacity in bits stored per synapse , defined as in Eq . ( 6 ) , . Note that in the situation when retrieval is not always perfect this expression is only an approximation of the true information content . The coding level that optimizes the information capacity in bits per synapse is larger ( ) than the one that optimizes the number of stored patterns ( ) , since the information content of individual patterns decreases with . Finally , note that the information capacity is close to its optimum in a broad range of coding levels , up to . As we have seen above , the fluctuations in the number of selective neurons in each pattern lead to a reduction in storage capacity in networks of finite size ( e . g . Figure 5c , d ) . The detrimental effects of these fluctuations can be mitigated by adding a uniform inhibition to the network [19] . Using a simple instantaneous and linear inhibitory feed-back , the local fields become ( 34 ) For infinite size networks , adding inhibition does not improve storage capacity since fluctuations in the number of selective neurons vanish in the large N limit . However , for finite size networks , minimizing those fluctuations leads to substantial increase in storage capacity . When testing the stability of pattern , if the number of selective neurons is unknown , the variance of the field on non-selective neurons is , and for selective neurons ( for small ) . The variance for non-selective neurons is minimized if , yielding the variance obtained with fixed size patterns . The same holds for selective neurons at . Choosing a value of between and brings the network capacity towards that of fixed size patterns . In Figure 7a , we show the storage capacity as a function of for these three scenarios . Optimizing the inhibition increases the maximal capacity by ( green curve ) compared to a network with no inhibition ( blue curve ) . Red curve is the capacity without pattern size fluctuations . Inhibition increases the capacity from at to . In Figure 7b , information capacity measured in bits per synapse is shown as a function of in the same three scenarios . Note again that for , the capacity is quite close to the optimal capacity .
We have presented an analytical method to compute the storage capacity of networks of binary neurons with binary synapses in the sparse coding limit . When applied to the classic Willshaw model , in the infinite limit , we find a maximal storage capacity of , the same than found in previous studies , although with a different definition adapted to recurrent networks , as discussed in the section ‘Willshaw model’ . We then used this method to study the storage capacity of a network with binary synapses and stochastic learning , in the single presentation ( SP ) scenario [17] . The main advantage of this model , compared to the Willshaw model , is its palimpsest property , that allows it to do on-line learning in an ever changing environment . Amit and Fusi showed that the optimal storage capacity was obtained in the sparse coding limit , and with a balance between the effect of depression and potentiation . The storage capacity of this network has been further studied for finite size networks in [19] , [20] . We have complemented this work by computing analytically the storage capacity in the large limit . The optimal capacity of the SP model is , which is about times lower than the one of the Willshaw model . This decrease in storage capacity is similar to the decrease seen in palimpsest networks with continuous synapses - for example , in the Hopfield model the capacity is about , while in a palimpsest version the capacity drops to about . The reason for this decrease is that the most recently seen patterns have large basins of attraction , while older patterns have smaller ones . In the Willshaw model , all patterns are equivalent , and therefore they all have vanishing basins of attraction at the maximal capacity . We have also studied the network in a multiple presentation ( MP ) scenario , with in which patterns presented to the network are noisy versions of a fixed set of prototypes , in the slow learning limit in which transition probabilities go to zero [23] . In the extreme case in which presented patterns are the prototypes , all synaptic weights are initially at zero , and if the synapses do not experience depression , this model is equivalent to the Willshaw model with a storage capacity of , which is about times larger than the capacity of the SP model . A more interesting scenario is when depression is present . In this case then the network has generalization properties ( it can learn prototypes from noisy versions of them ) , as well as palimpsest properties ( if patterns drawn from a new set of prototypes are presented it will eventually replace a previous set with the new one ) . We have quantified the trade-off between generalization and storage capacity ( see Figure 3d ) . For instance , if the noisy patterns have of their selective neurons in common with the prototypes to be learned , the storage capacity is decreased from to . A key step in estimating storage capacity is deriving an accurate approximation for the distribution of the inputs neurons receive . These inputs are the sum of a large number of binary variables , so the distribution is a binomial if one can neglect the correlations between these variables , induced by the learning process . Amit and Fusi [17] showed that these correlations can be neglected when . Thus , we expect the results with the binomial approximation to be exact in the large limit . We have shown that a Gaussian approximation of the binomial distribution gives inaccurate results in the sparse coding limit , because the capacity depends on the tail of the distribution , which is not well described by a Gaussian . For larger coding levels ( ) , the binomial approximation breaks down because it does not take into account correlations between inputs . Following [19] and [20] , we use a Gaussian approximation that includes the covariance of the inputs , and show that this approximation captures well the simulation results in this coding level range . We computed storage capacities for two different learning scenarios . Both are unsupervised , involve a Hebbian-type plasticity rule , and allow for online learning ( providing patterns are presented multiple times for the MP model ) . It is of interest to compare the performance of these two particular scenarios with known upper bounds on storage capacity . For networks of infinite size with binary synapses such a bound has been derived using the Gardner approach [25] . In the sparse coding limit , this bound is with random patterns ( in which fluctuations in the number of selective neurons per pattern fluctuates ) , and if patterns have a fixed number of selective neurons [26] . We found a capacity of for the SP model and for the MP model , obtained both for patterns with fixed and variable number of selective neurons . The result for the MP model seems to violate the Gardner bound . However , as noticed by Nadal [21] , one should be cautious in comparing these results: in our calculations we have required that a given pattern is stored perfectly with probability one , while the Gardner calculation requires that all patterns are stored perfectly with probability one . As mentioned in the section ‘Willshaw model’ , the capacity of the Willshaw and MP models drops to in the case of fixed-size patterns , if one insists that all patterns should be stored perfectly , which is now consistent with the Gardner bound . This means that the MP model is able to reach a capacity which is roughly half the Gardner bound , a rather impressive feat given the simplicity of the rule . Note that supervised learning rules can get closer to these theoretical bounds [27] . We have also studied finite-size networks , in which we defined the capacity as the number of patterns for which the probability of exact retrieval is at least 50% . We found that networks of reasonable sizes have capacities that are far from the large limit . For networks of sizes storage capacities are reduced by a factor or more ( see Figure 4 ) . These huge finite size effects can be understood by the fact that the leading order corrections in the large limit are in - and so can never be neglected unless is an astronomical number ( see Methods ) . A large part of the decrease in capacity when considering finite-size networks is due to fluctuations in the number of selective neurons from pattern to pattern . In the last section , we have used inhibition to minimize the effect of these fluctuations . For instance , for a network of neurons learning in one shot , inhibition allows to increase capacity from to . For finite size networks , memory patterns that are not perfectly retrieved can still lead to fixed points where the activity is significantly correlated with the memory patterns . We have investigated with simulations how allowing errors in the retrieved patterns modifies storage capacity . For , the capacity increases from to , i . e . by approximately 30% . Our study focused on networks of binary neurons , connected through binary synapses , and storing very sparse patterns . These three assumptions allowed us to compute analytically the storage capacity of the network in two learning scenarios . An important question is how far real neural networks are from such idealized assumptions . First , the issue of whether real synapses are binary , discrete but with a larger number of states , or essentially continuous , is still unresolved , with evidence in favor of each of these scenarios [28] , [29] , [30] , [31] , [32] . We expect that having synapses with a finite number of states will not modify strongly the picture outlined here [17] , [33] , [20] . Second , it remains to be investigated how these results will generalize to networks of more realistic neurons . In strongly connected networks of spiking neurons operating in the balanced mode [34] , [35] , [36] , [37] , the presence of ongoing activity presents strong constraints on the viability of sparsely coded selective attractor states . This is because ‘non-selective’ neurons are no longer silent , but are rather active at low background rates , and the noise due to this background activity can easily wipe out the selective signal [35] , [38] . In fact , simple scaling arguments in balanced networks suggest the optimal coding level would become [3] , [39] . The learning rules we have considered in this paper lead to a vanishing information stored per synapse with this scaling . Finding an unsupervised learning rule that achieves a finite information capacity in the large limit in networks with discrete synapses for such coding levels remains an open question . However , the results presented here show that for networks of realistic sizes , the information capacity at such coding levels is in fact not very far from the optimal one that is reached at lower coding levels ( see vertical lines in Figure 5–7 ) . Finally , the coding levels of cortical networks during delay period activity remain poorly characterized . Experiments in IT cortex [40] , [41] , [42] are consistent with coding levels of order 1% . Our results indicate that in networks of reasonable sizes , these coding levels are not far from the optimal values . The SP and MP models investigated in this paper can be thought of as minimal models for learning in hippocampus and neocortex . The SP model bears some resemblance to the function of hippocampus , which is supposed to keep a memory of recent episodes that are learned in one shot , thanks to highly plastic synapses . The MP model relates to the function of neocortex , where a longer-term memory can be stored , thanks to repeated presentations of a set of prototypes that occur repeatedly in the environment , and perhaps during sleep under the supervision of the hippocampus . The idea that hippocampal and cortical networks learn on different time scales has been exploited in several modeling studies [43] , [44] , [45] , in which the memories are first stored in the hippocampus and then gradually transferred to cortical networks . It would be interesting to extend the type of analysis presented here to coupled hippocampo-cortical networks with varying degrees of plasticity .
We are interested at retrieving pattern that has been presented during the learning phase . We set the network in this state and ask whether the network remains in this state while the dynamics ( 2 ) is running . At the first iteration , each neuron is receiving a field ( 35 ) Where M+1 is the number of selective neurons in pattern , with . Where we use the standard ‘Landau’ notations: means that goes to a finite limit in the large limit , while means that goes to zero in the large limit . and . We recall that and . Thus is a binary random variable which is with probability , either if is a selective neuron ( sites such that ) , or if is a non-selective neuron ( sites such that ) . Neglecting correlations between and ( it is legitimate in the sparse coding limit we are interested in , see [17] ) , the 's are independent and the distribution of the field on selective neurons can be written as ( 36 ) where we used Stirling formula for , with defined in ( 13 ) . For non-selective neurons ( 37 ) Now write ( 38 ) In the limit we are considering in this section , and if , the sums corresponding to the probabilities are dominated by their first term ( corrections are made explicit in the following section ) . Keeping only higher order terms in in Eqs . ( 36 ) and ( 37 ) , we have: ( 39 ) and ( 40 ) yielding Eq . ( 15 ) with . Note that with the coding levels we are considering here ( ) , is of order . When the number of selective neurons per pattern is fixed at , we choose for the activation threshold and these equations become: ( 41 ) where For random numbers of selective neurons we need to compute the average over : . Since is distributed according to a binomial of average and variance , for sufficiently large , this can be approximated as where is normally distributed: ( 42 ) with ( 43 ) and ( 44 ) When goes to infinity , we bring the limit into the integral in Eq . ( 42 ) and obtain ( 45 ) where is the Heaviside function . Thus in the limit of infinite size networks , the probability of no error is a step function . The first Heaviside function implies that the only requirement to avoid errors on selective neurons is to have a scaled activation threshold below . The second Heaviside function implies that , depending on , has to be chosen far enough from . The above equation allows to derive the inequalities ( 19 ) and ( 20 ) . We now turn to a derivation of finite-size corrections for the capacity . Here we show two different calculations . In the first calculation , we derive Eq . ( 32 ) , taking into account the leading-order correction term in Eq . ( 43 ) . This allows us to compute the leading-order correction to the number of patterns that can be stored for a given set of parameters . However , it does not predict accurately the storage capacity of the large-size but finite networks that we simulated . In the second calculation presented , we focus on computing the probability of no error in a given pattern , including a next-to-leading-order correction . Eq . ( 32 ) is derived for a fixed set of parameters , assuming that the set of active neurons have a fixed size , and that the activation threshold has been chosen large enough such that the probability to have non-selective neurons activated is small . From the Stirling expansion , adding the first finite-size correction term in Eq . ( 41 ) , we get ( 46 ) with . For large , the number of stored patterns can be increased until . Setting , an expansion of in allows to write ( 47 ) The patterns are correctly stored as long as . This condition is satisfied for . For the SP model , we can deduce which value of yields this value of ( see Eq . ( 26 ) ) . This allows to derive Eq . ( 32 ) , ( 48 ) We now turn to a calculation of the probability of no error on a given pattern , taking into account the next-to-leading order correction of order one , in addition to the term of order in Eq . ( 41 ) . This is necessary to predict accurately the capacity of realistic size networks ( for instance for , ) . is computed for a memory pattern with selective neurons . The estimation of used in the figures is obtained by averaging over different values of , with drawn from a binomial distribution of mean . We first provide a more detailed expansion of the sums in Eq . ( 38 ) . Setting , with the Taylor expansions: ( 49 ) ( 50 ) where and . Using ( 37 ) we can rewrite: ( 51 ) In the cases we consider , we will always have so that we can consider only the term of order in . The sum is now geometric , and we obtain ( 52 ) The same kind of expansion can be applied for the selective neurons . Again if we are in a situation where ( 53 ) When is close to and thus , we are then left with: ( 54 ) ( 55 ) When is too close to , which is the case for the optimal parameters in the large limit , we need to use ( 55 ) . It only contributes a term of order in and does not modify our results . In Figures 6-7 , we use ( 53 ) , which gives from ( 38 ) and ( 36 ) , ( 37 ) and ( 53 ) , ( 52 ) : ( 56 ) ( 57 ) The probability of no error is ( 58 ) which leads to Eqs . ( 31 ) For a fixed number of selective neurons in pattern , approximating the distribution of the fields on background neurons and selective neurons with a gaussian distribution gives: ( 59 ) where ( 60 ) and ( 61 ) where ( 62 ) The probability that those fields are on the wrong side of the threshold are: ( 63 ) and ( 64 ) Following the same calculations presented , and keeping only terms that are relevant in the limit , the probability that there is no error is given by: ( 65 ) where the rate function is ( 66 ) Calculations with the binomial versus the gaussian approximation differ only in the form of . Finite size terms can be taken into account in the same way it is done in the previous Methods section for the binomial approximation . In all above calculations we assumed that fields are sums of independent random variables ( 35 ) . For small correlations are negligible [17] , [19] . It is possible to compute the covariances between the terms of the sum ( see Eq . ( 3 . 9 ) in [19] ) , and take them into account in the gaussian approximation . This can be done using ( 67 ) ( 68 ) in Eqs . ( 59 ) , ( 61 ) , where ( 69 ) | Two central hypotheses in neuroscience are that long-term memory is sustained by modifications of the connectivity of neural circuits , while short-term memory is sustained by persistent neuronal activity following the presentation of a stimulus . These two hypotheses have been substantiated by several decades of electrophysiological experiments , reporting activity-dependent changes in synaptic connectivity in vitro , and stimulus-selective persistent neuronal activity in delayed response tasks in behaving monkeys . They have been implemented in attractor network models , that store specific patterns of activity using Hebbian plasticity rules , which then allow retrieval of these patterns as attractors of the network dynamics . A long-standing question in the field is how many patterns ( or equivalently , how much information ) can be stored in such networks ? Here , we compute the storage capacity of networks of binary neurons and binary synapses . Synapses store information according to a simple stochastic learning process that consists of transitions between synaptic states conditioned on the states of pre- and post-synaptic neurons . We consider this learning process in two limits: a one shot learning scenario , where each pattern is presented only once , and a slow learning scenario , where noisy versions of a set of patterns are presented multiple times , but transition probabilities are small . The two limits are assumed to represent , in a simplified way , learning in the hippocampus and neocortex , respectively . We show that in both cases , the information stored per synapse remains finite in the large limit , when the coding is sparse . Furthermore , we characterize the strong finite size effects that exist in such networks . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"neural",
"networks",
"computational",
"neuroscience",
"biology",
"and",
"life",
"sciences",
"computational",
"biology",
"neuroscience",
"learning",
"and",
"memory",
"coding",
"mechanisms"
] | 2014 | Memory Capacity of Networks with Stochastic Binary Synapses |
Homeostatic maintenance of tissues is orchestrated by well tuned networks of cellular signaling . Such networks regulate , in a stochastic manner , fates of all cells within the respective lineages . Processes such as symmetric and asymmetric divisions , differentiation , de-differentiation , and death have to be controlled in a dynamic fashion , such that the cell population is maintained at a stable equilibrium , has a sufficiently low level of stochastic variation , and is capable of responding efficiently to external damage . Cellular lineages in real tissues may consist of a number of different cell types , connected by hierarchical relationships , albeit not necessarily linear , and engaged in a number of different processes . Here we develop a general mathematical methodology for near equilibrium studies of arbitrarily complex hierarchical cell populations , under regulation by a control network . This methodology allows us to ( 1 ) determine stability properties of the network , ( 2 ) calculate the stochastic variance , and ( 3 ) predict how different control mechanisms affect stability and robustness of the system . We demonstrate the versatility of this tool by using the example of the airway epithelium lineage . Recent research shows that airway epithelium stem cells divide mostly asymmetrically , while the so-called secretory cells divide predominantly symmetrically . It further provides quantitative data on the recovery dynamics of the airway epithelium , which can include secretory cell de-differentiation . Using our new methodology , we demonstrate that while a number of regulatory networks can be compatible with the observed recovery behavior , the observed division patterns of cells are the most optimal from the viewpoint of homeostatic lineage stability and minimizing the variation of the cell population size . This not only explains the observed yet poorly understood features of airway tissue architecture , but also helps to deduce the information on the still largely hypothetical regulatory mechanisms governing tissue turnover , and lends insight into how different control loops influence the stability and variance properties of cell populations .
All tissues and organs in our bodies can be deconstructed and arranged into phylogenetic cellular lineages . At the base of every lineage lie stem cells ( SCs ) , the long lasting , self-renewing and generally non-differentiated cell type . Progeny of SCs progressively reduce their proliferative potential and concomitantly acquire specialized differentiated characteristics and novel functions . Typically , fully differentiated cells are post-mitotic and have limited life span , and thus require to be constantly replenished from the SC compartment . Proper steady-state maintenance of the lineages , as well as their rapid responses to cellular loss or excessive expansion require checks and balances at all steps of lineage progression , from stem to terminally differentiated cells . Significant advances in our understanding of the SC biology , as well as high potential for SC modulation as a therapeutic solution to a broad range of regenerative disorders , from non-healing wounds to rapid tumor growth [1–4] , have inspired a lot of theoretical work in the field of lineage regulation . The focus of the present study is understanding control networks involved in the homeostasis of healthy tissues . For a given , two- or multi-compartment lineage system , the control of cellular decisions , such as division and death timing , or division type , can be mediated by feedback loops that depend on the current state of cellular population ( s ) , more precisely , on the relative numbers of distinct cell types within the lineage . For example , the decision for a SC to proliferate can depend on whether there is a deficiency either in the SC compartment , or in other downstream compartment ( s ) . Similarly , the decision for a non-SC progenitor to terminally differentiate could depend on the current number of other terminally differentiated cells . In addition to proliferation and differentiation , other cellular events include asymmetric cell divisions , de-differentiation , and apoptosis . Cell numbers change as the result of divisions and deaths . How can the cell lineage system as a whole be regulated to remain at a near-equilibrium ? Several cell populations can participate in signaling , and control loops can be both positive and negative , to regulate , in a self-correcting way , the rate at which all of these processes take place . Given a complex system of this kind , we need to be able to evaluate whether the control network is capable of producing stable homeostasis , quantify the magnitude of variances resulting from perturbations , and assess the robustness of the stochastic lineage turnover . In [5] we considered stochastic dynamics of cellular lineages in a two-compartment model , which included SCs and one type of differentiated cells . We assumed that in such prototypical lineage only three cellular events took place: ( i ) death of a differentiated cell , ( ii ) proliferation or , alternatively , ( iii ) differentiation of a SC . While valuable , this approach has limitations because it only allows two cell types and three processes in the system . More recently , we showed that such two-compartment model can be sufficient to faithfully describe and predict cellular behaviors in relatively simple lineages , such as mammalian epidermis [6] . Considering the value of this methodology , it is important to generalize it and make it applicable for studying a larger class of more complex cellular lineages . Examples of complex lineages are numerous . Commonly , there are multiple intermediate proliferating cell types , sometimes referred to as transit amplifying cells , between SCs and terminally differentiated post-mitotic cells . Such intermediate progenitors are prominent in the hematopoietic [7 , 8] , intestinal epithelium [9 , 10] and hair follicle epithelium lineages [11–16] . For example , in the hematopoietic lineage , bona fide hematopoietic SCs give rise to common lymphoid and common myeloid progenitors . The latter , in turn , produce granulocyte-macrophage and megakaryocyte–erythroid progenitors [7 , 8] . Moreover , lineages often contain more than one type of SCs and more then one distinct type of differentiated cells . For example , there are two principal types of epithelial SCs in the intestinal epithelium , rapidly proliferating crypt base columnar SCs and quiescent +4 SCs [9 , 10 , 17] . There are also seven distinct differentiated cell types that these SCs can produce: absorptive enterocytes , enteroendocrine cells , Tuft cells , Goblet cells , Paneth cells , M-cells and cup cells [9 , 18] . In lineages with more than one SC type , there is often SC-to-SC interchangeability . For instance , crypt base columnar SCs and quiescent +4 SCs in the intestine can interconvert , depending on the conditions—crypt base columnar SCs are sensitive to damage , become largely depleted after irradiation and then restore from radiation-resistant +4 SCs [19–21] . In addition , following depletion , cells can be replenished from the non-SC progenitors via the so called de-differentiation or reprogramming mechanisms . Such is biliary epithelial cells regeneration via hepatocytes reprogramming in the liver following toxin-induced depletion [22 , 23] . In the lung , alveolar type-2 cells can reprogram into type-1 cells when the latter are selectively ablated by the hyperoxic injury [24] . Similarly , in the stomach , differentiated secretory Troy+ chief cells can de-differentiate into SCs following genetic depletion of the SC compartment [25] . Ideally , a mathematical framework is needed that is not restricted by a small number of cell types , and can handle this biological variety . In this paper we present a theoretical framework that allows to study stability , fluctuations , and robustness of near equilibrium cell dynamics for multi-process , multi-compartment lineages . We obtain analytically , in a general case , ( i ) constraints on the equilibrium rates of all the processes compatible with the existence of a steady state; ( ii ) the stability conditions for the steady state , and ( iii ) solutions for the second moments for all the cell populations . The latter describe comprehensively how different components of the control network affect fluctuations of different cell populations . This versatile mathematical framework , which we call “near equilibrium calculus of stem cells” , allows one to perform computations for arbitrarily complex cell lineages , under any regulatory control network . With this new tool , one can attempt several conceptual types of inquiries . One is explanatory: given an observed pattern ( for example , the symmetry of cell divisions , or the type and direction of control loops observed ) , one can attempt to explain why any particular kind of tissue architecture and cell population management logic have evolved , or , more precisely , evaluate if the given control network and the resulting division patterns are in any sense optimal in the context of stability of the system and robustness of its homeostatic maintenance . The second type of application is predictive: if a network regulating a certain system is unknown ( or not completely understood ) , one can hypothesize what type of a network would be compatible with the given observables and at the same time optimal from the viewpoint of robust homeostatic maintenance . Finally , given a regulatory network , one can evaluate the importance of its different components and the influence they exert on the amount of variance experienced by the cell population . To illustrate the versatility of the method , we apply it to the studies of the airway epithelium system . This system has recently attracted a lot of attention because ( 1 ) its key cell types , including stem cells , are well defined , ( 2 ) it has tractable two-dimensional organization , and ( 3 ) multiple genetic tools have become available to target each of the lineage’s cell types , either to induce cell depletion or gene deletion/mis-expression . In particular , airway epithelium lineage has proven to be a great model system for tracking responses to cell depletion in a semi-quantifiable way—following genetic depletion of a given cellular type , the response of the remaining cells can be precisely measured and tracked in time . Semi-quantitative nature of these recently published experiments provides a plethora of valuable numerical information , which can be modeled . By using our methodology , we were able to incorporate the available data and come up with a set of control networks that are compatible with the observed recovery patterns of the airway epithelium [26–28] . Further , we concerned ourselves with the general question of tissue design . It has been reported in the recent literature [29] that in the three-compartment cellular lineage of the airway epithelium , the SCs are characterized by mostly asymmetric divisions , while the secretory cells ( SecrCs ) , the next cell type in the differentiation hierarchy , are characterized by mostly symmetric divisions . By using the mathematical approach developed here , we show that ( 1 ) predominantly symmetric divisions of the SecrCs is a necessary feature that makes the lineage system compatible with the reportedly slow dynamics of the most differentiated ciliated cells ( CilCs ) , and that ( 2 ) predominantly asymmetric divisions of the SCs may be the consequence of the mathematical fact that asymmetric SC divisions , under the other existing constraints of the airway epithelial system , minimize the fluctuations of both SC and SecrC populations . Our work contributes to the growing computational literature on SC dynamics . Many aspects of SC dynamics have been modeled and studied mathematically . Methodologically , both discrete and continuous computational models have been used , particularly in the context of SC mutagenesis and carcinogenesis [30–41] . In addition to cancer , normal SC behaviors , such as ( i ) symmetry vs . asymmetry of SC divisions , ( ii ) SC quiescence vs . proliferative activation , and ( iii ) progressive lineage specification have been modeled , such as in the hematopoietic system [42–46] . Here , again , both deterministic and stochastic models have been introduced and studied ( see the review in [47] ) . Lineage decision-making controls have been studied deterministically both in the context of minimalistic two-compartment , as well as multi-compartment models [48–52] . Stochastic lineage systems have been considered as well in [53–59] , and feedback regulation of SC dynamics has been modeled in [48 , 48 , 60] . The present approach attempts to generalize the description of SC dynamics in the context of healthy tissue turnover . We strive to create a framework general enough to describe any feasible control network for any hierarchical organization , but at the same time to find a way for analytical understanding of the resulting dynamics , focusing on the role of various control loops in homeostatic maintenance .
The equilibrium is defined by n algebraic equations for the n variables , ( i * 1 , … , i * n ) , which are the equilibrium population sizes of all the compartments: ∑ k = 1 K Q k ( i * 1 , … , i * n ) Δ k i 1 = 0 , … , ∑ k = 1 K Q k ( i * 1 , … , i * n ) Δ k i n = 0 . ( 2 ) If the functional form of all the rates Qk is known , then the equilibria can be determined . In reality , the equilibrium population values can be measured , but the functional form of Qk is unknown . Therefore , it is more useful to interpret eq ( 2 ) as a linear system of equations for the equilibrium rates , which imposes K − n constraints on the rate values . In other words , only K − n out of K rates can be assigned an independent value at the equilibrium . Controls of the different processes combine with the cellular increments to form the Jacobian corresponding to the equilibrium point , J = { a m j } , a m j = ∑ k = 1 K ∂ Q k ∂ i j Δ k i m , 1 ≤ m , j ≤ n , ( 3 ) where the derivatives are assumed to be taken at the equilibrium . It is demonstrated in S1 Text , Section 1 , that the eigenvalues of J inform us not only of the stability of the deterministic equations , but also of the stability of the system involving higher moments . The control loops define how sparse matrix J is . System robustness can be investigated alongside with stability in the following way . Suppose that the control loops are fixed in the sense that we know the topology of the control network ( which cell population controls which process ( es ) ) , and the sign of controls . Let us vary the values of nonzero derivatives Eq ( 1 ) within some bounds . What portion of the set of parameters corresponds to a stable system ? In the most robust scenario , we have a sign-stable matrix J , that is , it is stable for all parameter values of the given signs . In a less robust scenario , only a small portion of parameter space corresponds to stability . While the equilibrium constraints and the Jacobian are obtained by deterministic methods , the next step of the analysis is stochastic . Here we extend the methodology developed in [5] and [61] to describe multi-compartment , multi-process systems . Let us denote by ypq the covariance of the populations in compartments p and q . It is convenient to form the variance vector , y → = ( y 11 , y 12 , … , y n n ) T . Quantities y11 , … , ynn correspond to second central moments , or the variances , of the cell populations . The covariances and the variances can be determined analytically from ( i ) the equilibrium rates , ( ii ) the control values , and ( iii ) the increments associated with the processes . It is convenient to define matrix W as the Kronecker sum W = J ⊕ J ≡ J ⊗ I + I ⊗ J , ( 4 ) where matrix I represents the identity matrix ( see S1 Text , Section 1 for details ) . This matrix contains the information about the controls at the equilibrium . The information about the equilibrium rates and the increments is combined in an n2 × 1 vector s → = ( s 11 , s 12 , … , s n n ) T , which has elements s p q = ∑ k = 1 K Q k * Δ k i p Δ k i q , p , q = 1 , 2 , … , n . ( 5 ) The covariances are then given as solutions of the linear system W y → = - s → . ( 6 ) Because of the special form of the matrix W , eq ( 6 ) is equivalent to the continuous Lyapunov equation , J Y + Y J T = - S . Here , Y is an n × n matrix with elements ypq , J is the Jacobian ( eq ( 3 ) ) , and S is an n × n matrix with elements spq ( eq ( 5 ) ) . This equation arises in the Lyapunov stability theory and several applications of control theory [62 , 63] . Its unique solution Y can be expressed in terms of matrix J in the following way , Y = ∫ 0 ∞ exp ( J t ) S exp ( J T t ) d t , ( 7 ) see e . g . [64] . This integral converges as long as all the eigenvalues of the matrix J have negative real parts . The diagonal elements of the matrix Y give the variance of the cell population numbers . Both types of analysis ( the regular stability analysis and analysis of variance ) share some important features , which is expected . For example , if the real parts of all the eigenvalues become larger ( and negative ) in a certain direction of the parameter space , the variances of all the populations will decrease in the same direction . If on the other hand , different eigenvalues become more “stable” for different parts of the phase space , we expect that variances of different populations might be minimized in different regions of the parameter space . The variance analysis however provides more information . These additional insights are as follows: Therefore we conclude that the study of the variances , while sharing some important features with the usual linear analysis , contributes additional information that can allow us to argue about aspects of tissue design and the functioning of stem cell lineages . In the next section we demonstrate the power of this methodology by using the example of the airway epithelium .
Airway epithelium lines the inner surface of the trachea and bronchi in the lung . It is organized as a two-dimensional sheet of cells sitting on top of the basement membrane . Because all cells are attached to the basal membrane , it is technically a single-layered epithelium . Its lineage consists of three principle cell types: ( 1 ) stem cells ( SCs ) , and two distinct differentiated cell types: ( 2 ) secretory ( SecrCs ) and ( 3 ) ciliated cells ( CilCs ) [65–67] . At homeostasis all cells are distributed at the following approximate stem-to-secretory-to-ciliated ratio: 30%-15%-55% . In terms of lineage control , the available experimental data suggests that most of the control mechanisms are autonomous , i . e . to a significant extent SCs , SecrCs and CilCs regulate each other’s dynamics . Importantly , additional regulatory signals can also come from the fibroblasts and immune cell types located beneath the basal membrane [65] . In this work , however , we will focus on the autonomous lineage controls . Airway epithelium demonstrates the following types of lineage behavior: ( i ) SC quiescence vs . activation , ( ii ) SecrCs differentiation into SecrCs or CilCs , ( iii ) SecrCs de-differentiation into SCs , and ( iv ) trans-differentiation of SecrCs into CilCs . Moreover , it is established that SecrCs can undergo proliferation , while CilCs are considered post-mitotic and their half-life is around 150 days . Most of the above behaviors can be activated upon lineage injury , when one or several cell types are depleted and the lineage repairs toward restoring homeostasis . Below we will outline three distinct previously reported airway epithelium injury experiments , the types of lineage responses that they invoke , as well as the types of regulatory mechanisms that they reveal . Below we will demonstrate the application of the modeling methodology developed here in the context of the airway epithelium regulation . In particular , we will show how the equilibrium rates are constrained , perform the stability analysis , and calculate variances . Analysis of stability and fluctuation magnitudes will allow us to argue about possible control network architectures compatible with the biological observations , and to explain the observed preferences for division symmetries and de-differentiation strategies . In Fig 1 and Table 1 we show eleven cellular processes that can happen in the airway epithelium . Each of these processes results in a change in the abundance of at least one of the three cell types , SCs , SecrCs , and CilCs . Controls are incorporated by assuming that the rate of each of the processes can be influenced by any of the existing population , such that near the equilibrium , Q k ( x , y , z ) ≈ Q k , 0 + Q k x ( x - x 0 ) + Q k y ( y - y 0 ) + Q k z ( z - z 0 ) , ( 8 ) where ( x0 , y0 , z0 ) are the equilibrium numbers of SCs , SecrCs , and CilCs respectively , Qk0 is the rate of process Qk at equilibrium , and quantities Qkx , Qky , Qkz ( which we call “controls” ) are derivatives of this rate with respect to the three population sizes . These three quantities describe how strongly , and in which direction , the intensity of a process changes if each of the populations experiences a fluctuation . A negative value of such a derivative corresponds to a negative control loop . To be precise , paper [29] identified more complexity in the dynamics of SCs in the airway epithelium . It was found that SCs do not divide directly into CilCs or SecrCs . Instead , they create ( by predominantly asymmetric divisions ) a different type of progenitor cell ( called luminal progenitors ) which later mature into SecrCs . Our model combines this into just one step , an asymmetric division into SecrCs . Adding this intermediate step effectively changes the rate of process Q9y and is therefore not implemented . Although each of the controls may be a nontrivial number , we strive to create the simplest model that is compatible with the existing observations . Such a model must include a negative regulation of SC divisions by SecrCs ( fact ( D ) above ) . Further , divisions and de-differentiation of SecrCs is negatively regulated by SCs ( facts ( B ) , ( C ) above ) . Finally , CilCs do not exert any known control over the processes happening in the SC and SecrC compartments ( fact ( A ) ) . We further assumed that the overall rate of CilC death increases with their abundance ( note that this is not a per-cell rate , but the overall intensity of apoptosis ) . Therefore , only some of the derivatives in eq ( 8 ) will be nonzero . We list these possible controls here: Q 1 y , Q 2 y , Q 3 y , Q 4 x , Q 5 x , Q 6 x , Q 8 z , Q 9 y , Q 10 , y , Q 11 , y . ( 9 ) All of these are nonpositive except Q8z , which is nonnegative . Interestingly , not all of the eleven processes make equal contribution to the maintenance of stable homeostasis . In [29] , it was shown that an overwhelming majority of SC divisions are asymmetric , and an overwhelming majority of the SecrC divisions are symmetric . Our first goal is to explain this type of design .
In this work we studied stochastic multi-compartment dynamics of SCs and their lineages . We developed a simple and effective method to mathematically describe any type of multi-compartment lineage system . We could find the analytical results for the expectation and variance of the population of any type of lineage-connected cells , assuming that we know the inverse of a simple deterministic matrix . Furthermore , the stability conditions for the multi-compartment SC dynamics were identified . The general method developed in this paper is applicable for studying a very large class of cellular lineages , and not just simple linear n-compartment models . The technique can naturally include any type of hierarchical or two-way relationships among cells . As described in the Introduction , in most tissues and organs , there are more than just two types of cells , and hierarchical cellular networks are sometimes arranged in a complex nonlinear fashion . Our technique ( and the symbolic algorithm developed ) are capable of handling such systems . The goal is to study the stability and robustness of control networks that maintain homeostasis in such complex systems . We also applied these techniques to interrogate the lineage dynamics in a particular biological system , the mouse airway epithelium . In this system , there are three principal types of cells ( SCs , SecrCs , and CilCs ) that are lineage-connected and can influence each other’s fate decisions . Symmetric and asymmetric divisions , deaths , differentiation and de-differentiation can all take place . There are significant available biological data with regards to the division types and the recovery dynamics in response to injury that take place in the airway epithelium , thus enabling us to compare mathematically derived behaviors with the actual cellular actions . Interestingly , we found that there are multiple ways in which all of the above cellular processes can be mathematically arranged and regulated so that they are stable and compatible with the existing experimental evidence on the lineage recovery dynamics . For example , mathematically , all divisions can be symmetric , or asymmetric , or mixed , and either one of these division types is compatible with the biologically observable lineage dynamics . Yet , recent biological data shows that airway epithelium SCs divide almost exclusively asymmetrically , while SecrCs divide almost exclusively symmetrically . By using the framework developed here , we offer an explanation of these symmetry patterns . It turns out that the predominantly symmetric divisions of SecrCs can be ascribed to the requirement of balance of various cellular processes at equilibrium . At equilibrium , one must expect to have cellular loss ( from say differentiation , de-differentiaion , or death ) to be balanced by cellular gain . In the case of the airway epithelium , we considered the peculiarly slow turnover dynamics of CilCs and derived a balance equation for CilC change . Then , from the requirement that the death rate of CilCs is slow [70] , we deduced that by necessity , SecrCs must divide predominantly symmetrically , to avoid unbalanced accumulation of CilCs . Further we provided an explanation of the predominantly asymmetric divisions of SCs . We considered a system where both symmetric and asymmetric division types for SCs were included , and asked what arrangement of equilibrium division rates will minimize the magnitude of cell number fluctuations . It turned out that strictly asymmetric divisions of SCs comprise the optimal solution for this linear minimization problem under the given biological constraints ( such as positivity of cell numbers and rates ) . Therefore , by using our methodology , we showed that the observed division pattern in the airway epithelium is the only one that is consistent with the steady cell numbers , slow turnover dynamics of the CilCs , and minimal variance of the cell populations at homeostasis . We have also focused on a particular lineage behavior revealed in the recent work suggesting the lack of negative feedback from the differentiated CilCs to SCs or SecrCs following genetic depletion of CilCs [29] . We used the smallest possible number of control loops to study this phenomenon mathematically . We show that minimally parameterized model can robustly mimic the biologically observable slow CilCs recovery dynamics . Furthermore , the same model can robustly mimic quick lineage recovery dynamics when both CilCs and SecrCs are depleted . Consistent with the speculated mechanism , we now show quantitatively that robust , biologically compatible airway epithelium lineage behaviors are possible when only one out of two differentiated cell types ( SecrCs ) provide negative feedback to SCs . This control arrangement explains why no lineage recovery mechanism gets triggered when only CilCs are injured . On its surface , it would appear that such lineage “blindness” to CilC depletion represents a major vulnerability of the airway epithelium . Clearly , biologically speaking , lack of quick epithelium repair would compromise its anatomical integrity and function . What can then explain this seemingly irrational control design ? We hypothesize that this way of lineage control represents an example of an evolutionary “economy” . Clearly , having two negative control loops to SCs ( both from the CilCs and SecrCs ) would lead to a robust and quick recovery following all types of differentiated cell loss . However , in a real-life situation it is likely unnecessary for the epithelium to be able to quickly recover from the loss of only one type of differentiated cells . To-date , there are no natural events that would deplete one but not the other type of differentiated cells; this can be only induced experimentally using an artificial genetic system . On the other hand , depletion of both CilCs and SecrCs happens , commonly following inhalation of the poisonous naturally occurring sulfur dioxide ( SO2 ) [71 , 72] , or as the result of acute viral infection , such as with the influenza virus [26 , 73] . Therefore , such naturally occurring injuries are enough to be able to trigger repair mechanisms by removing an inhibitory signal emanated by just one cell type . Conditions of scenario III ( specific CilC loss ) do not represent a situation for which an organism should be prepared . This interesting experiment reveals the absence of a signaling loop from CilCs back to SCs . We can think of this arrangement of control loops as an example of cooperation among different cell types . SecrCs signal back to SCs to help recover their own loss and the loss of CilCs . Another type of question that can be addressed with our framework is the necessity for various processes in control networks . For example , stability analysis shows that SecrCs de-differentiation to SCs in the airway epithelium is not observed under the equilibrium conditions ( to keep the balance of cell numbers , eq ( 15 ) , which coincides with earlier reports [29] ) . At the same time , de-differentiation is the process that has been experimentally shown to allow for the quick recovery of the SC numbers after their removal [26] , see Fig 4 ( c ) . The question arises whether de-differentiation may have another role in the system , because catastrophic SC depletion ( of the type created in the experimental setup of [26] ) is probably unlikely under natural conditions . Why did the mechanism of de-differentiation evolve in the first place ? The answer to this question comes directly from our theory . The presence of de-differentiation , and more specifically , de-differentiation controlled negatively by the SC population , is a necessary condition for the system’s stability , as follows from the expression for the first eigenvalue in Eq ( 18 ) . The biological explanation of this condition is that SC death occurs at low rates ( e . g . due to mutations ) . Since SCs divide strictly asymmetrically , they are not able to compensate for low rate of SC death over time . We propose that this can be compensated by SecrC de-differentiation , as follows from out analysis . Finally we need to mention the numerous limitations of this study in particular and the methodology developed in general . The biggest drawback is the absence of spatial considerations . In the literature , spatial models of SC dynamics have been studied by several authors [41 , 55 , 60 , 74–77] , see also the reviews [78–80] . Analytical results have only been obtained in the simplest systems , and did not include any considerations of regulatory networks . Our first attempts of the analytical treatment of spatially distributed SC systems are concerned with cell mutagenesis and cancer generation [81] . In [82] we provide analytical solutions of a very simple , spatially regulated SC lineage again in the context of carcinogenesis and tumor suppressor gene inactivation . The present framework can only mimic spatial tissue organization by weighing “local” and “global” control loops differently . An explicit treatment of spatial structures is subject of future work . Another limitation of this theory is the requirement of relatively small deviations from the equilibrium . Theoretical basis for this approach ( which stems from the linear noise approximation [61] ) requires a weak dependence of the control functions on cell population numbers . While injury recovery dynamics certainly can be modeled by means of stochastic simulations , as we did in the current paper , the theory is inherently “local” . More analysis is required to study the global stability and global dynamics of SC systems , see [50–52] . As the final message , we would like to propose that the current framework can be used to study the general principles that govern SC lineage dynamics , across tissues . Several such candidate principles come to mind , including ( 1 ) “economy” ( the non-existence of overlapping controls not needed for stability or robustness ) , ( 2 ) “cooperation” ( such as in the example given by SecrCs signaling back to SCs to help compensate for the CilC loss as well as their own ) , and ( 3 ) “robustness” in the sense that certain loop arrangements allow stability for very large parameter regions , compared to others . In the airway epithelium example , the observed network is stable for any parameter values as long as they have the correct sign , in contrast to some other network configurations considered in S1 Text , Section 4 . By using our methodology , one can study such patterns of cell regulation and ask how they trade off in the context of stability and variance minimization . This is one of future directions of research and immediate applications of the near equilibrium calculus of stem cells developed here , albeit in some tissue types application of this technique can be hindered by the time scale and spatial scale separation of cells within the lineages . | Tissue stability is the basic property of healthy organs , and yet the mechanisms governing the stable , long-term maintenance of cell numbers in tissues are poorly understood . While more and more signaling pathways are being discovered , for the most part it remains unknown how they are being put together by different cell types into complex , nonlinear , hierarchical control networks that , on the one hand , reliably maintain constant cell numbers , and on the other hand , quickly adjust to oversee the robust response to tissue damage . Theoretical approaches can fill the gap by being able to reconstruct the underlying control network , based on the observations about the aspects of cellular dynamics . We argue that while many hypothetical networks may be capable of basic cell lineage maintenance , some are much more efficient from the viewpoint of variance minimization . Thus , we developed a new methodology that can test various control networks for stability , variance , and robustness . In the example of the airway epithelium that we highlight , it turns out that the evolutionary selected , actual architecture coincides with the mathematically optimal solution that minimizes the fluctuations of cell numbers at homeostasis . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"cell",
"cycle",
"and",
"cell",
"division",
"population",
"dynamics",
"cell",
"processes",
"cell",
"differentiation",
"physiological",
"processes",
"developmental",
"biology",
"systems",
"science",
"mathematics",
"homeostasis",
"... | 2016 | Near Equilibrium Calculus of Stem Cells in Application to the Airway Epithelium Lineage |
Dengue virus ( DENV ) is the most prevalent and burdensome arbovirus transmitted by Aedes mosquitoes , against which there is only a limited licensed vaccine and no approved drug treatment . A Chromobacterium species , C . sp . Panama , isolated from the midgut of A . aegypti is able to inhibit DENV replication within the mosquito and in vitro . Here we show that C . sp . Panama mediates its anti-DENV activity through secreted factors that are proteinous in nature . The inhibitory effect occurs prior to virus attachment to cells , and is attributed to a factor that destabilizes the virion by promoting the degradation of the viral envelope protein . Bioassay-guided fractionation , coupled with mass spectrometry , allowed for the identification of a C . sp . Panama-secreted neutral protease and an aminopeptidase that are co-expressed and appear to act synergistically to degrade the viral envelope ( E ) protein and thus prevent viral attachment and subsequent infection of cells . This is the first study characterizing the anti-DENV activity of a common soil and mosquito-associated bacterium , thereby contributing towards understanding how such bacteria may limit disease transmission , and providing new tools for dengue prevention and therapeutics .
Dengue virus ( DENV ) is arguably the most prominent arboviral threat to humans in tropical areas , with 3 . 6 billion people at risk of infection and 100 million people symptomatically infected annually [1] . All four ( 1–4 ) DENV serotypes are able to cause significant disease , ranging from dengue fever to dengue hemorrhagic fever and dengue shock syndrome , these latter most often upon secondary infection [2] . The lack of approved drug regimens directly against the virus and the limited license of the only approved vaccine make preventing bites from the Aedes mosquito vector the most effective way of blocking disease transmission thus far . DENV is a single-stranded , positive-sensed RNA virus of the Flaviviridae family . The viral genome is encapsidated by the capsid ( C ) proteins , and enveloped by a lipid bilayer within which the envelope ( E ) proteins and the mature membrane ( M ) proteins are embedded [3] . Among these three viral structural proteins , the E protein is responsible for binding to cell surface receptors , therefore triggering receptor-mediated endocytosis and subsequent internalization of the virion [4] . Our previous work has established that Chromobacterium sp . Panama can inhibit DENV infection in Aedes aegypti when it colonizes the mosquito midgut , as well as in vitro infection of cells when the virus is previously incubated with this bacterium [5] . This Chromobacterium species was isolated from A . aegypti in Panama and has also been shown to limit the life-span of laboratory-reared A . aegypti and A . gambiae , as well as directly interfere with P . falciparum development and infection . While the anti-DENV mechanism ( s ) were unknown , the in vitro inhibition suggested it was mosquito-independent [5] , and indicated possible production of C . sp . Panama-derived antiviral factor ( s ) with transmission-blocking and therapeutic potential . In this study , we aimed to identify the anti-DENV factors produced by C . sp . Panama and to describe their mechanism of action . We employed a bioassay-guided fractionation approach to discover a secreted aminopeptidase as mediating C . sp . Panama’s anti-DENV activity through the degradation of the viral E protein , and thus preventing virus attachment to the host cell and subsequent infection .
In our previous work we explored the in vitro anti-DENV activity of C . sp . Panama by incubating unfiltered bacterial cultures with DENV particles , removing the bacterial cells only prior to assessing infection in mammalian cells or mosquito cells [5] . To understand if live bacteria are required to impair DENV infection , C . sp . Panama cultures were first filtered through a 0 . 22 μm membrane and only then mixed with the DENV viral stock . DENV replication in both BHK-21 cells ( mammalian ) and C6/36 cells ( mosquito ) was reduced by 3–5 logs ( BHK-21: p < 0 . 0001 , C6/36: p = 0 . 0009 ) upon treatment with this sterile supernatant when compared to the control ( Fig 1A ) . This indicates that the in vitro anti-DENV activity displayed by C . sp . Panama could be attributed to secreted or otherwise released mediators and that live bacteria are not required to impair DENV infection . Similar level of C . sp . Panama-mediated anti-DENV activity in both cell lines suggests that the target of such bacteria-derived factors are likely to be viral proteins or RNA , or host factors that are conserved across species [6] . To understand the nature of the C . sp . Panama-produced factors that influence DENV infection , we first partitioned the C . sp . Panama supernatant with commonly used organic solvents , both miscible and immiscible in water . C . sp . Panama molecules soluble in acetone , acetonitrile , butanol , chloroform , ethanol , ethyl acetate and hexanes showed no anti-DENV activity ( Fig 1B ) . Even though methanol partially extracted molecules with anti-DENV properties ( < 1-log reduction in DENV2 titer compared to control , p = 0 . 002 ) , water was the only solvent that was able to retain a significant majority of the activity observed in the initial extract ( 4-log reduction , p = 0 . 0002 ) , providing early evidence that the relevant C . sp . Panama-produced factors are likely charged or otherwise hydrophilic biomolecules , such as most proteins or sugars , and not lipidic in nature . Next , we assessed the thermostability of such factors , concluding that 1 hour incubation at 50°C partially inactivated ( 2-log reduction in DENV2 titer compared to control , p = 0 . 0005 ) the level of anti-DENV activity of samples handled at room temperature ( as before ) or 37°C , for which no viral replication was detected ( Fig 1C ) . Incubation at 70 or 99°C for 1 hour resulted in total loss of activity ( Fig 1C ) . Taken together , these data strongly suggest that the C . sp . Panama-produced anti-DENV factors represent hydrophilic proteins , given the very strong preference for the aqueous solvent–the only that would ensure proper protein folding [7 , 8] . The dramatic loss of activity at 50°C and above would not be as readily expected from sugars or other small metabolites . To further test our hypothesis , we used ammonium sulfate precipitation [9] to create protein extracts of the C . sp . Panama supernatant . The resulting protein suspension carried all the anti-DENV activity ( 4-log reduction in DENV titer compared to control , p < 0 . 0001 , Fig 1D ) , whereas the supernatant , after appropriate desalting , was ineffective in reducing DENV infectivity in vitro . This result clearly indicates that proteinous factors , such as peptides , proteins or protein complexes , are the likely mediators of C . sp . Panama’s anti-DENV activity . Attachment of viral particles to host cells is the first step that initiates virus entry , and thus precedes all the following steps necessary for viral replication . To understand if our C . sp . Panama proteinous extract exerts its anti-DENV effect by interfering with viral attachment , we performed neutralization assays pre- and post-attachment using low temperature to arrest receptor-mediated endocytosis as before [10] . In the pre-attachment assay , we incubated DENV virions with C . sp . Panama proteins at room temperature and then exposed this mixture to pre-chilled cells at 4°C to allow for viral attachment without endocytosis . Unbound virions were washed off , and attached virions were allowed to complete entry and infection in host cells at 37°C . In the post-attachment assay , DENV virions were first adsorbed to pre-chilled cells at 4°C and , only after extensive washing , C . sp . Panama proteins were added to treat attached virions . Our results showed a significant 3-log decrease in DENV titers compared to the control ( p < 0 . 0001 ) when DENV was exposed to the C . sp . Panama proteins before attachment ( Fig 2A ) . In contrast , C . sp . Panama proteins failed to inhibit DENV replication when attachment had already occurred ( Fig 2A ) , indicating that the C . sp . Panama-produced anti-DENV factor ( s ) is likely to either interfere with the viral particle itself and/or prevent its attachment to host cells . To investigate whether C . sp . Panama proteins affect viral particles , we incubated the protein extract with purified DENV2 virions and processed the samples for transmission electron microscopy ( Fig 2B ) . The DENV mature particle has a diameter of 50–60 nm [11 , 12] and , fully consistent with that , we obtained a measurement of 56 . 7±1 . 1 nm for the diameter of the viral particles in our control sample in Tris-HCl , pH 7 . 2 ( Fig 2C ) . Upon treatment with C . sp . Panama proteins in the same buffer , however , the viral particles exhibited a significantly different ( p < 0 . 0001 ) diameter of 34 . 0±3 . 2 nm ( Fig 2C ) . The DENV E protein occupies a 180–280 Å ( i . e . 18–28 nm ) radius of the particle , as shown by analysis of a class of virus with loose or absent E protein shell with diameter of ~36 nm [12] . As such , we hypothesized that treatment with our bacterial protein extract compromised the integrity of the DENV particle E protein coating layer . This is in agreement with our previous finding that viral attachment is compromised , as the E protein is a crucial mediator of viral attachment to , and fusion with , host cells [13 , 14] . To test this hypothesis , we exposed DENV to the C . sp . Panama protein extract as before and analyzed the presence of DENV E protein by Western blotting . As shown in Fig 2D , a significantly diminished signal for the E protein at the expected size of 53 kDa [14] is detected in the sample treated with C . sp . Panama protein extract , the same not being true for when DENV is treated with a heat-inactivated C . sp . Panama protein extract . Our immunofluorescence microscopy results were consistent with this finding by failing to detect attached virions to BHK-21 cells with another anti-E protein antibody ( 4G2 , ATCC; Fig 2E ) . However , as the cells were washed prior to staining , it is possible to interpret that any remaining bound virus ( as hinted in Fig 2A ) was simply below the detection limit for this technique . Nonetheless , the Western blotting data show E protein degradation upon treatment with C . sp . Panama proteins , pointing to a potential C . sp . Panama-secreted protease as the most likely factor responsible for this effect . In a parallel approach , we also sought to further partition the C . sp . Panama culture supernatant protein extract in search of the causative agent for the anti-DENV activity . For this purpose , we employed hydrophobic interaction chromatography ( HIC ) to separate proteins based on their content in hydrophobic amino acids . Following an initial fractionation in butyl Sepharose under a salt gradient , we were able to obtain a discrete fraction ( H2 , Fig 3A ) that preserved the anti-DENV properties . This fraction was further purified by a second round of fractionation where conductivity along the gradient was monitored to match ~60 mS/cm , resulting in the collection of fraction H2p ( Fig 3B ) . SDS-PAGE gels of these fractions revealed that fraction H2p is considerably less diverse than H2 , with the exception of a band at approximately 30 kDa and other small proteins or peptides below 16 kDa ( Fig 3C ) . Most importantly , this fraction retained proteolytic activity , as shown in Fig 3C , by its ability to degrade DENV proteins , including the E protein whose signal can be seen at ~53 kDa band . Given the low complexity of the H2p fraction , we next chose to perform a proteomic analysis by mass spectrometry to understand which C . sp . Panama proteins were contained within this sample and the results are summarized on Table 1 ( list of protein hits presented in S1 Table ) . Of the eight detected proteins , two ( CSPP0261 and CSPP0262 ) were of particular relevance given their assignment as proteolytic . The neutral protease ( CSPP0261 ) and the aminopeptidase ( CSPP0261 ) are encoded in tandem in the C . sp . Panama genome , and both are assigned to be secreted according to the most closely related UniProt reviewed entries: P14756 from Pseudomonas aeruginosa for CSPP0261 and Q01693 from Vibrio proteolyticus for CSPP0262 , making them ideal candidates for subsequent studies . Having putatively narrowed down the anti-DENV activity of the C . sp . Panama culture supernatant to two proteases , we aimed at a better understanding of these two enzymes by comparing their sequence to similar proteins already discussed in the literature . For the neutral protease CSPP0261 , BLAST analysis returned significant similarity to the P . aeruginosa pseudolysin ( 59 . 3% pairwise identity ) and the V . proteolyticus vibriolysin ( 49 . 7% pairwise identity ) , two secreted zinc metallopeptidases of the M4 family that form a divergent branch within the clan [15] . They have been described as having a nutritional role by scavenging amino acids for the bacteria , as well as a virulence role in pathogenic species , either by directly mediating the destruction of tissue or indirectly by interfering with host defense mechanisms or the normal function of host proteases [16] . Sequence alignment showed that the neutral protease CSPP0261 conserves the active and metal binding sites of these enzymes ( Fig 4A ) , and that it lacks the secondary C-terminal catalytic domain of vibriolysin [17] . For the aminopeptidase CSPP0262 , the only matching fully reviewed entry in UniProt corresponded to the Vibrio aminopeptidase ( 39 . 6% pairwise identity ) from V . proteolyticus , also a secreted zinc metallopeptidase but belonging to the M28 family as it binds two zinc ions that co-catalyze the proteolytic reaction [18] . This enzyme removes free N-terminal amino acids of peptides or proteins that are hydrophobic and large ( preferably ) , basic or proline , but never aspartyl , glutamyl or cysteic acid residues; this capability has been harvested in biotechnological settings namely in the production of the antiviral and anticancer agent interferon alfa-2b [18 , 19] . Sequence alignment to the aminopeptidase CSPP0262 showed perfect agreement in all the annotated active metal binding sites for the V . proteolyticus ortholog ( Fig 4B ) ; CSPP0262 lacks the C-terminal propeptide present in the Vibrio aminopeptidase for which , however , no function has been assigned [20 , 21] . Of note , both proteases contain a conserved signal peptide sequence potentially also governing their secretion by the Chromobacterium species . Also conserved is an N-terminal propeptide that has been shown for both ortholog proteins to be crucial for correct folding of the mature protein , having the ability to inhibit its activity until secretion to prevent intracellular damage [16 , 20 , 21] . After processing , both C . sp . Panama enzymes would assume an active form of ~32 kDa , consistent with our observation that our proteolytically active H2p fraction is enriched with a protein of this size ( Fig 3C ) . A difference between the two peptidase families , however , is their susceptibility to inhibitors: phosphoramidon is known to inhibit the activity of the M4 proteases [22] whereas bestatin has been described as an inhibitor of the M28 peptidase [18] . In an attempt to determine which of these enzymes is responsible for the anti-DENV activity of C . sp . Panama , we grew the bacteria in media supplemented with each of these inhibitors and evaluated the culture supernatant for its ability to degrade the DENV2 E protein . As shown in Fig 4C , treatment with 10 μM bestatin , and not 100 μM phosphoramidon , was able to revert E protein degradation by the C . sp . Panama culture supernatant . This indicates that the activity of the CSPP0262 aminopeptidase is crucial for digestion of the E protein , with minimal contribution of CSPP0261 . These results are replicated when assaying the effect of these inhibitors on DENV titers by plaque assay with a total ablation of anti-DENV activity in the presence of bestatin ( Fig 4D ) , but a more prominent role can now be postulated for CSPP0261 . Even though supplementation of bacterial growth media with phosphoramidon is not able to fully counteract the inhibitory activity of the C . sp . Panama culture supernatant , it is apparent that this inhibitor aids in rescuing the phenotype . Taken together , the results of our analyses point to the CSPP0262 aminopeptidase as the causative agent of the anti-DENV activity by C . sp . Panama , and support a facultative role of the CSPP0261 neutral protease in the process .
The anti-DENV activity of C . sp . Panama was further investigated in the present work , building on the initial observation that this bacterium can inhibit viral replication in vivo and in vitro [5] . First , we validated that the activity is mediated by a secreted factor and does not require live bacteria; retention of activity by protein extracts of the culture supernatant further indicated that a protein was responsible for the effect . Then , we demonstrate that the C . sp . Panama protein extract interfered with viral attachment by promoting the degradation of its E protein . Successive fractionation of this extract allowed for the isolation and identification of two tandemly expressed proteases , CSPP0261 and CSPP0262 , that appear to function somewhat synergistically towards the anti-DENV activity of C . sp . Panama , as indicated by experiments conducted with specific inhibitors to each of the peptidases . A mechanism for this synergy can be speculated based on studies conducted with the ortholog enzymes expressed and secreted by V . proteolyticus . As mentioned earlier , the CSPP0262 aminopeptidase contains an N-terminal propeptide that has been shown in the Vibrio ortholog to be inhibitory towards its proteolytic activity until it is removed . This cleavage can happen autocatalytically [23] , but a role has been proposed for vibriolysin as the biologically relevant activating peptidase [24] . Vibriolysin and the Vibrio aminopeptidase are co-expressed [24] , as also CSPP0261 and CSPP0262 appear to be , which further validates this observation . P . aeruginosa also secretes an aminopeptidase belonging to the M28 family that is processed to its mature form by pseudolysin [25] , but BLAST analysis revealed weak identity between this aminopeptidase and CSPP0262 ( 31% identity; 24% query cover ) . Given this , our findings manipulating this system with the different protease-specific inhibitors suggest an effector role for the CSPP0262 aminopeptidase , with the CSPP0261 neutral protease perhaps acting to facultatively cleave and activate the aminopeptidase . This is compatible with the observation made in V . proteolyticus that , even though the aminopeptidase is able to be activated by autocatalysis , that route is less efficient when compared to the activation by the respective neutral protease [23] . Ours is the first study characterizing the mechanisms of the anti-DENV activity of a bacterium that was originally isolated from the Aedes mosquito midgut . It is improbable that C . sp . Panama has evolved to secrete the CSPP0262 aminopeptidase to benefit from its antiviral properties; a more likely scenario has this aminopeptidase as having a primordial role in digestion and nutrient scavenging similar to what has been put forward for the ortholog in V . proteolyticus . Interestingly , while the highly proteolytic environment of the blood-fed mosquito midgut [26] does not significantly impact DENV infection , the C . sp . Panama-produced protease does , thereby representing a potent anti-DENV effector molecule . This knowledge could be used towards developing transmission-blocking strategies , based on the presence of C . sp . Panama itself in the midgut , or through transgenic or paratransgenic expression of the anti-DENV protease in the midgut tissue [27 , 28] . While it is not unconceivable that DENV could acquire mutations in its E protein that render the activity of the aminopeptidase from C . sp . Panama ineffective , we can hypothesize that that would come at a high fitness cost for the virus given the high specialization of this envelope protein and its sequential role in both attachment and entry . As such , a potent anti-DENV protease could also be developed into a therapeutic agent to treat viral infection in humans . Often overlooked as a class of therapeutics , proteases have been approved by the U . S . FDA to treat conditions such as hemophilia , acute myocardial infarction or muscle spasms since the 1980s [29 , 30] . While their antiviral potential has only been explored in topical applications against cold-causing rhinoviruses and , more recently , herpes simplex virus [31] , the great amount of research on alternative-to-parenteral delivery methods for biologics brings promise on the ability to deliver therapeutic proteins through non-invasive routes such as orally or intranasally in the near future [32–34] . Other potential hindrances relate to their propensity for immunogenicity in the form of anti-drug antibodies and the lack of specificity of their proteolytic activity as measured by the number of putative substrates , but these obstacles have also been extensively addressed and strategies developed to overcome them [35–37] . For example , the botulinum toxins are also zinc metalloproteases initially characterized from bacteria: significant advances since their introduction have allowed for new-generation drugs with reduced immunogenicity and increased stability that can be absorbed transdermally [38 , 39] . The evaluation of the aminopeptidase from C . sp . Panama as an antiviral could benefit and be expedited from the progress made in this research arena . The potential versatile applications of C . sp . Panama aminopeptidase in dengue control warrants further basic and translational investigations .
Cell lines were maintained and viral stocks prepared as previously described [40] . BHK-21 hamster kidney cells ( ATCC CCL-10 ) were grown at 37°C , 5% CO2 , in DMEM containing 10% fetal bovine serum ( FBS ) , 1% penicillin- streptomycin and 5 μg/ml Plasmocin . C6/36 Aedes albopictus cells ( ATCC CRL-1660 ) were grown at 32°C , 5% CO2 , in MEM , with the same additives , except Plasmocin , plus 1% non-essential amino acids . C6/36 cells were used to propagate DENV2 New Guinea C strain; after 6 days post infection ( dpi ) , cells were lysed by three successive freeze-thaw cycles and clarified by centrifugation at 1 , 500–2 , 000 g for 10 minutes , from which the soluble content was combined with cell supernatant as viral stock . This viral stock was aliquoted and stored at −80°C for later use . Unless otherwise indicated , Chromobacterium sp . Panama was grown in LB broth at 30°C for 72 hours at 250 rpm . Culture supernatants were obtained by filtering this preparation through a 0 . 22 μm syringe or disk filter . Levels of DENV2 infection were assayed as before: by plaque assay in BHK-21 cells [40] or focus forming assay in C6/36 cells [41] . Both cell lines were seeded at ~8 . 4×105 cells/well and then incubated at 37°C or 32°C overnight , respectively , before reaching 80–90% confluence . Cell monolayers were then incubated for 1 hour with untreated DENV2 viral stock or a 1:1 mixture of the stock with the different treatments after 1 hour of pre-incubation and covered with a 0 . 8% methylcellulose overlay DMEM medium with 2% FBS . After 5–6 dpi BHK-21 cell monolayers were fixed and stained with 1% crystal violet in methanol/acetone ( 1:1 ) . For the focus forming assay , C6/36 cells were fixed with methanol/acetone ( 1:1 ) and blocked with 5% skim milk in PBS . The C6/36 monolayer was probed with mouse anti-flavivirus envelope protein antibody ( 4G2 , ATCC; 1:1 , 000–1:2 , 000 ) followed by goat anti-mouse IgG-HRP ( 1:1 , 500 ) , and developed with True Blue Peroxidase Substrate ( KPL ) . In each case , the number of plaque or focus forming units ( PFU , FFU ) per ml was determined , and unpaired t-tests used to conclude on the significance of differences seen to the respective controls , correcting for multiple comparisons with the Welch’s method when appropriate . For the water-immiscible solvents , C . sp . Panama supernatant was mixed with butanol , chloroform , ethyl acetate and hexanes , at 1:1 and the non-aqueous phase was recovered . For acetone , acetonitrile , ethanol , methanol and water , C . sp . Panama supernatants were desiccated in an Eppendorf Vacufuge and the resulting residue resuspended in the same volume of the different solvents for 4 hours under continuous shaking ( 1400 rpm ) . Particles that did not resuspend were then pelleted by 10 minutes centrifugation at 8 , 000 g in a table top centrifuge and the supernatants were recovered . All extracts were then desiccated in an Eppendorf Vacufuge and the resulting residue resuspended in the same volume of water for 4 hours under continuous shaking ( 1400 rpm ) . After being filtered through a 0 . 22 μm syringe filter , each extract was added 1:1 to the DENV2 stock and incubated at room temperature for 45 minutes , and infectivity was assayed by plaque assay in BHK-21 cells . Desiccated LB medium resuspended in water was used as negative control . Ammonium sulfate , ( NH4 ) 2SO4 , was slowly added to the C . sp . Panama supernatant until reaching 70% saturation . After rapid stirring at 4°C for 1 hour , proteins were pelleted down by centrifugation at 10 , 000 g for 20 minutes . The protein pellet was then gently resuspended in 0 . 1 M Tris-HCl pH 7 . 2 buffer and concentrated using an ultra-centrifugal filter unit ( nominal molecular weight limit , 30 kDa; Amicon ) . Protein concentrations were determined by BCA assay ( Pierce , Thermo Scientific ) . For comparison , the ammonium sulfate-rich supernatant was desalted and concentrated to the same extent . The following protocols were adapted from the literature [10] . Pre-attachment: BHK-21 cell monolayers and reagents were cooled to 4°C for 1 hour . Virus and C . sp . Panama extract were incubated at 4°C for 1 hour , and then adsorbed to pre-chilled BHK-21 monolayers for another hour at 4°C . After adsorption , cell monolayers were washed 3 times with cold PBS , overlay medium was added and cells were incubated at 37°C for 5 days before fixing and staining for plaque detection . Post-attachment: BHK-21 monolayers and reagents were pre-chilled to 4°C for 1 hour . Virus was added to cells and allowed to attach for 1 hour at 4°C . Unbound virus was washed off by washing twice with cold PBS . C . sp . Panama extract was then added to virus bound cells and incubated for another hour at 4°C . Following incubation , cell monolayers were washed once with cold PBS . Overlay medium was added and cells were incubated at 37°C for 5 days before fixing and staining for plaques . For purification of viral particles , DENV2 stocks were pelleted in 8% w/v PEG 8000 at 14 , 000 g for 1 hour , further purified by 24% w/v sucrose cushion for 1 hour at 248 , 000 rpm ( Beckman SW28 rotor ) and separated in potassium tartrate-glycerol gradient 10–30% for 2 hours at 175 , 000 g ( Beckman SW55Ti rotor ) , as described previously [42] . Pure DENV2 viral particles suspended in 1:1 v/v 0 . 1 M Tris-HCl ( pH 7 . 2 ) or in the C . sp . Panama proteinous extract were incubated at room temperature for 1 hour and virus within the samples was sequentially inactivated in 4% paraformaldehyde in PBS for 30 minutes at 25°C followed by 30 minutes at 37°C . The samples in 0 . 08 M phosphate were then applied to a carbon grid , washed 3 times with 50 mM TBS and negatively stained with 1% w/v uranyl acetate 0 . 04% w/v trehalose . The grids were allowed to air-dry and images were acquired in a Hitachi 7600 TEM microscope at 80 . 0 kV . Viral stocks were incubated 1:1 v/v with the differentially treated C . sp . Panama culture supernatants , their protein extracts or control buffers for 1 hour at room temperature . For heat inactivation , the protein extract was incubated at 90°C for 1 hour prior to being added to the virus suspension . Samples were then incubated at 85°C for 2 minutes in SDS-containing loading buffer and ran on 4–20% tris-glycine gels under denaturing conditions . Protein bands were dry-transferred to a nitrocellulose membrane , followed by overnight blocking at 4°C in 5% skim milk 0 . 1% Tween 20 PBS . Membranes were probed with an anti-DENV2-E rabbit polyclonal ( 1:2 , 000 ) followed by anti-rabbit HRP-linked donkey ( 1:20 , 000 , Amerhsam ) antibodies for 1 hour each , with abundant wash with 0 . 1% Tween 20 PBS after each probing . Chemiluminescence signals were detected using Amersham ECL Prime Western Blotting detection reagents . Immunostaining and fluorescence microscopy was conducted as described before [43] , with adaptations . BHK-21 cells were seeded on a glass coverslip , grown at 37°C up to ~50% confluency , and incubated with DENV2 in the presence and absence of C . sp . Panama proteins for 1 hour at 4°C . Cells were washed with cold PBS and fixed with 4% paraformaldehyde-PBS for 15 minutes at room temperature . Fixed cells were blocked with 0 . 5% BSA in PBS for 1 hour at room temperature , washed with PBS and incubated with mouse 4G2 antibody ( ATCC ) for 2 hours at room temperature . Alexa Fluor 568 goat anti-mouse IgG ( Invitrogen ) was used as secondary antibody . Beta-actin was stained with Alexa Fluor 488 Phalloidin ( Invitrogen ) and nuclei were stained with DAPI . Samples were mounted with ProLong Gold antifade reagent and visualized on a Leica DM 2500 fluorescence microscope . Fast protein liquid chromatography was performed using an ÄKTA Purifier system . Hydrophobic interaction FPLC was conducted on protein extracts of C . sp . Panama resuspended using HiTrap Butyl FF or Butyl HP columns ( GE Healthcare Life Sciences ) under gradient elution from 1 . 5 to 0 M ammonium sulfate in 50 mM sodium phosphate ( pH 7 . 0 ) at constant 1 mL/min flow rate . Collected fractions were dialyzed using ultra-centrifugal filter units ( nominal molecular weight limit , 30 kDa; Amicon ) for solvent replacement as needed . Fraction H2p in 0 . 1M Tris-HCl was submitted to the Johns Hopkins Medicine Mass Spectrometry Core for identification using iTRAQ as before [44] . Peptides obtained were searched against the theoretical proteomes of Chromobacterium spp . following trypsin digestion using Mascot ( Matrix Science ) ; Scaffold 4 . 8 ( Proteome Softwares , Inc . ) was used to validate protein identification at 99 . 0% protein probability threshold , 4 as the minimum number of peptides and 95% as the peptide probability threshold . Amino acid sequences of identified peptides and proteins were further searched against the C . sp . Panama theoretical proteome based on its genomic information ( NCBI Identifier QARX00000000 ) for validation of the findings . Sequences of proteins of interest were subjected to basic local alignment tools ( BLAST ) against the UniProt database and further aligned to the closest related annotated entry using the Geneious v5 . 4 algorithm ( Biomatters Limited ) . Bestatin and phosphoramidon ( Sigma ) were diluted in water and supplemented to LB at the indicated concentrations prior to inoculation of C . sp . Panama . Following 72 hours incubation at 30°C , bacterial culture supernatants were collected and either processed as before for Western blotting or evaluation of effect on DENV2 titer by plaque assay on BHK-21 cells . | The global burden of dengue cannot be overstated , and lack of effective control measures , besides avoiding contact with the mosquito vector , creates an urgent need to develop new strategies to control the disease . Here we elucidate the mechanism by which a natural soil bacterium inhibits DENV infection . Chromobacterium sp . Panama secretes a protease that is able to undermine the integrity of the viral particles , preventing their attachment and thus infection of host cells . This study contributes towards understanding how natural bacteria of the vector gut may limit disease transmission , and provides new candidate tools for the development of dengue transmission-blocking and therapeutic agents . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"dengue",
"virus",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"viral",
"transmission",
"and",
"infection",
"enzymes",
"pathogens",
"geographical",
"locations",
"enzymology",
"microbiology",
"viral",
"structure",
"panama",
"nor... | 2018 | Aminopeptidase secreted by Chromobacterium sp. Panama inhibits dengue virus infection by degrading the E protein |
The reshaping and decorrelation of similar activity patterns by neuronal networks can enhance their discriminability , storage , and retrieval . How can such networks learn to decorrelate new complex patterns , as they arise in the olfactory system ? Using a computational network model for the dominant neural populations of the olfactory bulb we show that fundamental aspects of the adult neurogenesis observed in the olfactory bulb – the persistent addition of new inhibitory granule cells to the network , their activity-dependent survival , and the reciprocal character of their synapses with the principal mitral cells – are sufficient to restructure the network and to alter its encoding of odor stimuli adaptively so as to reduce the correlations between the bulbar representations of similar stimuli . The decorrelation is quite robust with respect to various types of perturbations of the reciprocity . The model parsimoniously captures the experimentally observed role of neurogenesis in perceptual learning and the enhanced response of young granule cells to novel stimuli . Moreover , it makes specific predictions for the type of odor enrichment that should be effective in enhancing the ability of animals to discriminate similar odor mixtures .
Contrast enhancement and decorrelation are common steps in information processing . They can reshape neuronal activity patterns so as to enhance down-stream processing like pattern discrimination , storage , and retrieval . The activity patterns can be complex and new patterns may become relevant due to changes in the environment or in the life circumstances of the animal . How can networks adapt to such demands , as they arise , for instance , in the olfactory system ? What are neural substrates that would allow the necessary network restructuring ? In the olfactory system initial sensory processing is performed in the olfactory bulb . Its inputs consist of activation patterns of its 100–1 , 000 glomeruli , each of which can be considered as an individual input channel representing a specific olfactory receptive field . The bulbar network reshapes the patterns representing odor stimuli and typically reduces the correlation between output patterns representing similar odors as compared to the respective input patterns [1]–[3] . It does so despite the fact that even simple odors evoke complex activation patterns due to the fractured representation of the high-dimensional odor space on the two-dimensional glomerular surface [4] . Unlike spatial contrast enhancement in the retina [5] , this decorrelation can therefore not arise from local lateral inhibition that is confined to neighboring glomeruli [3] , [6] . What types of network connectivities can then underlie the enhancement of small , but significant differences in the representation of similar odors ? Previously , a number of different decorrelation mechanisms have been proposed , each of which exploiting a different aspect of the nonlinear dynamics of the bulbar network . The network connectivities were taken to be fixed , either without any lateral inhibition [4] , with all-to-all inhibition [7] , or with sparse random connections across large portions of the bulb [3] . These networks were shown to reduce quite effectively the correlation between the representations of moderately similar stimuli . A different perspective is suggested by two distinctive features of the olfactory system: i ) many odors do not have an intrinsic meaning to the animal and their significance is likely to be learned by experience [8]–[10]; ii ) the bulbar network structure is not static but undergoes persistent turn-over due to neurogenesis and apoptosis even in adult animals [11] , [12] . So far , the specific role of adult neurogenesis for olfactory processing is only poorly understood [13] , [14] . It is known that environmental changes like sensory deprivation [15]–[18] and odor enrichment [19]–[21] , associative learning [22]–[25] , and life circumstances like mating [26] and pregnancy [27] affect anatomical and functional aspects of the olfactory bulb . Moreover , genetic [28] , [29] , pharmacological [30]–[32] , and radiational manipulations [33] , [34] have identified the significance of neurogenesis in these modifications . Here we ask whether the neuronal turnover associated with adult neurogenesis can provide a neural substrate for the adaptation of the network to the decorrelation of different relevant stimuli that may be highly similar . Such a contribution of neurogenesis to pattern separation has been proposed for the olfactory bulb as well as the dentate gyrus [35] . We use a minimal computational network model of neurogenesis in the olfactory bulb that incorporates the persistent addition of new inhibitory interneurons ( granule cells ) into the olfactory bulb [36] , their connection with the principal mitral cells via reciprocal synapses through which the mitral cells excite the granule cells and the granule cells inhibit the mitral cells [37] , and the activity-dependent apoptosis of the granule cells [15] , [32] , [38]–[41] . Using stimulus ensembles based on glomerular excitation patterns observed in rat [42] we find that the networks learn to decorrelate even very similar stimuli . This results largely from the surviving granule cells detecting strongly co-active mitral cells and providing lateral inhibition between them . Our modeling gives a natural interpretation of recent experiments on the role of neurogenesis in the perceptual learning of a non-associative odor discrimination task [40] and the detection of novel odors [20] . Our computational model predicts that learning to decorrelate highly similar mixtures comprised of dissimilar components requires the exposure to a mixture of the components rather than the individual components themselves . This can be tested in behavioral experiments using suitable enrichment protocols [40] , [43]–[45] .
In our computational model we consider the recurrent network formed by principal mitral cells and inhibitory granule cells . We focus on the adaptive restructuring of the network connectivity in response to a stimulus ensemble and model the individual neurons in a minimal fashion using linear firing-rate dynamics ( cf . METHODS , Discrete Adaptive Network Model ) . Focusing on the evolution of the network structure we ignore transients in the evolution of the neuronal activities and consider only their steady states in response to any given odor stimulus . The network is persistently rewired by adding in each time step of the network evolution randomly connected new granule cells and removing granule cells that are not sufficiently active during the steady state reached in response to odor stimulation ( Fig . 1 ) . Specifically , the survival probability of a granule cell depends in a sigmoidal fashion on its ‘resilience’ , which we introduce as its thresholded activity summed over the stimulus ensemble . In most of the computations we use input patterns that are based on a set of experimentally obtained glomerular activity patterns in rat [42] corresponding to the odorants -limonene , -carvone , 1-butanol , 1-hexanol , 1-heptanol , and acetic acid ( Fig . 2A ) . They drive 424 mitral cells , which in turn excite about 10 , 000 granule cells . Due to the reciprocal character of these synapses each granule cell provides self-inhibition to each of the eight mitral cells that drive it as well as lateral inhibition between them ( Fig . 1A ) . All synaptic strengths are taken to be fixed . Unless noted otherwise , all excitatory and all inhibitory synapses have equal strengths , respectively . The network that eventually emerges as a statistically steady state from the persistent rewiring substantially reshapes the representation of the stimuli ( Fig . 2B ) . In particular , the mitral cell activation patterns , which represent the output of the olfactory bulb , differ from each other significantly more than the glomerular input patterns . To quantify this reduction in similarity we use the Pearson correlation of the patterns associated with stimuli and ( cf . Eq . ( 10 ) ) , as has been done in previous , experimental studies [1]–[3] . Thus , the network achieves a substantial decorrelation of the stimulus representations ( Fig . 2Ci , ii ) . This is the case for the highly similar -limonene- and -carvone-pairs as well as the less correlated , remaining stimuli of the odor ensemble . Moreover , through the enhanced inhibition of mitral cells that are strongly driven in this stimulus ensemble and the spontaneous activity of mitral cells that receive very little or no input [3] , [46] the network reshapes the quite focal input patterns into output patterns in which the activity is more broadly distributed over the whole network ( Fig . 2B ) . Such a reduction of the focality of the output patterns has been observed for mitral cell activity in zebrafish [3] . Particularly for stimuli that predominantly overlap in these focal areas such a reshaping of the pattern can reduce the correlation significantly . Insight into the mechanisms underlying the decorrelation by the network is gained by following the evolution of the connectivity and the associated decorrelation performance as the network builds up from a network without any granule cells ( Fig . 3 ) . The early stages of this evolution are not meant to mimic the peri-natal development of the bulb , which is controlled by mechanisms other than those included in this model . To visualize the network connectivities the stimuli are down-sampled to 50 channels ( cf . METHODS , Natural Stimuli ) and the two-dimensional activation patterns are re-arranged into one-dimensional vectors in which the mitral cells that are strongly activated during the -limonene presentation are located at the beginning of the vector and those that dominate during the -carvone presentation at the end . Because the overlap between the activation patterns of these two pairs of enantiomers is small there are only few mitral cells that receive significant input for both types of stimuli . They end up towards the middle of the activity vector . For visual clarity the diagonal elements of the connectivity matrices are divided by 10 . During the initial phase the granule cell population is small and provides only little inhibition to the mitral cells . Their activities and with them the activities of the granule cells are therefore high and none of the granule cells are removed ( Fig . 3A ) . Since the granule cells establish random connections with the mitral cells the resulting effective connectivity between the mitral cells is essentially random ( Fig . 3Bi ) and the activity patterns are only reduced in amplitude without any qualitative changes; the correlations remain high . As the mitral cell activities decrease , some granule cells fall in their activity and resilience below the soft survival threshold ( cf . Fig . 1Bi ) and their survival probability drops drastically ( . This apoptosis is selective , resulting in a structured connectivity , in which more highly active mitral cells receive stronger inhibition ( Fig . 3Bii ) , and a reduction of the mean pattern correlation . The correlation between the highly similar stimuli is , however , still high . In the third phase of the network evolution the size of the granule cell population remains constant , but the connectivity evolves slowly towards establishing strong effective mutual inhibition between mitral cells that are highly co-active during -limonene or -carvone presentations ( marked by circles in Fig . 3Biii ) . In parallel , the correlation of these highly similar enantiomers is strongly reduced . The effectiveness of the inhibition of highly co-active mitral cells in decorrelating activity patterns is illustrated using a very simple example with stimuli exciting only three mitral cells ( the relevant two stimuli are shown in Fig . 4 ) . Like the highly similar olfactory stimuli in Fig . 2 , these stimuli overlap in strongly co-active glomeruli . This allows the population of granule cells that connect to the mitral cells driven by these glomeruli to be much larger than the other two populations . The reciprocity of the synapses implies that these mitral cells receive substantially stronger inhibition than the other mitral cell . The resulting reduction in amplitude reduces also the correlation between the two mitral cell activity patterns . In Fig . 3Biii the corresponding enhanced connectivity between mitral cells that are highly co-active during -limonene ( or -carvone ) stimulation is marked by black circles . What determines the performance of the networks arising from the persistent turn-over ? The granule-cell survival is controlled by two thresholds: i ) for each stimulus for which the granule cell activity surpasses the resilience threshold its resilience increases ( cf . Eq . ( 8 ) ) and ii ) the resilience accumulated across all stimuli of the ensemble has to be above the soft survival threshold in order for the granule cell to have a significant survival probability ( cf . Eq . ( 9 ) ) . The survival threshold controls in particular the total number of granule cells and with it the overall level of inhibition . In general , the overall correlation of the outputs decreases with increasing inhibition ( data not shown ) at the expense of the output amplitudes . In our comparisons we adjust therefore to keep the mean output amplitudes fixed . A more subtle and interesting role is played by the resilience threshold . For the network achieves an overall decorrelation that is quite comparable to that of the network of Fig . 2 with ; the highly similar stimuli -limonene and -carvone , however , are only very poorly decorrelated ( Fig . 5A ) . The origin of this poor performance is apparent in the effective connectivity obtained with ( Fig . 5B ) . A comparison with the connectivity arising for ( Fig . 3 ) reveals that the connections among the mitral cells that are co-active in response to -limonene ( or -carvone ) stimulation ( black circles ) are not stronger than among mitral cells that are not co-active ( red circle ) . As had been observed in Fig . 3 , it is the connections among co-active mitral cells , however , that are essential for decorrelating these stimulus representations . How does the threshold provide a co-activity detector ? Why do the connections among mitral cells that are not co-active interfere with the decorrelation ? The function of the threshold can be illustrated with a minimal set of two pairs of strongly correlated stimuli activating four glomeruli , and with ( Fig . 6A ) . Stimuli and may be viewed as caricatures of the limonene and carvone enantiomers , respectively . The granule cells in population inhibit the mitral cells that are co-active in stimuli ( cf . Fig . 6A ) and are therefore needed for decorrelation . The granule cells in population , however , are connected to mitral cells that are not co-active in any of the stimuli; they may interfere with the performance of the network . The resilience of the granule cells in population is comprised of two large contributions due to the strong inputs in stimuli and and two small contributions from stimuli and , while the resilience of the cells in population is determined by 4 intermediate contributions . In our model ( 3 , 4 ) for the neuronal dynamics the granule cell activities are linear in the mitral cell activities . For the activity of the interfering population is almost the same for all four stimuli and is close to the average of the activity across the four stimuli . As a result , the rectifier , which makes the resilience function ( 8 ) concave , renders the granule cells that establish interfering connections less resilient than the granule cells connecting co-active mitral cells , . This suppresses the interfering population relative to , as is apparent in a comparison of Fig . 3Biii and Fig . 5B . Within the framework of the population formulation eqs . ( 15 , 16 , 17 ) the simplicity of the minimal stimulus set of Fig . 6A allows a detailed analysis of the role of the threshold in the balance between the suppression of interfering connections and a reduction of the beneficial inhibition of co-active mitral cells . Due to the symmetry of the stimulus ensemble only two granule-cell populations have to be analyzed , and . Their dynamics can be understood using a phase-plane analysis . For steep survival curves the nullclines of , which are defined by , are very well approximated by ( cf . Fig . 1Bi ) . Starting from , both population sizes increase linearly in time until they reach one of the two nullclines . Then the system follows slowly that nullcline until a fixed point is reached . This can be the intersection of the two nullclines ( Fig . 6Bi ) . In addition , since cannot become negative , an intersection of the nullcline with the axis also represents a fixed point if at that point ( Fig . 6Biii ) and similarly with the roles of and interchanged . A straightforward expansion shows that for highly similar stimuli , , the correlation between the two output patterns is given by ( 1 ) Thus , as expected , the correlation decreases with increasing reciprocal inhibition of co-active mitral cells and increases with increasing strength of the interfering connections . As discussed above , the relation between these two populations can be controlled using the threshold . For fixed resilience threshold the correlation is minimized for ( cf . Eqs . ( 23 , 22 ) ) ( 2 ) This is the smallest value of for which the interfering connections vanish , . Thus , it maximizes the inhibition between co-active mitral cells without inducing interference . This leads to optimal decorrelation , as is also apparent in the output activity patterns in the bottom panels of Fig . 6B . Thus , the threshold in the resilience suppresses interfering connections between mitral cells that are not co-active and promotes a connectivity that is based on co-activity . To provide a context of the performance of this co-activity based connectivity we compare the decorrelation achieved by the resulting networks with that obtained by a number of other types of adaptive networks . In some of them the inhibition is also based on co-activity , in others on distance , correlation , or covariance ( see Text S1 with Figs . S1 , S2 , S3 therein ) . We find that the networks whose adaptation mechanism is based on some form of co-activity of mitral cells or glomeruli are able to decorrelate representations of highly similar stimuli and achieve a reduction of the overall correlations without and with significant spontaneous mitral cell activity . Among these networks are networks motivated by an earlier model for neurogenesis [47] as well as networks that aim to orthogonalize the stimulus representations by orthogonalizing ( and normalizing ) the activity vectors of pairs of mitral cells [48] . Alternatively , the connectivities can also be based on the correlations or covariances of the inputs . For instance , a correlation-based connectivity was found to capture the outputs of the bee antennal lobe , which is the insect homolog of the olfactory bulb , better than random or local connectivities [49] . We find that correlation- and covariance-based recurrent networks do not decorrelate stimulus representations very well . In various situations they even tend to increase rather than decrease the correlations . This reflects , in part , the fact that they are not sensitive to the spontaneous activity of the mitral cells . Anatomically , the dendrodendritic synapses between mitral cells and granule cells are found to be predominantly reciprocal , i . e . each granule cell has inhibitory connections only to those mitral cells from which it receives excitatory connections [37] . In combination with the threshold this establishes effectively inhibitory lateral connections selectively between highly co-active mitral cells and allows the networks to decorrelate their highly correlated inputs . As implemented in our model so far , the reciprocal synapses not only provide an anatomical connection between co-active mitral cells but due to the homogeneity of the inhibitory synaptic weights they also induce a symmetric connectivity matrix and the amount of self-inhibition that a given mitral cell experiences is directly related to the amount of lateral inhibition it provides to other mitral cells . What roles do these different aspects play in the decorrelation ? To test the importance of the correct anatomical connections we redirect a fraction of the inhibitory connections of each granule cell to randomly chosen mitral cells instead of the mitral cells that drive that granule cell . As expected , as the fraction of such non-reciprocal synapses increases the correlations increase as well . Without any reciprocal synapses the network does not decorrelate the stimuli at all ( Fig . 7 ) . The network performance is , however , quite robust: the overall decorrelation deteriorates noticeably only when more than 50% of the connections have been rewired . The highly correlated stimuli are , however , more sensitive to the rewiring with increasing from to when 50% of the connections are rewired , while changes only from to . The granule cells deliver their inhibitory inputs onto the secondary dendrites of the mitral cells at highly variable distances from the mitral cell somata . Their effect on the mitral cell firing will therefore vary over quite some range; in fact , some synaptic contacts will be too far away from the mitral cell soma to have any noticeably effect on that mitral cell's firing . To assess the impact of such heterogeneities we modify the inhibitory synaptic weights , which so far had the same value for all synapses , by picking them with equal probability from the two values . This breaks the symmetry of the inhibition and for half of the inhibitory connections are completely ineffective . The overall decorrelation is , however , not affected by this heterogeneity and even the decorrelation of the highly similar stimuli deteriorates only slightly over the whole possible range ( Fig . 7B ) . Essentially the same result is obtained if the synaptic strengths are distributed uniformly in the interval . While for very large granule cell populations the heterogeneities of different granule cells are expected to average out each other , for the parameters used in our study the effective connectivity matrix is still noticeably asymmetric: its anti-symmetric component amounts to about 20% of the symmetric one . Through the reciprocal character of the dendrodendritic synapse a granule cell mediates lateral inhibition between the mitral cells that drive it as well as self-inhibition of each of them . Due to the complex dendritic dynamics of granule cells [50] , [51] these two types of inhibition can be of different strength . In fact , recent observations suggest that self-inhibition is significantly weaker than lateral inhibition [52] . While our minimal model does not capture any explicit dendritic processing , the strength of self-inhibition and lateral inhibition that a mitral cell receives is given by the diagonal and off-diagonal coefficients of the effective connectivity matrix , respectively . We can therefore change the balance between self-inhibition and lateral inhibition phenomenologically by rescaling the diagonal terms , with , at the expense of the off-diagonal terms , for , while keeping the row-sum of the matrix fixed through the normalizing factor . Reducing self-inhibition in this fashion ( ) enhances the decorrelation of the representations of the natural stimuli significantly ( Fig . 7C ) , because it further enhances the competition between dominant , co-active mitral cells . Conversely , increasing the self-inhibition ( ) reduces the competition . In the complete absence of lateral inhibition ( ) granule cells are effectively coupled only to a single mitral cell . This provides still good overall decorrelation , but the representations of the highly similar odors are only poorly decorrelated . Thus , the experimentally observed reduction of self-inhibition may contribute to an improved decorrelation performance of the bulbar network . These comparisons show that for effective decorrelation the most important aspect of the reciprocity of the dendrodendritic synapse is that it provides mutual anatomical connections between the relevant mitral cells , i . e . between those that are co-active for some stimuli . The effective synaptic strengths can be quite heterogeneous without compromising the performance of the network . In fact , reduced self-inhibition can enhance the decorrelation substantially . One possible role of neurogenesis is to provide a persistent supply of new neurons , which may play a different role than old , mature neurons . An aspect of this type has been identified in experiments focusing on the responsiveness of young and old adult-born granule cells [20] , [53] . In the experiments , adult-born precursor cells , which develop into granule cells , were marked in the subventricular zone . After they have migrated to the olfactory bulb and have integrated into the bulbar network their response to odor stimulation was measured using the expression levels of various immediate early genes . It was found that the fraction of adult-born granule cells that respond to novel odors is significantly higher shortly after their arrival in the olfactory bulb than a few weeks later . It has been argued therefore that one important function of the young granule cells may be to serve as novelty detectors [20] . In our computational model a differential response of young and older adult-born granule cells to novel odors arises quite naturally . After establishing a network by exposing the system to the stimulus ensemble , , we mark granule cells as they are integrated into the network and measure their response to various stimuli as a function of their age . Assuming that the granule cell activity has to surpass a minimal value to activate the expression of the immediate early genes , we consider granule cells as responding if they reach an activity above a threshold . As the network evolves the less active granule cells die and are removed from the network ( Fig . 8Ai ) . As in the experiments , we find that the fraction of young adult-born granule cells that respond to a novel stimulus , i . e . a stimulus that is quite different from the stimuli in the background ensemble , decreases as the granule cells become older ( Fig . 8Aii ) . This decrease results from the reduced survival probability of these cells , which is due to the weak drive they receive by the stimuli in the stimulus ensemble that determines granule-cell survival . In contrast , the fraction of granule cells that respond to a familiar stimulus , i . e . a stimulus in the background ensemble , decays very little or even increases over the same time frame , reflecting their higher survival rate . For what range of the threshold does our model yield results that agree qualitatively with the experiments in [20] ? When the threshold is increased beyond the resilience threshold ever fewer marked granule cells respond and the fraction of marked granule cells that respond to the stimuli - averaged over all stimuli - drops from 1 to 0 ( Fig . 8B bottom panel ) . Thus , the experimentally obtained response fractions of 10–20% [20] set an upper limit for relative to . At the same time , decreasing reduces the difference between the temporal evolution of the response to novel and to familiar stimuli . We characterize the evolution by the ratio between the fraction of granule cells responding to the stimulus at the final time of the simulation and the fraction immediately after the end of the marking period . On average this ratio increases with increasing for the familiar stimuli , but it decreases for the novel stimulus ( Fig . 8B top panel ) . For the response to novel odors to differ significantly from that to familiar odors cannot be much smaller than . It is worth noting that varying the steepness of the survival curve does not affect the decorrelation of the odor stimuli substantially ( symbols at in the top panel of Fig . 3 ) , but the difference in the response to novel compared to familiar odors is significant only if the survival curve is not too steep ( Fig . 8B top panel ) . Thus , the activity-dependent survival of the granule cells combined with their random connections to the mitral cells is sufficient to capture the experimentally observed enhanced response of young adult-born cells to novel stimuli if the threshold for the activation of the immediate early genes is close to the resilience threshold , which is an essential determinant of the survival of the granule cells . In a wide range of experiments possible connections between adult neurogenesis and animal performance have been investigated employing various tests of odor detection , odor discrimination , short-term and long-term memory , and fear conditioning [19] , [29]–[34] , [40] , [43]–[45] . No simple picture regarding the role of neurogenesis in odor discrimination and odor memory has , however , emerged so far . This may in part be due to the fact that higher brain areas are likely involved in many of the behavioral tasks; they may well compensate for some changes occurring in the olfactory bulb and therefore possibly mask certain effects of the neurogenesis . A behavioral task that may reflect bulbar odor representations relatively directly is the spontaneous , non-associative odor discrimination based on habituation , which has been shown to result predominantly from bulbar processes [54]–[56] . These experiments exploit the decreasing interest an animal typically displays to repetitions of the same stimulus: the animal's response to a second stimulus after if has habituated to a first stimulus is a measure of the degree to which the animal discriminates the two stimuli [55] . Exposing animals to extended periods during which their environment is enriched with additional odors enhances their spontaneous odor discrimination [40] , [43]–[45] . This is indicative of perceptual learning . The dominance of bulbar processing in this task [54]–[56] suggests that the enrichment induces changes in the bulbar odor representations [56] . Since the enhancement is significantly suppressed if neurogenesis is halted pharmacologically [40] , it is likely that the changes in the odor representations reflect a restructuring of the bulbar network . Importantly , for the enrichment to improve the performance the odors employed have to be related to the odors that are to be discriminated [43] . The perceptual learning observed in the experiments is captured in our minimal computational model . We use an ensemble of background stimuli , which establishes a default network connectivity , and test the performance of the network with two test stimuli ( -limonene and -limonene ) . They are not included in the stimulus ensemble that drives the network evolution . For the default network the correlation between the representations of the test stimuli is high , consistent with the fact that naive animals do not discriminate these odors spontaneously . Then the stimulus ensemble is enriched with additional odors ( ) . As the network adapts and evolves to a new steady state characterized by different effective connectivity matrices ( Fig . 9C , right panels ) , the correlation between the two test stimuli evolves as well . If the odors used for the enrichment have sufficient overlap with the test odors the correlation between the test odors decreases substantially ( red line in Fig . 9B ) . However , if the enrichment odors are only weakly related to the test odors the correlation of the test odors does not decrease ( black line in Fig . 9B ) . In fact , in some cases the correlation between the test odors can even increase . As expected , if the influx of new granule cells is stopped with the onset of the enrichment the odor representations and their correlations are unaffected by the enrichment , even if the enrichment odor is close to the test odor ( green line ) . In experiments , odor enrichment enhances the ability of the animals to discriminate similar odors only if there is sufficient overlap between the activation patterns of the stimuli used in the enrichment and those of the stimuli to be discriminated [43] . Our network model allows more specific predictions for the type of enrichment protocols that should be effective in enhancing the ability of the animals to discriminate a given set of test odors . We consider the decorrelation of very similar mixtures comprised of dissimilar components . Specifically , we use as components limonene ( 50% –limonene and 50% –limonene ) and carvone ( also both enantiomers in equal proportions ) , whose activation patterns have very little overlap ( Fig . 2A ) . We employ two different enrichment protocols . In the first one pure limonene and pure carvone are added to a background of alcohols and acetic acid in an alternating fashion ( Fig . 10B , top panel ) . Experimentally , this would correspond to presenting limonene and carvone separately at different times . In the second protocol an equal mixture of limonene and carvone is added to the background ensemble ( Fig . 10B , bottom panel ) . In both protocols the activity-dependent removal of interneurons occurs only after the complete set of background and enrichment stimuli has been presented . To implement the mixtures in the model we assume that the glomerular activation patterns for mixtures are approximated sufficiently well by a linear combination of the patterns for the individual components . While using the pure components in the enrichment decreases the correlation between the representations of the mixtures at all mixture ratios ( ‘alternating’ in Fig . 10Di ) it does so substantially less than the network enriched with the 50∶50 mixture ( ‘mixture’ in Fig . 10Di ) . The stronger decorrelation obtained with the mixture protocol compared to the alternating protocol can also be recognized directly in the output patterns ( Fig . 10C , bottom vs . top panel ) . This substantial difference arises because the decorrelation of the mixtures is strongly enhanced by mutual inhibition between mitral cells that are driven by limonene as well as carvone . This inhibition is provided by ‘mixed’ granule cells . By contrast , ‘pure’ granule cells are connected to mitral cells that are activated ( almost ) exclusively by limonene or carvone . As discussed in Sec . Threshold Promotes Lateral Inhibition Based on Co-Activity , in the context of interference , in the alternating protocol the mixed granule cells have a lower survival probability than the pure granule cells . In the mixture protocol , however , both types of granule cells have very similar survival probabilities . To assess the inhibition provided by these populations we consider the sum of the synaptic weights in the four quadrants of the effective connectivity matrix ( cf . Fig . 3Biii ) . We find that the inhibition provided by the mixed granule cells in the mixture protocol is stronger than in the alternating protocol ( see ( 27 ) ) . Insight into what controls these populations can be gained by considering again the simple caricature of Fig . 6A . Within that framework the mixture protocol can be viewed as a stimulus set in which all four glomeruli receive essentially equal input . Both types of granule cells have then equal survival rates . Within that model it is easily seen that the size of their populations falls between that of the mixed granule cells and the pure granule cells in the alternating protocol because the total resilience of the mixed granule cells has to be the same in both protocols ( see Sec . METHODS , Alternating vs Mixture Protocol ) . Thus , compared to the alternating protocol the mixture procotol enhances the relevant inhibition and improves the decorrelation of the limonene-carvone mixtures . If neurogenesis were to affect only interneurons that provide non-topographic inhibition and no lateral inhibition [4] both enrichment protocols would be expected to lead to the same level of decorrelation . Specifically , if in the model each granule cell makes only connections with a single mitral cell the alternating protocol leads to the same decorrelation of the limonene-carvone mixtures as the mixture protocol ( Fig . 10Dii ) . Comparing the influence of the two enrichment protocols on the animals' ability to discriminate such mixtures may therefore give insight into the type of neurogenesis-dependent connectivity that dominates the decorrelation mechanism . Thus , even though in both protocols the enrichment odors - taken together - have the same overlap with the test odors the model predicts that enrichment with the mixture protocol achieves substantially better decorrelation of the test stimuli than the alternating protocol .
It has been observed that young granule cells are more likely than mature ones to respond to odors that are novel for the animal [20] , [53] . This has been interpreted as a mechanism for novelty detection . Our model captures the enhanced response of young cells in a natural way . Since granule cells that respond to novel odors but not to the odors in the ongoing environment receive only little ongoing input , they do not survive for a long time and the fraction of granule cells responding to the novel odor decreases with their age . Thus , the observation of an enhanced response of young granule cells to novel odors suggests that new granule cells do not have a strong bias towards connecting to highly active mitral cells but connect also to mitral cells that have only been weakly active in the past . Such a strategy enables the network to learn to process novel odors . Experimentally , the response of the granule cells was measured in terms of the expression of various immediate early genes ( c-fos , c-jun , EGR-1/zif-268 ) . The fraction of granule cells responding to the novel odors was found to be 10–25% for young cells and lower for older cells [20] , [53] . Such an intermediate response fraction is obtained in our model if the threshold for the expression of the immediate early genes is close to that for the survival of the granule cells . This is suggestive of a common step in the pathways controlling IEG-expression and cell survival . The decorrelation of highly similar stimuli like the two pairs of enantiomers used in our computation hinges upon the presence of an activity threshold that the granule cells have to surpass to increase their survival probability . It enhances the connections between mitral cells that are highly active simultaneously and suppresses those between mitral cells that are strongly active albeit only in response to different stimuli . Biophysically , a threshold for the survival of the granule cells may arise from the need to drive L-type Ca channels , which activate the MAPK pathway that leads to the stimulation of genes that are essential for neuronal survival [57] , [58] . With the strengthening of inhibition between co-active mitral cells the mechanism underlying the adaptation in our model is somewhat related to that underlying other adaptive networks that have been studied previously . In an early neurogenesis model for the olfactory bulb the evolution of the effective pairwise inhibition between mitral cells was based directly on the scalar product of the mitral cell activities [47] . Adaptive networks that aim to orthogonalize the stimulus representations can do so via a connectivity that is based on the pairwise scalar products of input activities [48] . A somewhat different adaptive connectivity has been suggested in a modeling study of the bee antennal lobe . There it was found that a connectivity in which the inhibition is proportional to the correlations between the glomerular activities was able to match the observed output patterns better than random or local center-surround connectivities [49] . We have compared a few types of networks that exploit different adaptation algorithms and find that connectivities that are based on the co-activity of mitral cells or glomeruli achieve significantly better decorrelation than networks based on the correlations or covariances of the inputs . A particular problem of the latter algorithms is that they are not sensitive to mean activities of the cells and do not take the spontaneous activity of the mitral cells adequately into account . An anatomically characteristic feature of the olfactory bulb is the reciprocal nature of the dendrodendritic synapses between mitral and granule cells . The purpose of this reciprocity is not well understood . Our computational modeling shows that it can play an essential role in exploiting the activity-dependent survival of the granule cells to establish a connectivity whose lateral inhibition reflects the co-activity of the mitral cells . This provides a mechanism for the network to learn to decorrelate even highly similar stimuli . Biologically , the reciprocity may be imperfect in a number of ways . In principle , an inhibitory synapse could connect the granule cell to a mitral cell that is not the origin of the associated excitatory synapse . Modeling such a situation by a random rewiring of a fraction of inhibitory connections we find that the network performance is reasonably robust to such perturbations . However , when more than 50% of the synapses are rewired the performance deteriorates significantly and without any reciprocity the stimulus representations are not decorrelated at all . A second type of imperfection of the reciprocity is likely to arise if the dendrodendritic synapse is located far from the soma of the mitral cell . In such a case the inhibition exerted by the granule cell may not have much effect on the mitral cell firing , although the granule cell is driven strongly by that mitral cell . This asymmetry can arise because excitation is driven by action potentials , which can travel long distances along the dendrite , whereas the shunting provided by the inhibition is confined to a distance comparable to the electrotonic length of the dendrite [59] . Thus , the effective inhibitory strength may vary substantially between synapses depending on their location relative to the soma . Mimicking such a heterogeneity by random variations in the synaptic strength we find that the network performance is only moderately affected by such effects . Since mitral cells are connected to many granule cells the heterogeneity of the combined synaptic strengths is likely to be reduced compared to the heterogeneities within individual granule cells . Such an averaging may be reduced if correlations between the strengths of different synapses , which may arise due to correlations in the physical distances between the cells , should be significant . The reciprocity may also be perturbed because the strength of the self-inhibition that a mitral cell experiences on account of a given granule cell may differ from that of the lateral inhibition that said granule cell provides to other mitral cells . In fact , recent experiments suggest that self-inhibition is significantly weaker than lateral inhibition [52] . One cause for this difference may be the complex physiology of granule cells , which includes local dendritic calcium signaling , dendritic calcium spikes , and action potentials driven by sodium conductances [51] . Our minimal single-compartment model for the granule cell does not allow to capture these rich dynamics . However , on a phenomenological level the balance between self-inhibition and lateral inhibition can be modified by rescaling the diagonal and off-diagonal terms in the effective connectivity matrix . Our model shows that reducing the self-inhibition while strengthening the lateral inhibition can substantially enhance the ability of the network to decorrelate the representations of highly similar stimuli . The decorrelation of similar stimulus representations that is obtained in our model provides a natural interpretation of recent experiments on spontaneous odor discrimination via habituation [40] . Only with neurogenesis intact does enriching an animal's odor environment enhance its ability to discriminate similar odors . Since the habituation used in these discrimination experiments reflects predominantly changes in the olfactory bulb rather than higher brain areas [54]–[56] , the improvement in odor discrimination resulting from odor enrichment likely reflects modifications in the encoding of the test stimuli in the olfactory bulb . Our modeling shows that fundamental features underlying the neuronal turn-over in the bulb – activity-dependent survival and reciprocal synapses – suffice to allow perceptual learning by changing the odor encoding so as to decrease their similarity and enhance their discriminability . In laboratory experiments that allow many repetitions animals can learn to discriminate highly similar odor stimuli [31] , which may have highly correlated representations in the olfactory bulb . Outside the laboratory the animals are likely to face the challenge to form associations with stimuli given only a few trials . This task may be very difficult if not even impossible if the relevant odors are represented in the bulb in a highly correlated fashion . In line with experiments on odor enrichment [40] , our computational model shows that neurogenesis may facilitate this task by reducing the correlation of odors in an ensemble to which the animal is exposed . These odors could represent a repertoire of potentially relevant odor types that the animal can easily discriminate , should the need arise . In our model the survival of the granule cells depends on the inputs they receive from mitral cells via their dendro-dendritic synapses . Their relevance could be determined by the context in which the animal is exposed to the odor . Such contexts are likely to affect modulatory inputs to the olfactory bulb , which can modify the excitability of granule cells [60]–[62] and mitral cells [60] , [62] as well as mitral cell inhibition [63] , all of which will affect granule cell survival [64] . Contexts could also induce specific , direct inputs from cortical areas like piriform cortex to granule cells at proximal or basal synapses , both of which are functional in young granule cells with the proximal synapses developing even before the basal and dendro-dendritic ones [65] , [66] . Enrichment enhances odor discrimination only if the enrichment odors overlap in their glomerular excitation patterns with those of the test stimuli [43] . Our modeling confirms this . Moreover , it makes specific predictions with regard to the decorrelation of similar mixtures comprised of dissimilar components . If neurogenesis affects predominantly granule cells that provide lateral inhibition , our model predicts that animals will learn to discriminate such mixtures more easily if the enrichment is performed using the odor mixture rather than alternating its individual components . This difference is predicted even though both enrichment ensembles have the same overall overlap with the test stimuli . If non-topographical self-inhibition [4] should dominate neurogenetic restructuring , no difference between the protocols is expected . The change of odor representations that our neurogenesis model predicts to arise from odor enrichment might also be testable in reward-associated discrimination tasks by focusing on the initial learning stages . Suitable enrichment protocols are expected to enhance the differences in the encoding of the test odors . Applied before the animals learn the odors that are to be discriminated , such enrichment should lead to a shortening of the initial learning phase if the odors are very similar . Moreover , it has been found that animals tend to follow different strategies during the early stages of a 2-alternative choice odor discrimination task depending on the degree of similarity of the two odors [67] . In fact , for very similar stimuli their early strategy suggests that they actually can not yet tell the test odors apart . In that case suitable prior enrichment may even allow the animals to employ their coarse-discrimination strategy for odor pairs for which without enrichment they would use the fine-discrimination strategy . Divisive response normalization has been discussed extensively in sensory processing , in particular in the visual system [68] . In this type of normalization the response of each cell , which corresponds to a channel with given response characteristics like preferred orientation or spatial frequency of visual grid patterns , is divided by the sum of the activities of cells covering a wider range of response characteristics . The gain control implemented by this process is consistent with various experimentally observed neural responses ( e . g . contrast independence and contrast adaption ) [68] . In olfaction it has been proposed that such a normalization may arise from the lateral inhibition provided by the network of peri-glomerular cells , short-axon cells , and external tufted cells in the glomerular layer of the olfactory bulb [69] . Divisive normalization has been observed in the antennal lobe , which is the insect analogue of the olfactory bulb [70] . Implemented in simulations by global lateral inhibition , it was found to reduce the correlation between the different channels ( activities of the principal neurons ) across a large set of odors [70] , [71] . Further analysis of our neurogenetic model suggests that the olfactory bulb performs a complementary type of divisive normalization ( unpublished data ) . Rather than reshaping the mitral cell activities such that their pattern average is the essentially the same for all stimuli , the activity-dependent survival of the granule cells tends to equalize ( normalize ) the activity of all mitral cells when averaged across the stimulus set . Correspondingly , it foremost contributes to a reduction of the correlations between pairs of activity patterns rather than between pairs of mitral cells ( channels ) . For stimuli whose similarity is dominated by highly co-active mitral cells the normalization of the activity of individual mitral cells achieves , however , only quite limited decorrelation . The joint normalization of the activities of multiple mitral cells , which results from the lateral inhibition of granule cells connected to multiple mitral cells , can preserve some differences in the mitral cell activities and , as a consequence , can achieve considerably better pattern decorrelation . In our minimal model we have focused on the impact of the structural plasticity afforded by the turn-over of the granule cells . We have therefore treated the individual mitral and granule cells in a minimalistic fashion . In particular , we have described them in terms of linear rate dynamics without any threshold . Previous studies have shown that nonlinearities can induce stimulus decorrelation even in non-adaptive networks [3] , [4] , [7] . An interesting question is therefore whether neuronal nonlinearities could further enhance the decorrelation achieved by the adaptive networks studied here . Moreover , we have modeled each neuron as a single compartment . Both , mitral and granule cells , have , however , elaborate dendrites , which most likely increase the complexity of their interaction . Thus , while action potentials can propagate with little attenuation along the mitral cell dendrite and can excite granule cells even at large distances , the inhibition that a granule cell imparts to a mitral cell is expected to depend strongly on the distance of the GABAergic synapse from the mitral cell soma . The mutual inhibition that a granule cell mediates between mitral cells will then not be symmetric . We have mimicked such an asymmetry by modifying the effective connectivity matrix and found that the network performance is quite robust with respect to such perturbations . The dendritic computations in the granule cells are tied in with their complex multi-level calcium dynamics [50] , [51] . Even quite small depolarizations of a granule cell spine can induce local GABA release , which results in graded self-inhibition of the driving mitral cell . Stronger inputs can induce low-threshold calcium spikes that can spread within the dendritic tree . Finally , suitable inputs can trigger somatically evoked conventional sodium-driven action potentials that invade the whole dendrite . This complexity may endow the granule cell with additional computational power like a dynamically regulated range of inhibition . Since the ability to generate sodium spikes develops last in adult-born granule cells [13] , the balance between local signaling , calcium spikes , and sodium spikes may change with the age of the cell . While our phenomenological modeling does not capture this biophysical complexity , it shows that a reduction of the self-inhibition and a concomitant enhancement of lateral inhibition can substantially improve the decorrelation of stimulus representations . The mechanisms controling the survival and apoptosis of granule cells are not understood in detail . It is known that larger fractions of granule cells survive if the animal is kept in odor-enriched enviroments [19] or if the excitability of the granule cells is genetically enhanced [41] . In our minimal model we therefore assumed that the survival of the granule cells increases with their activity . It has been found , however , that certain associative odor discrimination tasks can not only enhance but also reduce the survival of the adult-borne granule cells , depending on their age [23] . Recent experiments have also indicated that apoptosis of specific neurons can be elevated when associative memories are erased [72] . It would be interesting to extend our minimal model , which aims to capture the impact of neurogenesis on non-associative odor discrimination tasks [40] , [44] , [45] , to such more complex situations . In conclusion , using a minimal computational model we have shown that adult neurogenesis with activity-dependent apoptosis of inhibitory interneurons that are reciprocally connected with the principal neurons is sufficient to restructure a network like that of the olfactory bulb such that it learns to decorrelate representations of very similar stimuli . The network performance is quite robust with respect to various types of deviations from reciprocity that are likely to be present in the olfactory bulb . The model makes predictions regarding the impact of different enrichment protocols on the performance of animals in spontaneous and award-related odor discrimination tasks . Their outcome is expected to give insight into the type of network connectivity that is associated with the interneuronal turn-over .
We consider a minimal computational network model that focuses on the turn-over of inhibitory interneurons caused by neurogenesis and activity-dependent apoptosis and study the networks' ability to learn to decorrelate similar stimulus representations . The recurrent network comprises two types of neurons , principal neurons ( mitral cells ) and inhibitory interneurons ( granule cells ) , which are coupled through reciprocal synapses ( Fig . 1 ) . Within the framework of threshold-linear rate equations the activity of mitral cell in response to stimulus with afferent activity pattern , , and the corresponding activity of granule cell , , satisfy ( 3 ) ( 4 ) with the rectifier defined as ( 5 ) Here denotes the spontaneous activity of the mitral cells [46] , in the absence of any odor stimulus and without any inhibitory inputs from granule cells . The stimuli are taken from a stimulus ensemble . Throughout this paper we consider only the steady states that the mitral and granule cell activities reach in response to long stimulus presentations . The temporal evolution that we discuss is that of the network connectivity , which has a time scale that is much slower than that of the neuronal activities . For strong inhibition and in particular for asymmetric connectivity matrices ( cf . RESULTS , Imperfect Reciprocity of Synapses is Sufficient ) the steady states of ( 3 , 4 ) may become unstable . While we did find complex eigenvalues in the spectrum of the linear operator of ( 3 , 4 ) , which correspond to oscillatory modes , in none of the cases we considered did any of the eigenvalues have positive real parts . Thus , the steady output patterns remained stable for all parameters values we considered . For sufficiently large spontaneous activity essentially all mitral cell activities – and with them the granule cell activities – are positive and the results are changed only slightly if the rectified coupling Eq . ( 5 ) is replaced by a linear coupling . We therefore use in the following only the linear coupling , which reduces the computational effort substantially . The restructuring of the network due to neurogenesis is implemented by adding granule cells to the network at a steady rate and removing them with an activity-dependent probability . No detailed information is available to what extent the formation of synapses between granule and mitral cells depends on their activity or the previous presence of synapses at that location . Since the secondary dendrites of the mitral cells , onto which the granule cells synapse , extend over large portions of the olfactory bulb we assume in this minimal model that each granule cell has the potential to establish a connection with any of the mitral cells . Thus , we assume that each of the new granule cells connects to randomly chosen mitral cells . Consistent with observations [74] we assume that no granule cell connects to any mitral cell twice . A characteristic feature of the dendro-dendritic synapses connecting mitral and granule cells is the prevalent juxtaposition of a glutamatergic synapse onto the granule cell and a GABAergic synapse onto the mitral cell [37] . There are indications that the glutamatergic and the GABAergic component form at very similar times [65] , with possibly the glutamatergic component formed somehwat earlier [75] . Anatomically , the connections between these two cell types are therefore predominantly reciprocal . This reciprocity is important for the ability of the resulting network to decorrelate stimulus representations . It implies ( 6 ) with if granule cell is receiving input from mitral cell and otherwise . Due to the linearity of the neuronal dynamics the effective connectivity matrix is given by ( 7 ) Focusing on the structural plasticity provided by the persistent turn-over of granule cells rather than any plasticity of the synapses [76] , [77] , we assume in most of this work that all inhibitory synapses have fixed equal strength and all excitatory synapses have strength 1 . To probe the role of reciprocity we consider in Sec . RESULTS; Imperfect Reciprocity of Synapses is Sufficient also connectivities violating ( 6 ) and heterogeneities in the synaptic strength . We model activity-dependent apoptosis of the granule cells [41] by discrete events [78] during which the survival of any given granule cell is assessed based on the history of its activity . The duration of the time interval over which the activity influences cell survival is currently not known . We assume that it is long enough for the animal to be exposed to a number of relevant odors defining a stimulus ensemble S . Thus , at each of these events , which we assume for simplicity to occur regularly in time defining a time step of length , granule cells are removed probabilistically . Their survival probability is taken to depend in a sigmoidal fashion on their cumulative , thresholded activity across the stimulus ensemble . Introducing the resilience of granule cell via ( 8 ) with a resilience threshold , we take ( 9 ) with a soft survival threshold . Since little is known about the specifics of the survival probability we take here and . In this model a granule cell has to reach an activity beyond at least for some of the stimuli in the ensemble in order to trigger the signaling pathway that controls its survival [41] , [57] , [58] . The probabilistic network evolution eventually leads to a statistically steady state as characterized by the output patterns and their correlations fluctuating around constant values . The magnitude of the fluctuations decreases with an increase in the overall number of granule cells in the system , which can be achieved by a suitable decrease in the synaptic weight . Fig . 9B shows the typical size of fluctuations in the correlation for the parameters used in our study . The network evolution is self-regulated by the balance between proliferation and apoptosis: with increasing granule-cell population the overall inhibition of the mitral cells increases , leading to a reduction in granule cell activity . This lowers the survival probability of the granule cells and provides the saturation of the network size . As observed experimentally , reduced odor stimulation leads to a reduction in the size of the granule cell population [15] . We quantify the reshaping of the stimulus representations by the resulting network using the Pearson correlation of patterns ( 10 ) where . The average correlation is given by ( 11 ) To capture certain aspects of the network evolution analytically we also consider the weak-coupling limit , . The number of granule cells is then large and the network restructuring can be described in terms of differential equations for the mean size of the various populations of granule cells that have established the same connections with mitral cells . For simplicity we give here only the equations for two-connection networks in which each granule cell makes connections with two mitral cells . The probability for the population of granule cells connecting mitral cells and to have size evolves during a small time step according to ( 12 ) where is the fixed influx of new granule cells and is the removal rate . With giving the probability for a granule cell to survive for the duration , the removal rate is given by ( 13 ) Here we have used that different cells are removed independently of each other . The resilience is given in terms of the activity of the granule cells analogous to Eq . ( 8 ) . For large mean population size the probability distribution , t ) will be sufficiently sharply peaked to allow to approximate the evolution equation for , ( 14 ) by ( 15 ) Here and in the following we drop the brackets indicating the mean value . The steady-state neuronal activities are given by ( 16 ) ( 17 ) Note that for networks with a realistic number of mitral cells the number of possible different granule cell populations is extremely large , much larger than the total number of granule cells in the olfactory bulb . Thus , the size of most populations will be small and fluctuations in the number of granule cells , which have been neglected in the population description eqs . ( 15 , 16 , 17 ) , may become relevant . The main purpose of the population formulation is to allow analytical approaches for simple cases , which can provide insight that may be hard to extract from numerical simulations of the discrete model . When interpreting the analytical results the limitations of the formulation need to be kept in mind . For analytical calculations considering a steep sigmoid , , for the survival probability in Eq . ( 15 ) is particularly attractive . The steady state of the population description eqs . ( 15 , 16 , 17 ) can then be analyzed quite easily because the nullcline for the population , which is defined by , is then very well approximated by since switches quickly from to as passes through . To obtain analytical results for the threshold that minimizes interfering connections between mitral cells that are strongly active but only during the presentation of different stimuli we consider a set of four stimuli activating four glomeruli , ( 18 ) ( 19 ) The symmetry of this stimulus ensemble has been chosen such that for networks in which each granule cell connects to two mitral cells only two granule-cell populations have to be analyzed , and . Independent of the values of the thresholds the remaining populations are given byFor these stimulus pairs are highly correlated . We consider their reshaping by networks that are trained using the slightly simplified ensemble with . The approximate nullclines and for the evolution of the two populations and are then given by ( cf . Eq . ( 8 ) ) ( 20 ) ( 21 ) Without loss of generality we have absorbed into the definition of . Depending on the system has two fixed points . For one has ( 22 ) where is given by ( 23 ) For the fixed point is given by ( 24 ) Thus , the interference induced by population vanishes for , while the inhibition of the co-active cells starts to decrease at Since the correlation decreases with decreasing but increases with decreasing it is minimal for . Two comments regarding the solution ( 24 ) with are in order . The nullclines are given by ( 20 , 21 ) only in the limit For finite values of corrections arise that render non-zero ( cf . ( 15 ) ) . Moreover , the description of the granule cell populations solely in terms of their mean values requires that the means are sufficiently large . In particular , since the population is always non-negative its mean cannot strictly vanish . Nevertheless , for small influx and large the population will become very small as is increased beyond . A simple model with four stimuli and four glomeruli can also be used to obtain insight into the difference between the alternating stimulus protocol and the mixture protocol of Sec . RESULTS; Effective Enrichment: Overall Overlap is Not Sufficient . To mimic the alternating protocol we use stimuli ( 18 , 19 ) with and for the training with the mixture protocol we use . For this protocol all granule cell populations are equal , which we denote by . The nullcline determining is given by ( 25 ) For comparison with in the alternating protocol ( 21 ) gives , implying ( 26 ) where and are the granule cell populations given by ( 22 ) . Thus , within this simple model the mixture protocol induces stronger inhibition than the alternating protocol between the first and second pair of mitral cells , . This inhibition enhances the decorrelation of stimuli like and , which mimic the test stimuli of Sec . RESULT; Effective Enrichment: Overall Overlap is Not Sufficient . This relationship among the populations is also found in the simulations of the full discrete model of Fig . 10 . Excluding the terms on the diagonal , which provide self-inhibition , the sums of the synaptic weights in the four quadrants of the effective connectivity matrix are found to be ( 27 ) The terms above the braces indicate which population is considered to correspond to which block . Note that each quadrant contains many connections that are not specific to limonene or carvone ( cf . Fig . 9 ) ; neither will they be affected much by the difference in the protocol nor will they contribute substantially to the discrimination of the test stimuli . To test the ability of the model network to decorrelate stimulus representations we use an ensemble of stimuli modeled after the activation patterns in the glomerular layer of rat that have been obtained experimentally via -deoxyglucose uptake in response to odor exposure ( published in the Glomerular Activity Response Archive http://gara . bio . uci . edu/ , cf . [42] ) . In these data the individual glomeruli have not been identified . Clearly , not each of the pixels corresponds to a glomerulus . We have down-sampled the experimentally determined pixel patterns to 424 input channels ( or 50 channels in cases in which we illustrate the connectivity ) , and take each channel as a proxy for a glomerulus . In the down-sampling we avoid excessive smoothing of the resulting patterns by retaining in each set of adjacent pixels the highest value rather than their average ( Fig . 2A ) . The stimulus set includes 2 pairs of enantiomers , -limonene and -carvone , which are difficult to discriminate . Specifically , without training mice do not discriminate between the two enantiomers of limonene [40] . When addressing the ability of the model network to learn to decorrelate highly similar stimuli we focus on these 4 stimuli . In addition , to mimic a background odor environment we include four additional stimuli , 1-butanol , 1-hexanol , 1-heptanol , and acetic acid . | The olfactory bulb is one of only two brain regions in which new neurons are added persistently in substantial numbers even in adult animals . This leads to an ongoing turnover of interneurons , in particular of the inhibitory granule cells , which constitute the largest cell population of the olfactory bulb . The function of this adult neurogenesis in olfactory processing is only poorly understood . Experiments show that it contributes to perceptual learning . We present a basic computational model that is built on fundamental aspects of the granule cells and their connections with the excitatory mitral cells , which convey the olfactory information to higher brain areas . We show that neurogenesis can reshape the network connectivity in response to olfactory input so as to reduce the correlations between the bulbar representations of even highly similar stimuli . The neurogenetic adaptation of the stimulus representations provides a natural explanation of the perceptual learning and the different response of young and old granule cells to novel odors that have been observed in experiments . The model makes experimentally testable predictions for training protocols that enhance the discriminability of odor mixtures . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"computational",
"neuroscience",
"biology",
"computational",
"biology",
"sensory",
"systems",
"neuroscience"
] | 2012 | Neurogenesis Drives Stimulus Decorrelation in a Model of the Olfactory Bulb |
One of the crucial steps in endochondral bone formation is the replacement of a cartilage matrix produced by chondrocytes with bone trabeculae made by osteoblasts . However , the precise sources of osteoblasts responsible for trabecular bone formation have not been fully defined . To investigate whether cells derived from hypertrophic chondrocytes contribute to the osteoblast pool in trabecular bones , we genetically labeled either hypertrophic chondrocytes by Col10a1-Cre or chondrocytes by tamoxifen-induced Agc1-CreERT2 using EGFP , LacZ or Tomato expression . Both Cre drivers were specifically active in chondrocytic cells and not in perichondrium , in periosteum or in any of the osteoblast lineage cells . These in vivo experiments allowed us to follow the fate of cells labeled in Col10a1-Cre or Agc1-CreERT2 -expressing chondrocytes . After the labeling of chondrocytes , both during prenatal development and after birth , abundant labeled non-chondrocytic cells were present in the primary spongiosa . These cells were distributed throughout trabeculae surfaces and later were present in the endosteum , and embedded within the bone matrix . Co-expression studies using osteoblast markers indicated that a proportion of the non-chondrocytic cells derived from chondrocytes labeled by Col10a1-Cre or by Agc1-CreERT2 were functional osteoblasts . Hence , our results show that both chondrocytes prior to initial ossification and growth plate chondrocytes before or after birth have the capacity to undergo transdifferentiation to become osteoblasts . The osteoblasts derived from Col10a1-expressing hypertrophic chondrocytes represent about sixty percent of all mature osteoblasts in endochondral bones of one month old mice . A similar process of chondrocyte to osteoblast transdifferentiation was involved during bone fracture healing in adult mice . Thus , in addition to cells in the periosteum chondrocytes represent a major source of osteoblasts contributing to endochondral bone formation in vivo .
Long bones , ribs , vertebrae , and other parts of the vertebrate skeleton are formed through a precisely synchronized process known as endochondral ossification . The highly complex endochondral bone tissue , which is generated through a cartilage intermediate , consists of multiple types of cells , including mesenchymal-derived chondrocytes , osteoblasts and osteocytes , as well as osteoclasts and bone marrow cells , which have a hematopoietic origin [1] . Endochondral bone is made of an outer compact bone ( cortex ) and an inner spongy bone tissue within the bone marrow cavity . The conversion from the nonvascular cartilage template to fully mineralized endochondral bones proceeds in distinct and closely coupled steps [2] . The first step is initiated when chondrocytes in the center of the cartilage models undergo hypertrophic differentiation and cells in the perichondrium surrounding the hypertrophic zone differentiate into osteoblasts to form the interim bone cortex ( bone collar ) . Concurrently , the initial vascular invasion occurs in the same region importing blood vessel-associated pericytes , osteoclasts and progenitor cells in the circulating blood . Immediately following the onset of bone collar formation , hypertrophic chondrocytes and the mineralized cartilage matrix in the center of the cartilage template are replaced by a highly vascularized trabecular bone tissue as well as bone marrow . Bone trabeculae in the primary spongiosa are formed by deposition of osteoid by osteoblasts on the surface of calcified cartilage spicules . Until recently , the precise origins of the trabeculae-producing osteoblasts had remained largely undefined . A lineage tracing study using tamoxifen-inducible CreER transgenic mice harboring a ROSA26R conditional allele showed that Osx-expressing immature osteoblast precursors , labeled in the perichondrium before the development of the primary ossification center , were able to migrate into the cartilage along with blood vessels and were responsible , at least in part , for trabeculae formation [3] , [4] . Besides perichondrium cells , several other cellular sources have also been examined as potential candidates to account for the formation of the primary spongiosa , including pericytes associated with the invading blood vessels , bone marrow progenitor cells and hypertrophic chondrocytes [4] . During endochondral ossification , terminally differentiated hypertrophic chondrocytes are eventually completely removed from the initial cartilage template or growth plates . Some of these chondrocytes have been shown to be eliminated through either apoptosis or autophagy ( type II programmed cell death ) [5] , [6] , [7] , [8] . However , the fate of the hypertrophic chondrocytes not accounted for by either of these processes of cell death continues to be debated . It has been shown that hypertrophic chondrocytes also express osteoblast markers , such as Alkaline Phosphatase ( ALPL ) , Osteonectin ( SPARC ) , Osteocalcin ( BGLAP ) , Osteopontin ( SPP1 ) and Bone sialoprotein ( IBSP ) , implicating potential complex functions of these cells [5] , [6] . Indeed , a number of studies have reported that in cell cultures containing ascorbic acid or in organ cultures , hypertrophic chondrocytes , instead of becoming extinct , resume cell proliferation and undergo asymmetric cell division , giving rise to cells with morphological and phenotypic characteristics of osteoblasts capable of producing a mineralized bone matrix in vitro [6] , [7] , [8] , [9] . In addition , histological experiments suggested that in embryonic chicks , long bone chondrocytes differentiated to bone-forming cells and deposited bone matrix inside their lacunae [10] , further promoting the notion of a direct role of hypertrophic chondrocytes in trabecular bone formation . However , most of the evidence supporting transdifferentiation was based on either histological observations during skeletal development or on in vitro studies [11] . Overall , it was suggested that these studies were not fully conclusive ( 3 ) . The result of an earlier ex vivo experiment that modified mouse embryonic limb tissue was consistent with a hypothetical transdifferentiation of chondrocytes into osteoblasts but the cells were not further characterized [12] . However , the conclusions of two more recent lineage tracing studies did not support a contribution of mature chondrocytes to the osteoblast/osteocyte pool in the central metaphyseal regions below the growth cartilage [3] , [13] . Mature osteoblasts develop from Runx2-expressing osteoblast precursors that are derived from mesenchymal progenitors [14] , [15] . Osterix ( Osx , Sp7 ) is a key transcription factor required for the full differentiation of Runx2-expressing osteoblast precursors into mature osteoblasts and osteocytes [16] . Osx is expressed in osteoblasts and osteocytes but also , at a lower level , in prehypertrophic and hypertrophic chondrocytes and in bone marrow mesenchymal progenitor cells during and after embryonic development [17] . Inactivation of Osx during and after embryonic development completely arrested osteoblast differentiation and bone formation [16] , [17] . The purpose of this study was to examine whether hypertrophic chondrocytes may acquire an osteogenic fate in vivo . We specifically labeled either hypertrophic chondrocytes with a Cre recombinase driven by a Collagen 10 BAC transgene ( Col10a1-Cre ) [18] or chondrocytes with an inducible Cre recombinase , the DNA of which was inserted by knock-in in the 3′ UTR of the Aggrecan gene ( Agc1-CreERT2 ) [19] . Labeling occurred with either EGFP [20] , LacZ [21] or Tomato expression [22] . LacZ and Tomato expression were from conditional alleles in the ROSA locus . The EGFP DNA preceded by a LoxP site was inserted 3′ to the poly-A site of Osx whereas in this allele the other LoxP site was placed in the first intron of the Osx gene [23] . In mice harboring this allele , high expression of EGFP occurs only in Osx-expressing cells after the LoxP sites recombine ( S1A Figure ) [23] . Neither of the two Cres was expressed in the perichondrium or the periosteum of endochondral bones [18] , [19] . Upon recombination , ROSA26R reporter mouse expresses secreted β-galactosidase ( LacZ ) , ROSA-Tomato reporter mouse expresses cytoplasmic tandem dimer Tomato , and Osx floxed mouse expresses cytoplasmic EGFP . Whereas labeling of mature chondrocytes in mice harboring Col10a1-Cre occurred constitutively once its expression begun and persisted as long as the Col10a1 promoter remained active , the timing of labeling of chondrocytes by Agc1-CreERT2 was controlled by the administration of tamoxifen and this labeling period persisted for a short period . One advantage of the Osx/EGFP allele in cell fate experiments was that if one would detect non-chondrocytic cells expressing EGFP , these cells would likely be osteoblast lineage cells [16] , [23] . Our data show that labeled non-chondrocytic cells appeared in the primary spongiosa of Col10a1-Cre or of tamoxifen activated Agc1-CreERT2 embryos and mice . In the case of Col10a1-Cre embryos and in Agc1-CreERT2 embryos treated with tamoxifen earlier than E14 . 5 , these non-chondrocytic reporter+ cells started to appear at the onset of primary ossification . Later they were found throughout the primary ossification centers and subsequently in the endosteum and within the bone matrix . Their appearance could also be induced in the primary spongiosa postnatally . Many of these cells expressed the mature osteoblast marker Osteocalcin and exhibited osteoblast-specific Col1a1-promoter activity . Likewise , in tamoxifen treated Agc1-CreERT2 mice chondrocyte-derived reporter+ non-chondrocytic cells were present in the repair callus of fractured tibiae . Later these reporter+ cells , which were associated with the ossified bone matrix in the calluses , also displayed Col1a1-promoter activity . Our results provide in vivo evidence that chondrocytes , both in cartilage primordium and in established growth plates , as well as chondrocytes in bone repair calluses , have the capacity to transdifferentiate into osteoblasts and represent a major source of osteoblasts in endochondral bones .
In Col10a1-Cre transgenic mice , Cre recombinase activity was previously detected specifically in all hypertrophic chondrocytes starting from E13 . 5 throughout endochondral skeletal development and into the postnatal stage [18] . Here we further confirmed that in the femurs and tibias of E15 . 5 Col10a1-Cre; ROSA26R mice , only hypertrophic chondrocytes , not cells in the perichondrium and periosteum , were positive for LacZ ( S1B-a , b Figure . ) , indicating that Cre activity driven by the Col10a1 regulatory elements occurred specifically in hypertrophic chondrocytes in these Col10a1-Cre transgenic mice . This was also confirmed by in situ hybridization of Col10a1 and Cre mRNA which was only observed in the hypertrophic zone ( S1C Figure ) . To test the hypothesis that some of the Col10a1-expressing hypertrophic chondrocytes might transdifferentiate into osteoblasts , we crossed Col10a1-Cre with Osx flox/flox mice to generate Col10a1-Cre; Osx flox/+ embryos . In these embryos EGFP expression labels Osx-expressing cells , which either expressed Col10a1 or were derived from Col10a1-expressing cells ( Figure S1A ) . Immunofluorescence ( IF ) with anti-EGFP showed that in the femurs of Col10a1-Cre; Osx flox/+ embryos , abundant EGFP-positive ( EGFP+ ) cells were present throughout the primary ossification centers ( Fig . 1 ) , where only very few , if any , Col10a1- or Cre-expressing cells were detected ( S1C Figure ) . The appearance of these EGFP+ cells was concurrent with the onset of primary ossification at E15 . 5 ( Fig . 1A ) . These EGFP+ cells continued to be present in the primary spongiosa at E16 . 5 ( Fig . 1B ) , E18 . 5 ( Fig . 1C ) to after birth ( Fig . 1D ) . The EGFP expression levels of these cells increased from E15 . 5 to E18 . 5 in parallel to the increasing levels of Osx expression in the same regions . In the 2-week-old Col10a1-Cre; Osx flox/+ mice , EGFP+ cells were found throughout the trabecular surfaces , also lining the endosteum of the distal half of the femur , and even embedded within the bone matrix of the cortical bone and trabeculae ( Figure S1D , Fig . 1D ) . Such EGFP+ non-chondrocytic cells were completely absent in intramembranous skeletal elements such as the calvariae of E18 . 5 Col10a1-Cre; Osx flox/+ embryos ( Figure S1E ) . In addition to its expression in hypertrophic chondrocytes and osteoblasts in the primary spongiosa , Osx was also highly expressed in periosteum and perichondrium [16] . However , EGFP+ cells were only detected in primary ossification centers and , as to be expected at lower levels in the hypertrophic zone , not in periosteum and perichondrium . This indicated that these EGFP+ non-chondrocytic cells were cells in which the Osx floxed allele was recombined by Col10a1-Cre and that these cells were unlikely to be derived from periosteum or perichondrium cells ( Fig . 1 ) . Staining of EGFP+ hypertrophic chondrocytes was not strong enough to be clearly shown due to strong autofluorescence from mineralized cartilage and bone matrix and from blood cells in the bone marrow . Similarly , Col10a1-Cre induced Tomato+ cells were found throughout the primary spongiosa of Col10a1-Cre; ROSA-tdTomato embryos ( Fig . 2A , B and C ) in the same pattern as in the Col10a1-Cre; Osx flox/+ embryos . This result confirmed our observation that cells derived from Col10a1-Cre expressing hypertrophic chondrocytes were indeed present in the primary spongiosa . In the femurs of 6-month-old Col10a1-Cre; ROSA-tdTomato;Osx flox/+ mice , Tomato+ cells continued to be present , however , the number of these cells on the endosteum surface was reduced and these cells were distributed more evenly from endosteum to periosteum within the cortex ( S1F Figure ) . By 8 months , there were almost no Tomato+ cells present in the primary spongiosa . The Tomato+ cells were distributed mostly within the bone matrix or in the growth plate . Besides a few strong Tomato+ chondrocytes in the growth plate , the fluorescence intensity of a majority of Tomato+ cells were much weaker than in the younger mice ( Figure S1G ) . To verify the presence in the primary spongiosa of reporter+ cells that were labeled by Col10a1-Cre , and to further delineate the origin of these cells we generated inducible Agc1-CreERT2; ROSA26R mouse models . In Agc1-CreERT2; ROSA26R mice , tamoxifen-induced Cre recombinase activity occurred efficiently in all chondrocytes including hypertrophic chondrocytes during and after development [19] . In the limbs of E14 . 5 Agc1-CreERT2; ROSA26R embryos retrieved from a female treated with tamoxifen at E13 . 5 , the majority of chondrocytes , including pre- and hypertrophic chondrocytes , were positive for LacZ , while essentially no LacZ+ cells were found in the perichondrium , indicating that the Agc1-CreERT2-mediated recombination took place specifically in chondrocytes , not in perichondrium cells ( Figure S2A-a , a′ , b , b′ ) . As described in our previous report [19] , E12 . 5 Agc1-CreERT2; ROSA26R embryos were negative for LacZ when treated with tamoxifen at E11 . 5 ( Table S1 ) , while E13 . 5 Agc1-CreERT2; ROSA26R embryos were positive for LacZ when treated with tamoxifen at E12 . 5 . Experiments detailed in the next paragraph imply that it took between 9 and 18 hours after tamoxifen administration to accumulate enough β–galactosidase in order to obtain a clear readout in Agc1-CreERT2; ROSA26R embryos . Hence it is likely that E12 . 5 was the earliest embryonic stage for Agc1-CreERT2 activity in the femur of these embryos . To determine the length of time that tamoxifen remains capable of activating Agc1-CreERT2 after intraperitoneal injection , E13 . 5 Agc1-CreERT2;ROSA26R embryos were retrieved from pregnant females treated with tamoxifen at either E8 . 5 , E9 . 5 or E11 . E13 . 5 Agc1-CreERT2;ROSA26R embryos treated with tamoxifen at E11 were positive for LacZ staining , while E13 . 5 Agc1-CreERT2; ROSA26R embryos treated with tamoxifen at E8 . 5 or E9 . 5 were negative for LacZ , indicating that tamoxifen has a Cre activation window of three days or less after injection ( Figure S2B and Table S1 ) . To further assess the dynamics of the LacZ labeling after tamoxifen injection , we administered tamoxifen to pregnant females of E15 . 5 Agc1-CreERT2; ROSA26R embryos and followed the appearance of LacZ+ cells in the femur of these embryos at successive times post-injection . At 9 hours post-injection weak LacZ+ chondrocytes including hypertrophic chondrocytes started to appear , at 18 hours many chondrocytes and hypertrophic chondrocytes were strongly positive for LacZ and very few LacZ+ cells were detected in the primary spongiosa . At 24 hours post injection significantly more cells in the primary spongiosa were positive for LacZ ( Figure S2C ) . This suggested that there was a sequential appearance of LacZ staining first in chondrocytes between 9 and 18 hours and then in cells of the primary spongiosa after tamoxifen injection at E15 . 5 the time when ossification is initiated in femurs . We then injected tamoxifen in pregnant females of E11 . 5 Agc1-CreERT2; ROSA26R embryos four days earlier than the onset of primary ossification at E15 . 5 in the femur , a time when the Cre-inducing activity of tamoxifen was no longer retained in this tissue . These embryos were then sacrificed at E16 . 5 . LacZ staining was observed not only in chondrocytes but also in non-chondrocytic cells in the primary spongiosa ( Fig . 3A ) . This experiment largely ruled out that the presence of LacZ+ cells in the primary spongiosa might be due to transient ectopic expression of Agc1-CreERT2 in nascent osteoblasts at E15 . 5 coupled with the tamoxifen induction of Cre recombinase in these cells . This experiment suggested that the labeled cells in the primary spongiosa were derived from chondrocytes , in which the ROSA locus was recombined by tamoxifen induction of Cre recombinase . Also , in the limbs of E15 . 5 Agc1-CreERT2; ROSA26R embryos collected 1 day after tamoxifen injection at E14 . 5 , only chondrocytes ( including hypertrophic chondrocytes ) , and a few non-chondrocytic cells residing immediately below the proximal growth plate , were positive for LacZ , but no LacZ+ cells were present in the center of the marrow cavity in these embryos ( Fig . 3B ) . By comparison , E15 . 5 Col10a1-Cre;Osx flox/+ and E15 . 5 Col10a1-Cre; ROSA-tdTomato embryos ( Figs . 1A and 2A ) showed more abundant labeled cells in the primary spongiosa because recombination and labeling of the hypertrophic chondrocytes occurred earlier . However , when embryos were retrieved at E16 . 5 , 2 days after tamoxifen injection , non-chondrocytic LacZ+ cells were found from right under the growth plates gradually extending to the center of the marrow cavity ( Fig . 3B ) . In the 1-month-old Agc1-CreERT2; ROSA26R mice , born to a female treated with tamoxifen at E14 . 5 , LacZ+ cells were localized on the surfaces of trabeculae and the endosteum and embedded within the bone matrix ( Fig . 3B ) . Together these results substantiated the notion that the tamoxifen-mediated appearance of non-chondrocytic LacZ+ cells in the primary spongiosa was specifically linked to the CreERT2 activity in chondrocytes and could be dissociated from the onset of primary ossification at E15 , 5 . To further validate our observations in Agc1-CreERT2;ROSA26R embryos and mice , we generated Agc1-CreERT2; Osx flox/+ embryos and mice , which were retrieved from tamoxifen-treated pregnant females or were the offspring of these female mice . In the femurs of E14 . 5 Agc1-CreERT2; Osx flox/+ embryos collected 1 day after tamoxifen treatment at E13 . 5 pre- and hypertrophic chondrocytes were positive for EGFP , consistent with the notion that the Osx floxed allele was recombined only in chondrocytes ( Fig . 3C ) . In the femurs of E15 . 5 and E16 . 5 Agc1-CreERT2; Osx flox/+ embryos collected 2 and 3 days after tamoxifen treatment , not only were hypertrophic chondrocytes positive for EGFP but so were non-chondrocytic cells in priFmary ossification centers , similar to the findings in the Col10a1-Cre; Osx flox/+ embryos ( Fig . 1 ) . However when we pulsed tamoxifen in E14 . 5 pregnant mice to tag the EGFP+ ( Osx−/+ ) cells 1 day prior to the onset of primary ossification , almost no non-chondrocytic EGFP+ ( Osx−/+ ) cells were detected in the primary ossification centers at E15 . 5 , except a few at the proximal chondro-osseous interface ( Fig . 3C ) . Whereas at E16 . 5 , in addition to hypertrophic chondrocytes , abundant numbers of non-chondrocytic cells in the primary ossification centers were also positive for EGFP ( Fig . 3C ) . Furthermore , in a 2-week-old Agc1-CreERT2; Osx flox/+ mice , an offspring of a female treated with tamoxifen at E14 . 5 , the EGFP+ ( Osx−/+ ) cells were lining the surfaces of trabeculae and endosteum . They were also found within the bone matrix ( Fig . 3C ) , but were totally absent in calvariae ( S2E Figure ) , similar to what was seen in the 2-week-old Col10a1-Cre; Osx flox/+ mice ( Fig . 1D ) . Nevertheless , unlike the findings in the 2-week-old Col10a1-Cre; Osx flox/+ mice , in which Cre activity persisted in Col10a1-expressing cells , the hypertrophic chondrocytes were no longer positive for EGFP in the Agc1-CreERT2; Osx flox/+ mice . Together these experiments in Agc1-CreERT2; Osx flox/+ embryos indicated that the labeling of chondrocytes preceded the time of appearance of labeled non-chondrocytic cells in the primary spongiosa . The timing of these distinct processes was controlled by the time of tamoxifen injection in the pregnant females . In a separate experiment a pregnant Agc1-CreERT2;ROSA26R female was treated with tamoxifen at E17 . 5 when growth plates in the femurs were fully established ( Figure S2D ) and her offspring was sacrificed two days after birth . In these pups LacZ+ non-chondrocytic cells were present in the primary spongiosa and were distributed in an dome-shaped area under the growth plate , suggesting that mature chondrocytes in the established growth plate continued to be an important source of these LacZ+ cells . Collectively , the distinct temporal and spatial presence of reporter+ non-chondrocytic cells observed after recombination by either Col10a1-Cre or Agc1-CreERT2 strongly suggested that these reporter+ non-chondrocytic cells were derived from mature chondrocytes . If the reporter+ cells , labeled at the time they existed as chondrocytes but later found in the trabecular region and in the endosteum , were functional osteoblasts , these cells should express typical osteoblast markers . To test this hypothesis we first performed double immunofluorescence ( DIF ) experiments with anti-EGFP and anti-Osteocalcin ( Ocn ) on the femur of 1-month-old Col10a1-Cre;Osx flox/+ mice . Non-chondrocytic EGFP+ cells represented cells derived from hypertrophic chondrocytes whereas cells positive for intracellular Osteocalcin ( Ocn+ ) corresponded to mature osteoblasts . Hence cells that are positive for both markers ( EGFP+Ocn+ ) are cells derived from chondrocytes that are functional osteoblasts . Shown in Fig . 4A , the EGFP+Ocn+ cells were indeed present in the femur of 1-month-old Col10a1-Cre;Osx flox/+ mice . High resolution confocal microscopy revealed that these Ocn+EGFP+ cells exhibited a characteristic mature osteoblast morphology ( Fig . 4B ) . It was evident that not all Ocn+ cells on the bone surfaces were positive for EGFP , implying that there were other sources of osteoblasts besides hypertrophic chondrocytes . In addition , not all EGFP+ cells were positive for Ocn , since EGFP+ cells represented cells derived from mature chondrocytes at different stages of osteoblast differentiation . In the trabecular region , approximate 63% of Ocn+ cells were Ocn+EGFP+ cells , and about 62% Ocn+ endosteal cells were Ocn+EGFP+ cells ( Fig . 4C and Table S2 ) . Consistent with this result , DIF with anti-EGFP and anti-Col1a1 antibodies revealed that the EGFP+ ( Osx−/+ ) cells in the primary spongiosa were tightly associated with type I collagen in the femurs of E18 . 5 Col10a1-Cre; Osx flox/+ embryos ( S3A Figure ) . Likewise , IHC with anti- BSP showed that many of the LacZ+ cells in the primary spongiosa were closely associated with BSP in the femur of E16 . 5 Agc1-CreERT2; ROSA26R treated with tamoxifen at E14 . 5 ( Figure S3B ) . We further generated Col10a1-Cre; 2 . 3Col1-GFP; ROSA-tdTomato triple transgenic mice to assess co-localization of Col10a1-Cre induced reporter+ cells and an osteoblast-specific marker in vivo . In these mice , a Tomato+ non-chondrocytic cell represented a Col10a1-Cre marked cell whereas a EGFP+ cell corresponded to a functional osteoblast [24] . Thus a Tomato and GFP double positive ( Tomato+EGFP+ ) cell represented an osteoblast derived from Col10a1- expressing chondrocytes . In the femur of 3-week-old Col10a1-Cre;2 . 3Col1-GFP;ROSA-tdTomato mice there was an abundance of Tomato+EGFP+ cells throughout the trabecular and endosteal surfaces ( Fig . 5A ) . These Tomato+EGFP+ cells in both trabeculae and cortical region were further examined by high resolution microscopy ( Fig . 5B ) . As the Ocn+EGFP+ cells shown in Fig . 4B , the Tomato+EGFP+ cells in the femur of 3-week-old Col10a1-Cre;2 . 3col1-GFP;ROSA-tdTomato mice had a typical osteoblast morphology . In the trabecular area , approximately 60% of total EGFP+ cells were also Tomato+ , and around 68% of EGFP+ cells on the endosteum surface were positive for Tomato ( Fig . 5C and Table S3 ) . This result indicated that the hypertrophic chondrocyte-derived osteoblasts contributed to more than half of the osteoblast population in the 3-week-old mice in vivo . As in the case of Col10a1-Cre;2 . 3Col1-GFP;ROSA-tdTomato mice , Tomato+EGFP+ cells were also present in the femur of newborn Agc1-CreERT2;2 . 3Col1-GFP;ROSA-tdTomato mice treated with tamoxifen at E14 . 5 ( Fig . 6 ) . The Tomato+ and Tomato+EGFP+ cells were specifically located in the primary spongiosa , but not in perichondrium and periosteum , tissues that were also marked by EGFP expression . This result further substantiated our hypothesis that hypertrophic chondrocytes were capable of differentiating into bone–forming osteoblasts in vivo . Together , these results suggested that osteoblasts , which were derived from chondrocytes in cartilage anlagen constitute a major source of osteoblasts responsible for trabecular and endosteal bone formation in the growing endochondral skeleton . To evaluate whether the growth plate chondrocytes during postnatal growth also contribute to the osteoblast pool in the primary spongiosa as is the case during embryonic development , we generated inducible Agc1-CreERT2; ROSA-tdTomato and Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato mice to specifically tag and trace the growth plate chondrocytes after birth . At 2 weeks , the Agc1-CreERT2; ROSA-tdTomato or Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato and control mice were treated with tamoxifen and subsequently sacrificed at different time points after tamoxifen injections . ISH showed that Agc1 was specifically expressed in growth plate chondrocytes , not in cells in primary spongiosa of 2 and 3-week-old Agc1-CreERT2; ROSA-tdTomato and Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato mice ( Figure S4A ) . At post tamoxifen day 1 , a majority of growth plate chondrocytes were weakly positive for Tomato and very few Tomato+ cells were present at the chondro-osseous junction in the primary spongiosa of a Agc1-CreERT2; ROSA-tdTomato mouse , while no Tomato+ cells were detected in the ROSA-tdTomato control ( Fig . 7A ) . At post tamoxifen day 2 in Agc1-CreERT2; ROSA-tdTomato mouse , more non-chondrocytic Tomato+ cells appeared below the growth plates in the primary spongiosa in addition to the Tomato+ chondrocytes . The fluorescence signal of day 2 Tomato+ chondrocytes was more intense than the day 1 Tomato+ chondrocytes ( Fig . 7B ) . Unlike in the 2-week-old Col10a1-Cre; Osx flox/+ mice ( Fig . 1D ) , there were no Tomato+ cells on the endosteum or embedded within the cortex in 2-week-old Agc1-CreERT2; ROSA-tdTomato mice both 1 and 2 days post tamoxifen injection . At 8 days post tamoxifen treatment , the number of non-chondrocytic Tomato+ cells in the primary spongiosa was substantially increased and these cells were distributed in a broader area under the growth plates than in the post tamoxifen day 1 and day 2 mice ( Fig . 8Ab ) . A majority of the non-chondrocytic Tomato+ cells in the day 8 primary spongiosa were small in size and as those in day 1 and day 2 mice were not found within the trabeculae or cortex . Only some of the Tomato+ cells in the vicinity of periostea displayed a larger and elongated morphology and some of them were also positive for EGFP , suggesting the presence of chondrocyte derived osteoblasts in the primary spongiosa of these mice ( Fig . 8Ab′ ) . In the 6-week-old Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato mice , which were treated with tamoxifen at 2weeks , the number of non-chondrocytic Tomato+ cells was substantially increased and the cells were distributed throughout the breadth of the primary spongiosa . These cells were morphologically similar to those in the vicinity of periostea in the post tamoxifen day 8 mice . A considerable number of Tomato+ cells were also embedded within trabeculae and a few were found on the endosteum and within the cortex ( Fig . 8B ) . Moreover , the number of Tomato+EGFP+ cells were very substantially increased compared to the post tamoxifen day 8 mice . Thus there was a considerable time lag between the initial labeling of growth plate chondrocytes , the presence of abundant chondrocyte-derived non-chondrocytic cells in primary spongiosa and particularly the formation of large numbers of chondrocyte-derived functional osteoblasts . Note that the number of Tomato+ cells on the endosteum and within the cortex in these mice was much less than in the 3-week-old Col10a1-Cre; 2 . 3Col1-GFP; ROSA-tdTomato mice ( Fig . 5A ) . When 11-week-old Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato mice were treated with tamoxifen and were sacrificed two weeks later , the Tomato+ cells were present in a narrower area under the growth plate ( Figure S4B ) than in the mice sacrificed 8 days after tamoxifen injection at 2 weeks ( Fig . 8A ) ; in these 13-week-old mice there were also very few Tomato+EGFP+ cells . Overall , these results indicate that postnatal growth plate chondrocytes continued to contribute to the osteoblasts pool during the period of rapid postnatal growth in juvenile mice . Bone fracture healing occurs mostly through processes similar to developmental endochondral ossification , which involves a cartilage intermediate [25] , [26] . To test the hypothesis that mature chondrocytes present in the repair callus may be a source of osteoblasts contributing to bone formation during fracture healing , we generated semi-stabilized fractures of the tibia in 2 to 3-month-old Agc1-CreERT2; ROSA-tdTomato or Agc1-CreERT2;2 . 3Col1-GFP;ROSA-tdTomato and control mice and subsequently injected tamoxifen into these mice at 6 to 7 days post-surgery , a time which is prior to or around the time of chondrocyte differentiation [25] , [26] . In the tibia of a post-surgery day 9 Agc1-CreERT2; ROSA-tdTomato mouse treated with tamoxifen 7 days after surgery , the Tomato+ cells were found in the growth plate and in the repair callus but not in the region between these two areas ( Fig . 9Ab ) . No Tomato+ chondrocytes were detected in the ROSA-tdTomato control ( Fig . 9Aa ) . In these mice 9 days after surgery , chondrocyte differentiation occurred in some but not all of the callus cells . Interestingly , the areas of Tomato+ cells in the repair callus were completely matching the chondrocyte areas stained maroon red by Saf-O ( Fig . 9Ab′ ) , suggesting that these Tomato+ callus cells were in fact chondrocytes in the cartilage callus . In the tibia of a post-surgery day 14 Agc1-CreERT2; ROSA-tdTomato mouse treated with tamoxifen 6 days after surgery , Saf-O staining indicated that the repair callus was partially ossified , showing a mixture of both bone ( green ) and cartilage ( red ) tissues ( Fig . 9Bb′ ) . However , almost all cells in the repair callus , both in the cartilage and in the bone regions were positive for Tomato ( Fig . 9Bb ) , suggesting the presence of non-chondrocytic Tomato+ cells in the repair callus . As in the case of the post-surgery day 9 fractured tibia ( Fig . 9Ab ) , the Tomato+ cells were not found in the region between the growth plate and the repair calluses , implying that the Tomato+ cells in the repair callus were in fact either callus chondrocytes or cells derived from callus chondrocytes , and hence that these Tomato+ cells were not derived from growth plate chondrocytes . In the repair callus of a post-surgery day 14 Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato mouse treated with tamoxifen at 6 days after surgery , many of the Tomato+ cells were also positive for GFP ( Tomato+EGFP+ ) , implying that the mature chondrocytes present in the repair callus have the ability to become Col1a1-expressing bone forming osteoblasts ( Fig . 10Ab , Figure S5a ) . At day 29 post-surgery , ossification was almost complete in the repair callus of an Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato mouse treated with Tamoxifen 6 days post-surgery ( Figure S5b ) . The number of Tomato+EGFP+ cells was very substantially increased compared to the day 14 callus ( Fig . 10Ba ) . Thus the appearance of Tomato+ cells in the early callus , their localization corresponding to the cartilage callus , the absence of Tomato+ cells between the growth plate and the callus , and the presence of abundant Tomato+EGFP+ cells in the ossifying callus strongly suggested that chondrocytes in the repair callus were a source of osteoblasts involved in bone fracture healing .
Our objective was to follow the fate of chondrocytes in endochondral ossification and to test the hypothesis whether chondrocytes may undergo transdifferentiation into functional osteoblasts in vivo . In studying the fate of labeled chondrocytes in both Col10a1-Cre-containing embryos and tamoxifen treated Agc1-CreERT2–containing embryos , we observed that , in addition to labeled chondrocytes and hypertrophic chondrocytes , from E15 . 5 on numerous labeled cells with a different morphology than that of hypertrophic chondrocytes were present in the primary ossification center , where neither Col10a1-Cre nor Agc1-CreERT2 were expressed . Later labeled cells were found within the matrix of bone trabeculae and postnatally in the endosteum and within the cortical bone . The reporter+ cells were lining the endosteum continuously from the distal growth plate down to the femur shaft . The area of the endosteum covered by reporter+ cells was gradually increasing with time , from up to mid-shaft at 2weeks to about three quarts of the length at 1 month after birth . Although we occasionally observed very rare reporter+ cells in the perichondrium or periosteum area , the number of these reporter+ cells were too low to account for the abundant reporter+ cells in the primary spongiosa . In Col10a1-Cre;reporter embryos the appearance of the non-chondrocytic labeled cells in the primary spongiosa occurred synchronously with the onset of primary ossification . However , the use of Agc1-CreERT2;reporter embryos allowed us to control the timing of Cre recombination in chondrocytes and the subsequent appearance of labeled cells in the primary ossification centers by varying the timing of tamoxifen administration . The tamoxifen pulse experiments with Agc1-CreERT2; ROSA26R embryos ( Fig . 3 , Figure S2B , Figure S2C and Table S1 ) clearly established that the labeling of chondrocytes and the appearance of non-chondrocytic cells in the primary spongiosa were two sequential events and that the latter could be dissociated from the start of osteogenesis . These experiments provided compelling evidence that the labeled cells in the primary ossification centers were derived from chondrocytes , and that their presence was not due to a hypothetical Cre activity in emerging osteoblasts . The finding that the labeled cells present in the trabecular and endosteal regions of Col10a1-Cre-containing mice expressed two osteoblast markers namely Osteocalcin and a osteoblast-specific Col1a1 promoter-driven EGFP strongly suggested that these cells were functional osteoblasts . In the femur trabeculae of tamoxifen-treated Agc1-CreERT2-containing mice the presence of chondrocyte-derived osteoblasts was also confirmed by co-expression of the tomato reporter and EGFP driven by the osteoblast-specific Col1a1 promoter . Our results further revealed that about 60 percent of osteoblasts in the femurs of 3- and 4-week-old mice were derived from chondrocytes . These numbers were computed in Col10a1-Cre;reporter mice , although one cannot completely exclude that the 60 percent figure maybe an overestimate , if for the instance there was the unlikely possibility of an ectopic Col10a1-Cre promoter activity . The labeling of chondrocytes after a single tamoxifen injection in pregnant females of AgcCreERT2 embryos was unlikely to be complete . In the newborn Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato mice treated with tamoxifen at E14 . 5 ( Fig . 6 ) , and in the 3-week and 6-week-old mice treated with tamoxifen at 2-week ( Fig . 8 ) , some of the osteoblasts ( EGFP+ cells ) in the primary spongiosa were derived from mature chondrocytes prior to tamoxifen injections . This population of EGFP+ cells would therefore not be labeled by Agc1-CreERT2 and these cells would thus be negative for Tomato . Hence , the number of Tomato+ EGFP+ cells would not reflect the actual number of chondrocyte-derived osteoblasts present in these mice . Our data revealed that both mature chondrocytes in cartilage primordia prior to the establishment of growth plates , and mature chondrocytes in established growth plates , transdifferentiated and contributed to the osteoblast pools . The reporter+ cells labeled by Agc-CreERT2 around E14 . 5 , before the establishment of the growth plate persisted into the growth period after birth , since they were still found on the surfaces of trabeculae and endosteum , and embedded within the bone matrix at 2 weeks and 1 month ( Figs . 3B and 3C ) . After the establishment of growth plates , mature chondrocytes in growth plates maintained their ability to become osteoblasts both during late embryonic development and postnatal growth ( Figure S2D , Figs . 7 and 8 ) . During the rapid growth period at 2 weeks , the Agc1-CreERT2 induced more non-chondrocytic reporter+ cells in the primary spongiosa in less time ( Figs . 7 and 8 ) than during the late growth period at 11 weeks ( Figure S4B ) . These results suggested the hypothesis that the contribution of growth plate chondrocytes to the osteoblast pool could possibly be associated with growth plate activity and that this contribution might even eventually stop sometime after the growth period . After the growth period the Col10a1-Cre induced reporter+ cells were still present in the metaphyseal and cortical regions in 6-month-old mice ( Figure S1F ) . However , more of these cells were embedded within the bone matrix , and less of them were found on the bone surfaces , compared to the 2- and 3-week-old mice ( Fig . 1D and Fig . 5A ) , implying that the number of active osteoblasts derived from chondrocytes was likely reduced after the growth period . At 8-month , there were almost no Col10a1-Cre induced reporter+ cells in the primary spongiosa and very few were on the bone surfaces . In addition , the fluorescence intensities of Tomato+ cells were much weaker than that in the younger mice ( Figure S1G ) . These results implied that the Col10a1-Cre induced cells in the 8-month-old mice were probably derived from the growth plate chondrocytes during the growing period , further suggesting an association between the appearance of chondrocyte-derived osteoblasts and the growth plate activities . In the fractured tibiae , Agc1-CreERT2 induced Tomato+ cells were present specifically in the articular cartilage , growth plate and repair calluses when tamoxifen was injected at post surgery day 6 or day 7 around the time of chondrocyte differentiation in the callus [25] , [26] . The Agc1-CreERT2 induced Tomato+ cells continued to be present in calluses after 14 and 29 days post-surgery , when Agc1-CreERT2 was no longer active , suggesting that these Tomato+ cells were either callus chondrocytes labeled at the time of tamoxifen injection around post fracture day 7 or cells derived from these labeled chondrocytes . Importantly there were no significant numbers of Tomato+ cells observed in the region between growth plates and calluses during the repair process , indicating that the Tomato+ cells present in calluses were unlikely derived from growth plate chondrocytes ( Figs . 9 and 10 ) . The Tomato+EGFP+ cells were found in the partially ossified callus at post surgery day 14 , and the appearance of these Tomato+EGFP+ cells was temporally and spatially associated with callus ossification . These data substantiated that chondrocytes in the repair callus were a clear source of osteoblasts responsible for bone formation during fracture healing . Several laboratories have attempted to address the question whether chondrocytes had the ability to transdifferentiate into osteoblasts . In an earlier ex vivo experiment , in which vascular invasion and endochondral ossification were inhibited by removal of the perichondrium , Col1a1 expressing cells were observed in an area between the cartilage anlagens [12] . Although these cells were not further characterized , they were hypothesized to be either hypertrophic chondrocytes arrested at a terminal stage of differentiation or having undergone transdifferentiation . The drawback of this ex vivo experiment was the absence of perichondrium , which may be essential for complete chondrocytes transdifferentiation . We consider the present study as an extension of this previous experiment . In an in vivo study in Col2a1-CreER; ROSA26R embryos specific cells at the cartilage/perichondrium interface , which expressed the Col2a1-CreER transgene , were shown to give rise to some likely osteoblasts mainly in the endosteal region [3] . While this study concluded that hypertrophic chondrocytes did not make a detectable contribution to the osteoblast/osteocyte pool in the central metaphyseal regions under the growth plate within a 3-day time frame , this may be explained by the short time span after tamoxifen administration . In another in vivo study transgenic mice were used harboring both a Col10a1-mCherry and either a 2 . 3kbCol1a1-Emerald transgene or a 3 . 6kbCol1a1-Topaz osteoblast-specific transgene . MCherry fluorescence was detected in the primary spongiosa but not mCherry mRNA . Although much of the mCherry fluorescence was attributed to decaying hypertrophic chondrocytes some of it was present in live cells which did not express the emerald or topaze transgenes . It was speculated that these cells might be a rare population of hypertrophic chondrocytes showing a phenotype different from or preceding that of differentiated osteoblasts . However , the study concluded that transdifferentiation was unlikely to be involved in this phenotype . [13] . Also here , the time lag between labeling of chondrocytes and their appearance as osteoblast marker-expressing cells , as observed in our work , may explain the seeming discrepancy between the studies . During endochondral bone formation , osteoblast lineage cells first segregate from Sox9-expressing mesenchymal progenitor cells in a process that involves canonical Wnt/β-catenin signaling as well as Runx2 and Osx expression and silencing of Sox9 expression and Sox9 activity to form the perichondrium and periosteum [16] , [27] , [28] , [29] , [30] . These osteoblast precursors differentiate further into osteoblasts to form the initial cortical bone . They also invade the cartilage anlagen together with osteoclasts and blood vessels to differentiate into mature osteoblasts that generate bone trabeculae [3] . In parallel to the segregation of osteoblast lineage cells , chondrogenic cells segregate from the same progenitors and in a stepwise fashion differentiate into chondrocytes , which express high levels of Sox9 to form the initial cartilage anlagen [31] . The last step in this chondrogenic pathway is the formation of hypertrophic chondrocytes , in which Sox9 is no longer expressed . Our current results provide evidence that hypertrophic chondrocytes both before and after the establishment of the growth plate constitute a second major source of osteoblasts that participate in the formation of trabecular and cortical bones ( Fig . 11 ) . This property of these chondrocytes to be a direct source of osteoblasts persists until at least three months of age . The same property of chondrocytes to give rise to osteoblasts is also very active in the bone fracture repair process . Recent advances have reshaped the conventional view of differentiated mammalian cells in terms of their plasticity [32] . As is the case in other instances of transdifferentiation [33] it is possible that hypertrophic chondrocytes might first dedifferentiate , then proliferate before these cells redifferentiate into osteoblasts . Our preliminary unpublished experiments revealed the existence of cells in the bone marrow that were derived from Col10a1-expressing chondrocytes . These cells displayed properties of mesenchymal progenitor cells and were able to differentiate into osteoblasts , chondrocytes and adipocytes in vitro , although we do not know whether these cells had the ability to become osteoblast cells in vivo . This preliminary result suggests the possibility that “hypertrophic chondrocytes to osteoblasts” transdifferentiation may indeed involve a dedifferentiation and then a redifferentiation process . We noted that in juvenile mice the chondrocyte-derived reporter+ cells , which were negative for osteoblast markers , persisted in the primary spongiosa for a considerable amount of time before becoming functional osteoblasts ( Figs . 8A and 8B ) . We further speculate that these two hypothetical steps may be independently regulated by different signaling pathways . Collectively , our mouse models provide evidences that chondrocytes , both in cartilage anlagen and in growth plate , are a direct source of osteoblasts responsible for endochondral bone formation during development and postnatal growth . Likewise , chondrocytes in repair calluses undergo transdifferentiation to become osteoblasts contributing to callus ossification in fracture repair . We speculate that chondrocyte-derived osteoblasts may also be involved in other endochondral ossification processes , such as osteophyte formation in osteoarthritis [34] and heterotopic ossification in fibrodysplasia ossificans progressiva disorder [35] , [36] . While this manuscript was in the final stage of revision , a study [37] was published which reached similar conclusions as those in the present paper .
All mice work were conducted strictly according to the NIH guidelines . The Col10a1-Cre mice were generated by Dr . Klaus von der Mark [18] . The Agc1-CreERT2 and Osx flox/flox mice were previously generated in Dr . de Crombrugghe's laboratory [19] , [23] . The 2 . 3Col1a1-GFP mice were generously provided by Dr . David Rowe [24] . The ROSA26R ( B6 . 129S4-Gt ( ROSA ) 26Sortm1Sor/J , stock number: 003474 ) and ROSA-tdTomato ( B6;129S6-Gt ( ROSA ) 26Sortm9 ( CAG-tdTomato ) Hze/J , stock number:009705 ) mice were purchased from the Jackson Laboratory . The mice were injected intraperitoneally with1 . 5 mg/10 g body weight of tamoxifen solution on the desired embryonic days . Tamoxifen was dissolved in 10% ethanol and 90% corn oil . Tamoxifen ( Sigma , T5648 ) was first mixed with 1/10th volume of ethanol and then emulsified in corn oil ( Sigma , C8267 ) . The embryos or mice were fixed in 0 . 2% glutaraldehyde and 0 . 8% formaldehyde ( pH 7 . 5 for embryos , pH 7 . 8 for mice ) at room temperature for 1 hour and stained in 1 mg/ml X-gal overnight at room temperature . After staining , the samples were post-fixed in 4% paraformaldehyde in PBS ( pH 7 . 4 ) at 4°C overnight and dehydrated for paraffin embedding . The post-fixed bones were decalcified prior to embedding . ISH with digoxin ( DIG ) -labeled riboprobes was carried out on 10-µM frozen sections as previously described [38] . Instead of dehydration for paraffin sections , the frozen sections were washed twice in PBS ( pH 7 . 4 ) for 5 minutes . Bones fixed in 4% paraformaldehyde in PBS ( pH 7 . 4 ) overnight were decalcified and embedded for CryoJane frozen sections as previously described [39] . Sections were treated with hyaluronidase ( Sigma , H4272 , 2 mg/ml in PBS [pH 5 . 0] ) at 37°C for 20 minutes for embryonic samples or 30 minutes for postnatal samples and incubated with anti-EGFP ( Invitrogen A11122 ) at room temperature for 1 hour or with anti-mouse collagen type I ( Millipore AB765P ) at 4°C overnight . For IHC , the sections were then incubated with HRP polymer conjugates ( Invitrogen 87-9263 ) and developed with a Vectastain ABC kit ( Vector PK-4000 ) . For IF , the sections were then incubated with anti-rabbit Alexa Fluor 555 ( Invitrogen A21428 ) . For anti-EGFP and anti-col1a1 double IF , after incubation with anti-EGFP and anti-rabbit Alexa Fluor 555 , the sections were incubated with Alexa Fluor 488-labeled ( Invitrogen A10468 ) anti-mouse collagen type I at 4°C overnight . For anti-EGFP and anti-Ocn double IF , after incubation with anti-EGFP ( abcam ab13970 ) and anti-Ocn ( A93876 ) , the sections were incubated with anti-chicken Alexa 488 ( abcam ) and anti-rabbit Alexa Fluor 555 ( Invitrogen A21428 ) . All IF sections were mounted with Prolong Gold antifade reagent with DAPI ( Invitrogen P36931 ) . The 2 to 3-month-old Agc1-CreERT2; ROSA-tdTomato or Agc1-CreERT2; 2 . 3Col1-GFP; ROSA-tdTomato and control mice were used in the fracture surgery . The procedure was described in [3] . Fluorescence cell images were captured using a A1 Laser scanning confocal microscope made by Nikon Instruments . | During endochondral bone formation , which is responsible for the generation of most bones in mammals and many other species , osteoblasts deposit a bone-specific matrix on the surface of cartilage scaffolds made by chondrocytes and hypertrophic chondrocytes . It has long been thought that the terminally differentiated chondrocytes in this cartilage scaffold undergo cell death . Here we demonstrate that chondrocytes can transdifferentiate into osteoblasts and that these transdifferentiated osteoblasts represent a substantial fraction of the bone forming cells in mice , We also provide evidence that chondrocytes can transdifferentiate into osteoblasts during bone fracture repair , a process similar to endochondral bone formation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"biology",
"and",
"life",
"sciences"
] | 2014 | Chondrocytes Transdifferentiate into Osteoblasts in Endochondral Bone during Development, Postnatal Growth and Fracture Healing in Mice |
Phagocytosis and inflammation within the lungs is crucial for host defense during bacterial pneumonia . Triggering receptor expressed on myeloid cells ( TREM ) -2 was proposed to negatively regulate TLR-mediated responses and enhance phagocytosis by macrophages , but the role of TREM-2 in respiratory tract infections is unknown . Here , we established the presence of TREM-2 on alveolar macrophages ( AM ) and explored the function of TREM-2 in the innate immune response to pneumococcal infection in vivo . Unexpectedly , we found Trem-2−/− AM to display augmented bacterial phagocytosis in vitro and in vivo compared to WT AM . Mechanistically , we detected that in the absence of TREM-2 , pulmonary macrophages selectively produced elevated complement component 1q ( C1q ) levels . We found that these increased C1q levels depended on peroxisome proliferator-activated receptor-δ ( PPAR-δ ) activity and were responsible for the enhanced phagocytosis of bacteria . Upon infection with S . pneumoniae , Trem-2−/− mice exhibited an augmented bacterial clearance from lungs , decreased bacteremia and improved survival compared to their WT counterparts . This work is the first to disclose a role for TREM-2 in clinically relevant respiratory tract infections and demonstrates a previously unknown link between TREM-2 and opsonin production within the lungs .
Phagocytosis is the process by which cells ingest particles and is a major effector mechanism of the innate immune system . Professional phagocytes such as macrophages use a variety of surface receptors including scavenger- , Fc- and opsonin-receptors to internalize microbes . In addition , innate immune receptors such as Toll like receptors ( TLRs ) recognize conserved microbial structures and trigger the production of pro-inflammatory cytokines and chemokines , thereby shaping both the innate and adaptive immune response . TLR activation and phagocytosis are intimately linked . Upon phagocytosis of Gram positive and negative bacteria , TLR2 and 4 initially located at the plasma membrane accumulate into phagosomes , sample their contents and elicit immune responses to products which only become accessible after digestion of the cell wall of bacteria [1] , [2] , [3] . TLR activation also enhances the expression of phagocytic receptors such as scavenger receptor A or MARCO and elevates opsonin levels [4] , [5] , [6] , [7] . The prime aim of such inflammatory phagocytosis is to urgently remove invading pathogens before they multiply , invade other tissues and spread systemically . The rapid elimination of pathogens thus prevents excessive inflammation that can otherwise result in immunopathology and organ failure . Understanding the mechanisms that control phagocytosis and limit inflammation is important as this has implications in the survival from infectious diseases . The triggering receptor expressed on myeloid cells 2 ( TREM-2 ) has been proposed to regulate both processes . TREM-2 belongs to a conserved but functionally distinct gene family of proteins , with the best studied family members including TREM-1 and 2 [8] , [9] . TREM-2 is a receptor with an unknown ligand and is expressed by several cell types including bone marrow derived macrophages ( BMDM ) , microglia and osteoclasts [10] , [11] , [12] . Humans with mutations in TREM-2 develop Nasu-Hakola disease , which is characterized by progressive dementia and bone cysts [12] , [13] . Furthermore , recent evidence shows individuals who possess heterozygous rare variants of TREM-2 are at increased risk of Alzheimer's disease [14] , [15] . TREM-2 signals via the immunoreceptor tyrosine-based activation motif ( ITAM ) of the adaptor protein DNAX activation protein of 12 kDa ( DAP-12 ) to mediate its downstream effects . DAP-12 is rather promiscuous and is used by many other receptors having both activating and inhibitory functions [16] . When TREM-1 is engaged , it potentiates the immune response to bacteria and TLR ligands [17] , [18] , [19] . Conversely , TREM-2 was reported to function as a negative regulator of TLR mediated inflammation [10] , [11] , [20] . Therefore , TREM family members act as fine tuners of TLR mediated innate immune responses . Significantly , TREM-2 also plays an important role in phagocytosis . Over-expression of TREM-2 in Chinese hamster ovary ( CHO ) cells confers binding of both Staphylococcus aureus and Escherichia coli and TREM-2 mediates phagocytosis of E . coli in BMDM [21] . The in vivo relevance of the phagocytic capacity of TREM-2 in relation to E . coli peritonitis has been corroborated by a recent study showing that administration of bone marrow cells that over-express TREM-2 enhances bacterial clearance and improves survival in a cecal ligation and puncture ( CLP ) model [22] . In this report we examine the function of TREM-2 in the context of Streptococcus pneumoniae infection , the most frequent cause of community acquired pneumonia [23] . We chose to examine TREM-2 in this context for several reasons . Firstly , TREM-2 is expressed on human AM as well as mouse bronchial epithelial cells , suggesting a role for TREM-2 within the pulmonary compartment [24] , [25] . Secondly , pulmonary expression of TREM-2 and DAP-12 increases in mice following Mycobacterium bovis infection [26] , but the functional importance of TREM-2 in lung host defense is unknown . Lastly , both the early innate immune response and phagocytosis are critical for the outcome during pneumococcal pneumonia [27] , [28] , [29] , [30] . Thus , we hypothesized that studying TREM-2 in the context of pneumococcal pneumonia would provide an ideal model system for examining the cross-talk between TLR signaling and phagocytosis within the lungs .
To investigate the role of pulmonary TREM-2 in the context of bacterial pneumonia , we first established which cell types expressed TREM-2 within the lungs . We determined TREM transcript levels in primary AM and respiratory epithelial lung cells ( pEC ) as well as epithelial cell lines such as alveolar MLE-12 and bronchial MLE-15 cells , respectively [31] , and included RAW 264 . 7 macrophages as a positive control [32] , [33] . While TREM-1 expression was predominantly restricted to macrophages , we discovered TREM-2 to be strongly expressed in both AM and respiratory epithelial cells ( Fig . 1A ) . We confirmed these results by demonstrating expression of TREM-2 on primary AM using western blot ( Fig . 1B ) . Interestingly , we detected two bands in AM , with the upper band probably corresponding to a glycosylated form of TREM-2 as previously observed [34] . Specificity for the antibody was provided as minimal detection of TREM-2 was observed in TREM-2 deficient AM . We then sought to determine pulmonary TREM-2 expression upon S . pneumoniae infection and found a time dependent upregulation in whole lung transcript levels , with highest expression 48 h post infection ( Fig . 1C ) . This increase in pulmonary TREM-2 expression during infection most likely reflected the influx of TREM-2 expressing cells , since TREM-2 transcript levels on primary AM declined following S . pneumoniae treatment ( Fig . 1D ) . Together , TREM-2 was abundantly expressed within healthy lungs and further induced upon infection with S . pneumoniae . To exploit the functional role of this constitutive pulmonary TREM-2 expression in relation to early pulmonary inflammatory responses following bacterial infection , we next infected WT and Trem-2−/− mice with S . pneumoniae for 6 h . Much to our surprise , since TREM-2 was earlier considered a negative regulator of inflammation [10] , [20] , we did not identify any differences in levels of several inflammatory mediators tested , such as TNF-α , MCP-1 , IL-1β and IL-6 ( Fig . 1E and F ) . In fact Trem-2−/− mice only displayed elevated levels of KC , a chemokine required for neutrophil influx following bacterial respiratory tract infections [35] ( Fig . 1E and F ) . This was accompanied by a modest increase in recruited neutrophils ( Fig . 1G ) . Altogether , these data demonstrate that lung TREM-2 only partially dampened inflammation following S . pneumoniae infection in vivo . Given that TREM-2 is considered a negative regulator of inflammation , our observation of unaltered cytokine release in lungs of Trem-2−/− mice following S . pneumoniae infection was surprising . We therefore first re-examined the regulatory function of TREM-2 and concentrated on macrophage responses to bacteria . We found S . pneumoniae or LPS induced TNF-α and KC synthesis by Trem2−/− BMDM augmented ( Fig . 2A and B ) , supporting previous observations of TREM-2 negatively regulating TLR mediated cytokine synthesis [10] . Getting back to the specific role of TREM-2 within the lungs , we then determined whether Trem-2−/− AM would behave similarly to BMDM in response to S . pneumoniae or its lipotechoic acid ( LTA ) , a known TLR2 ligand [36] . Strikingly , Trem-2−/− AM displayed opposite effects to Trem-2−/− BMDM . We discovered that TREM-2 deficient AM exhibited decreased inflammation in response to either stimuli ( Fig . 2C and D ) . The mechanism whereby ITAM coupled receptors generate inhibitory or activating signals is not well understood . However , the avidity of receptor ligation and thus ligand density has been described as a potential reason for pro- versus anti-inflammatory responses by ITAM associated receptors [16] , [37] . To rule out that the dose of bacteria could alter the effect of TREM-2 on TLR signaling , we stimulated primary AM from WT and Trem-2−/− mice with increasing doses of S . pneumoniae . While TNF- α and KC release was not different between WT and Trem-2−/− AM at an MOI of 25 , synthesis was significantly lower in Trem-2−/− AM stimulated with an MOI of 100 S . pneumoniae ( Fig . 2E and F ) . These data suggest that TREM-2 regulated TLR2 mediated cytokine synthesis in a cell type specific manner , i . e . it diminished inflammation in BMDM , but partially enhanced it in AM . Furthermore , given that BMDM and AM were stimulated with an identical dose of bacteria , these data indicate that the cell type specific effects of TREM-2 cannot be explained by differences in receptor ligation and avidity . Considering that the elimination of pathogens is the crucial step in host defense during pneumonia , we then more closely examined the anti-bacterial properties of macrophages and the role of TREM-2 herein . In agreement with published reports that indicated TREM-2 to be a phagocytic receptor for E . coli [21] we first identified that Trem-2−/− BMDM exhibited a decreased uptake of S . pneumoniae compared to WT cells ( Fig . 3A ) . Although we anticipated that Trem-2−/− AM would behave similarly , this postulate proved to be incorrect as Trem-2−/− AM surprisingly exhibited an enhanced uptake of S . pneumoniae compared to WT AM ( Fig . 3B ) . We could rule out differences in phagocytosis between the cell-types due to the kind of bacterium used , since uptake of E . coli was equally enhanced in Trem-2−/− AM compared to WT AM ( Fig . 3C ) . We further confirmed the enhanced uptake of S . pneumoniae in Trem-2−/− AM using confocal microscopy ( Fig . 3D and Fig . S1 ) and corroborated previously published observations showing that Trem-2−/− BMDM exhibit decreased uptake of E . coli compared to WT BMDM [21] ( Fig . S2 ) . Increased phagocytosis of S . pneumoniae by TREM-2 deficient AM was also found following opsonisation of bacteria with either pneumococcal serotype 3 capsular antibodies or 10% WT serum , indicating that TREM-2 can inhibit both FcR-dependent and independent phagocytosis of S . pneumoniae ( Fig . 3E and F ) . Uptake of FITC labeled BSA was unchanged between the genotypes of AM , while in the same experiment uptake of S . pneumoniae was significantly higher , thus illustrating that TREM-2 affects phagocytosis of bacteria but not endocytosis of BSA ( Fig . 3G ) . Further , to extend these studies to other serotypes of S . pneumoniae , we tested uptake of serotype 19A S . pneumoniae , a common serotype that causes invasive pneumococcal disease in children [38] . By doing so , we could observe enhanced uptake of 19A S . pneumoniae in Trem-2−/− AM compared to their WT counterparts ( Fig . 3H ) . We next set out to investigate if the enhanced phagocytosis of S . pneumoniae by TREM-2 deficient AM would also be observed in vivo . To this end , we inoculated WT and Trem-2−/− mice with FITC-labeled S . pneumoniae and assessed the uptake of S . pneumoniae by phagocytes in vivo using flow cytometry . By doing so , we could verify that Trem-2−/− F4/80+ CD11c+ AM did indeed display augmented bacterial phagocytosis in vivo compared to their WT counterparts ( Fig . 3I and J ) . Considering the importance of infiltrating neutrophils in phagocytosing bacteria during pneumonia , we then examined the potential contribution of neutrophils to bacterial clearance in vivo . Following intranasal infection with FITC labeled S . pneumoniae , we found a tendency for decreased phagocytosis by Ly6G+CD11b+ lung neutrophils from Trem-2−/− mice compared to their WT counterparts , suggesting that the modest increase in neutrophil numbers early during infection ( Fig . 1G ) was not responsible for the improved bacterial clearance of Trem-2−/− mice ( Fig . 3K and L ) . We finally asked if the enhanced phagocytosis by Trem-2−/− AM would result in an improved bacterial clearance during pneumococcal pneumonia in vivo . We infected mice with S . pneumoniae for 24 h and recovered significantly fewer bacteria from lungs and bronchoalveolar lavage fluid ( BALF ) from Trem-2−/− mice ( Fig . 3M ) . These data were surprising and important because they challenged the general view of TREM-2 acting as a phagocytic receptor and negative regulator of inflammation but suggested that TREM-2 can modulate phagocytosis and inflammation in a very cell-type specific manner . With uptake of bacteria being higher in resident AM from TREM-2 deficient mice , this macrophage type obviously displays opposite effects from BMDM . Given the importance of lung macrophages in pneumococcal clearance [29] , [39] , we next focused on determining the molecular targets by which TREM-2 deficiency in AM was able to influence pneumococcal uptake and we conducted a genome wide transcriptome analysis of both genotypes of AM . While we discovered no difference in the expression levels of important phagocytic receptors such as the scavenger receptors Cd36 , Marco , Sra-1 and Lox-1 and the complement receptors Cr1 ( Cd35 ) and Cr3 ( itgb2/itgam; Cd11b/Cd18 ) ( Fig . 4A and Fig . S3 ) , we importantly found that Trem-2−/− AM expressed higher baseline levels of opsonins such as C1qa , C1qb , C1qc and Thbs1 ( encoding thrombospodin; Fig . 4A ) . We confirmed higher basal C1qb and Thbs1 levels in TREM-2 deficient AMs by RT-PCR ( Fig . 4B ) and verified enhanced intracellular C1q protein levels in Trem-2−/− AM by flow cytometry ( Fig . 4C ) . In search for specific factors that would explain differences in TREM-2 mediated responses between AM and BMDM , we found that C1qb levels were exclusively increased in Trem2−/− AM , whereas Thbs1 levels were elevated in both TREM-2 deficient AM and BMDM ( Fig . 4D ) . This led us to hypothesize that altered C1q expression could account for differential phagocytic effects between AM and BMDM . As microarray data suggested that TREM-2 had the ability to influence basal C1q production in AM , we decided to test this idea and to determine C1qb and C1qc transcript levels in RAW264 . 7 macrophages over-expressing TREM-2 . Over-expression of TREM-2 was able to lower basal C1qb and C1qc transcript levels , proving that TREM-2 had the ability to regulate C1q transcription ( Fig . 4E and Fig . S4 ) . The nuclear receptor peroxisome proliferator-activated receptor-δ ( PPAR-δ ) has previously been shown to regulate C1q and the C1q promoter contains binding sites for this transcription factor [40] . To understand the potential role of PPAR-δ in our cell system , we treated Trem-2−/− AM with the PPAR-δ inhibitor GSK0660 [41] and could revert C1qb transcript levels to those of WT AM in a dose dependent manner ( Fig . 4F ) . We observed similar results using another PPAR-δ inhibitor , arguing against inhibitor-specific effects in down regulating C1q expression ( Fig . S5 ) . Specificity of these compounds for PPAR-δ and not other PPARs such as PPAR-γ was evaluated , since activation of PPAR-δ with the PPAR-δ activator GW0742 [40] increased transcript levels of C1qb and C1qc in WT AM in a dose dependent manner and co-incubation with PPAR-δ inhibitors abrogated this increase ( Fig . S5 ) . These data confirmed that PPAR-δ activity determines C1q transcription in AM . There was no striking difference in transcript and total protein levels of PPAR-δ between WT and Trem-2−/− AM , suggesting that the ability of TREM-2 to regulate C1q via PPAR-δ was not due to enhanced PPAR-δ levels ( Fig . S6 ) . To demonstrate that TREM-2 itself downregulates PPAR-δ activity , we made use of TREM-2 over-expressing HEK cells and quantified PPAR-δ activation using a reporter system . Over-expression of TREM-2 was able to lower both basal and ligand-induced PPAR-δ reporter activity ( Fig . 4G ) . To rule out differences between HEK cells and macrophages , and to further confirm the reporter gene experiments , we next stimulated TREM-2 or GFP control over-expressing RAW264 . 7 macrophages with GW0742 or DMSO control and monitored nuclear PPAR-δ levels . These data corroborated the experiments in Fig . 4G and demonstrated that TREM-2 could suppress PPAR-δ activity in RAW264 . 7 macrophages ( Fig . 4H ) . To understand the specific effect of TREM-2 on PPAR-δ activation in AM , we next monitored nuclear and cytoplasmic levels of PPAR-δ in WT and Trem-2−/− AM following GW0742 treatment . Trem-2−/− AM exhibited an early translocation of ligand-induced PPAR-δ from the cytoplasm to the nucleus , compared to their WT counterparts ( Fig . 4I ) . Altogether , these data demonstrate that TREM-2 can suppress basal levels of C1q and suggest that the elevated C1q production observed in Trem-2−/− AM occurs via effects of TREM-2 on the nuclear receptor PPAR-δ , a known activator of C1q transcription [40] . C1q is unique among the complement factors , as it is exclusively produced by macrophages and not hepatocytes [42] , [43] , [44] , [45] , [46] . While C1q is well-known for its role in initiating the classical complement pathway , it also exerts complement-independent functions such as clearance of immune-complexes or apoptotic cells [47] , [48] , [49] . The importance of complement in general , including C1q , in protecting against pneumococcal infections is well established [30] , [50] and has been based on the importance of C3b-mediated opsonization of bacteria . Since we observed that TREM-2 deficient AM produced more C1q ( Fig . 4A , B , C and F ) and that Trem-2−/− mice showed a modestly elevated neutrophil influx early during pneumococcal pneumonia ( Fig . 1G ) , we speculated that classical complement activation could be increased in the pulmonary compartment of Trem-2−/− mice following S . pneumoniae infection and monitored levels of C3a and C5a , anaphylatoxins , which promote neutrophil influx [51] . However , we did not observe any differences in C3a and C5a in the BALF of Trem-2−/− mice both 6 h and 24 h post S . pneumoniae infection , suggesting that classical complement activation was not generally elevated in the pulmonary compartment of Trem-2−/− mice following S . pneumoniae infection ( Fig . S7 ) . We therefore studied the potential direct contribution of C1q to bacterial phagocytosis , and pre-incubated WT AM with C1q or control BSA and quantified the uptake of S . pneumoniae . C1q significantly increased bacterial phagocytosis by AM ( Fig . 5A ) , thus confirming a direct role for C1q in phagocytosis , independent of complement activation . Further , these data are consistent with data , showing that C1q can bind to S . pneumoniae , independent of IgG and increase internalization [52] . As our data indicated that PPAR-δ activation enhanced C1q levels in WT AM ( Fig . S5 ) and that PPAR-δ inhibition could lower C1q levels ( Fig . 4F ) , we next decided to examine the role of PPAR-δ mediated C1q production in the context of S . pneumoniae phagocytosis within WT AMs . AMs , where PPAR-δ had been inhibited exhibited lower S . pneumoniae uptake compared to DMSO control cells . Importantly , the attenuated S . pneumoniae phagocytosis upon PPAR-δ inhibition was increased to levels of DMSO control cells upon addition of C1q ( Fig . 5B ) . These data strongly suggest that in AM the principal target for PPAR-δ is C1q for the regulation of S . pneumoniae phagocytosis . Given that PPAR-δ inhibitors can lower basal C1q production in Trem-2−/− AM ( Fig . 4F and Fig . S5 ) , we next postulated that inhibiting C1q using PPAR-δ inhibitors would revert the enhanced phagocytosis by Trem-2−/− AM to WT levels . Significantly , at the dose of inhibitor where C1qb transcript levels in TREM-2 deficient AM were lowered to that of WT AM ( i . e . 10 µM GSK0660 ( Fig . 4F ) ) , phagocytosis was no longer different between WT and Trem-2−/− AM ( Fig . 5C ) . These data linked elevated C1qb levels , produced in a PPAR-δ dependent manner , with enhanced phagocytosis . To confirm the dependence of enhanced phagocytosis by Trem-2−/− AM on higher C1q levels , we blocked C1q using a blocking antibody and discovered that C1q blockage abolished any differences in phagocytosis between WT and Trem-2−/− AM ( Fig . 5D ) . To then test if exogenous administration of C1q would enhance phagocytosis of WT AMs to Trem-2−/− levels , we adhered WT AM to C1q coated plates and quantified uptake of S . pneumoniae compared to Trem-2−/− AM . Consistent with results in Fig . 5A , C1q enhanced phagocytosis of S . pneumoniae by WT AM in a dose dependent manner to reach Trem-2−/− levels ( Fig . 5E ) . Thus , using two independent approaches of inhibiting C1q in Trem-2−/− AM , as well as exogenously supplementing C1q in WT AM , we demonstrated that the enhanced phagocytosis of S . pneumoniae by Trem-2−/− AM depended on macrophage derived C1q . Bone marrow transplantation experiments indicate that transplantation of WT bone marrow into C1qa−/− mice restores levels of serum C1q to WT , clearly demonstrating that C1q is produced by myeloid cells [46] . To further ratify the importance of macrophage derived C1q versus serum C1q , we next opsonised S . pneumoniae with serum from WT or C1qa−/− mice and examined phagocytosis . Our hypothesis was that the difference in phagocytosis between WT and Trem-2−/− AM would still be visible in the presence of C1q-deficient serum , although overall phagocytosis levels would be reduced in both genotypes . Indeed , this was the case , strongly suggesting that AM are the source of C1q and that elevated C1q levels produced by Trem-2−/− AM mediate the enhanced phagocytosis of S . pneumoniae ( Fig . 5F ) . To finally study the importance of elevated C1q production by Trem-2−/− AM in vivo , we blocked C1q in lungs of mice before infection with S . pneumoniae and quantified bacterial clearance . Indeed , blocking pulmonary C1q could reverse the difference in bacterial counts between WT and Trem-2−/− animals ( Fig . 5G ) . We conclude that the improved bacterial clearance we observed in TREM-2 deficient animals was intimately linked to enhanced C1q production by TREM-2 deficient AM . The enhanced bacterial phagocytosis of S . pneumoniae in Trem-2−/− mice finally led us to monitor the survival of WT and Trem-2−/− mice during pneumonia . Ninety five hours post infection , before the first Trem-2−/− mouse succumbed , 60% of WT mice were dead , demonstrating conclusively that WT mice displayed enhanced mortality compared to their TREM-2 deficient counterparts during pneumococcal pneumonia ( Fig . 6A ) . To determine the reasons for this effect , we evaluated pulmonary bacterial loads and bacteremia shortly before the first mouse succumbed . TREM-2 deficient mice exhibited a thousand fold decrease in S . pneumoniae burden in lungs 48 h post infection ( Fig . 6B ) . These data corroborated our earlier time points of infection ( Fig . 3M ) . However , although , we had observed that Trem-2−/− mice reproducibly exhibit enhanced bacterial clearance compared to WT mice ( Fig . 3M , 5G and 6B ) , there were some differences in the degree of bacterial clearance . Natural variation in animal experiments or differences in experimental setup could explain this . Strikingly , while seven out of nine WT mice exhibited bacteremia , S . pneumoniae could only be detected in the blood of one Trem-2−/− mouse ( Fig . 6C ) . Elevated bacteremia in WT mice resulted in enhanced systemic inflammation as determined by plasma IL-6 measurements , compared to TREM-2 deficient mice ( Fig . 6D ) . We next evaluated lung pathology and inflammation . Lung histology revealed that the levels of interstitial inflammation , pleuritis and edema formation were greatly decreased in mice deficient for TREM-2 compared to WT mice ( Fig . 6E and 6F ) . In agreement with this , we detected diminished pulmonary cytokine levels such as TNF-α , IL-1β , MCP-1 and IL-6 ( Fig . 6G ) . Attenuated pulmonary inflammation is not only associated with decreased bacterial burden within the lungs but can also be attributed to improved clearance of apoptotic neutrophils by AM , which promotes the resolution of inflammation [53] , [54] , [55] , [56] . To examine the possibility that the higher inflammation in WT mice might be associated with more apoptotic cells , we monitored pulmonary neutrophil infiltration and active caspase 3 levels in WT and Trem-2−/− mice using Ly6G and active caspase 3 staining . We could observe that Trem-2−/− mice displayed attenuated neutrophil and active caspase 3 levels , particularly in the interstitial space , 48 h post S . pneumoniae infection ( Fig . 6H , 6I ) . Elevated pulmonary cell apoptosis of WT mice late during pneumococcal pneumonia was confirmed using TUNEL staining ( Fig . 6J ) . Specificity for all stainings was verified as no positive signal was detected in the respective isotype controls ( Fig . S8 ) . We next tested the hypothesis that aside from improved S . pneumoniae phagocytosis ( Fig . 3 and 5 ) , Trem-2−/− AM may exhibit elevated apoptotic cell uptake , also known as efferocytosis . To model this we measured the uptake of CFSE labeled apoptotic thymocytes , a well-established and widely used method for determining efferocytosis . Thymocyte apoptosis was confirmed using both DNA laddering and Annexin V/7-AAD positivity ( Fig . 6K and L ) . While there was no difference in the clearance of CFSE labeled apoptotic bodies at low doses between the genotypes of AM , importantly , at a MOI of 10 , Trem-2−/− AM exhibited significantly elevated efferocytosis compared to WT AM ( Fig . 6M ) . We conclude that TREM-2 deficiency improved bacterial and apoptotic cell clearance , lung pathology and prevented systemic inflammation during pneumococcal pneumonia , all of which ultimately led to improved survival .
In this study we examined the effects of TREM-2 on bacterial phagocytosis and pulmonary inflammation within the context of bacterial pneumonia . We unexpectedly discovered a cell-type specific role for TREM-2 , as TREM-2 suppresses bacterial phagocytosis via repression of C1q in AM . These findings demonstrate a previously unknown link between ITAM associated receptor expression and opsonin production in resident AM and explains the detrimental function of TREM-2 during pneumococcal pneumonia . C1q is a member of the defense collagen family that is important for initiating the classical complement pathway and thereby crucial for host defense against pneumococci [30] , [50] , [57] . C1q consists of 18 polypeptide chains that associate together in a “bouquet of tulips” like configuration , with each C1q chain containing a C-terminal globular region that recognizes PAMPs , and a N-terminal collagenous region that associates with phagocytic receptors on macrophages to enhance bacterial phagocytosis [45] , [51] . Within the pulmonary compartment , C1q is produced locally by AM [45] , [46] , i . e . by those cells that provide the first phagocytic line of defense against S . pneumoniae [29] , [39] . C1q has been shown to act as a molecular bridge between S . pneumoniae and host cells , independently of IgG and serotype , facilitating increased adherence and bacterial uptake [52] . Our data reveal a previously unknown link between C1q and TREM-2 and suggest that TREM-2 suppresses C1q production by AM via a mechanism that involves PPAR-δ associated pathways ( Fig . 4 ) . The consequence of this suppressive effect by TREM-2 in AM is important as it explains the detrimental impact of TREM-2 during pneumococcal pneumonia . Our findings that blocking the locally enhanced production of pulmonary C1q was sufficient to reverse the improved bacterial clearance by Trem-2−/− AM in vivo and in vitro support this argument ( Fig . 5 ) . Beside AM , neutrophils are considered important in phagocytosing bacteria upon lung infections and since we observed a modest increase in neutrophil influx early during pneumococcal pneumonia in Trem-2−/− mice , these cells could contribute to the enhanced bacterial clearance in these mice . Although we cannot exclusively rule out this possibility , we importantly discovered that neutrophils from TREM-2 deficient mice exhibit a tendency towards lower uptake of S . pneumoniae in vivo , while Trem-2−/− AM exhibit increased phagocytosis in vitro and in vivo ( Fig . 3 ) . Underscoring the importance of AM in bacterial clearance during pneumonia over neutrophils , previous studies by our group show that selective changes in KC secretion and neutrophil influx occurring in a TLR2 dependent manner are not sufficient to induce altered bacterial clearance or differences in outcome during pneumococcal pneumonia in vivo [58] . But how does TREM-2 influence C1q production ? PPARs are nuclear receptors that are ligand inducible transcription factors and activate target genes through binding to PPAR-response elements ( PPREs ) as heterodimers with the retinoid X receptor family [59] . PPARs shuttle between the cytoplasm and nucleus in response to ligand activation [60] . Three lines of evidence link PPAR-δ to the enhanced C1q production in AM: 1 ) PPAR-δ activation enhances production of C1q by AM; 2 ) PPAR-δ inhibition lowers levels of C1q in Trem-2−/− AM to WT AM; 3 ) Over-expression of TREM-2 lowers levels of C1q and PPAR-δ activation . Interestingly the early ligand induced activation kinetics and nuclear shuttling of PPAR-δ are elevated in Trem-2−/− AM , while PPAR-δ appears to be primarily localized in the cytoplasm of WT AM ( Fig . 4I ) . These data suggest that the manner by which TREM-2 influences C1q transcription is mediated by effects on PPAR-δ activation , possibly via interference with PPAR-δ ligand binding , as already suggested by our PPRE reporter experiments and activity assays ( Fig . 4G and H ) . Regardless of the exact mechanism whereby TREM-2 inhibits PPAR-δ activity , our data suggests that C1q is the principal target for PPAR-δ in the regulation of S . pneumoniae phagocytosis within AM ( Fig . 5B–C ) . Our observations of enhanced phagocytosis in TREM-2 deficient AM are important as they raise awareness of conceptually labeling “TREM-2 as a phagocytic receptor for bacteria . ” We do not dispute previous studies showing that TREM-2 deficiency in BMDM leads to lower bacterial phagocytosis [21] . In fact , we were able to reproduce these findings but believe that the effects of TREM-2 on phagocytosis are cell-type specific . While AM and BMDM are related cell types , they present substantial differences with the former cell type being a resident macrophage isolated using lavage and the latter isolated from bone marrow and in vitro differentiated with M-CSF . Although differential C1q expression upon TREM-2 deficiency may account for the phagocytic differences between the cell types , it is highly likely that in BMDM , additional factors play a role . The cell-type specific effects of TREM-2 on phagocytosis are reminiscent of its effects on antigen presentation and osteoclastogenesis , which clearly differ between different cell types . TREM-2 stimulation of immature dendritic cells induces expression of MHC class II and co-stimulatory molecules that are required for antigen presentation [61] , but this effect is not observed in microglia [11] . TREM-2 deficiency in RAW 264 . 7 macrophages and human monocytes leads to a reduced capacity to generate osteoclast precursors but bone marrow cells from Trem-2−/− mice exhibit accelerated osteoclastogenesis [12] , [32] , [62] . Furthermore , although TREM-2 has been shown to be a phagocytic receptor for apoptotic neurons [11] , uptake of microspheres is unchanged following knockdown of TREM-2 , indicating TREM-2 is not essential for all types of phagocytosis [63] . Interestingly , our data show that AM expressed TREM-2 not only suppresses bacterial phagocytosis but also efferocytosis ( Fig . 6M ) . These data are in contrast to previous studies suggesting that TREM-2 promotes the uptake of apoptotic neurons by microglia [11] , [63] . C1q is important for the uptake of apoptotic cells [47] , [48] , [49] . Indeed , as C1qa−/− mice age they develop multiple apoptotic bodies and autoimmunity compared to WT controls [47] . Within the lung , improved clearance of apoptotic neutrophils by AM promotes the resolution of inflammation [53] , [54] , [55] , [56] . Our data raise the intriguing possibility , that AM expressed TREM-2 might influence efferocytosis through C1q . This hypothesis is consistent with our observations of TREM-2 specifically modulating bacterial phagocytosis and efferocytosis but not the uptake of BSA by AM . These findings make perfect sense in the context of TREM-2 modulating these processes via C1q , as C1q acts as a bacterial opsonin and is also critical for the uptake of apoptotic cells . The ability of TREM-2 to regulate cellular responses in a cell-type specific manner is not limited to phagocytosis and efferocytosis . In this study we show that TREM-2 also modulates S . pneumoniae and TLR2 induced inflammation in a cell-type specific manner , with TREM-2 deficient AMs displaying less inflammation in response to either stimulus . Importantly , these observations are consistent with very recent studies by Correale et al . , who demonstrated that dendritic cells from Trem-2−/− mice display attenuated inflammation in response to TLR ligands [64] . Further , we find in vivo that , other than KC , no other inflammatory mediator tested was higher in Trem-2−/− mice following early pneumococcal infection compared to their WT counterparts . In this regard it is important to note that recent observations indicate that the anti-inflammatory activity of TREM-2 in vivo may differ depending on the disease context . For example , Trem-2−/− mice exhibit attenuated inflammation following DSS induced colitis and stroke , rather than augmented inflammation as expected [64] , [65] . In summary , our study , together with others present in the literature , provides credence for the claim that TREM-2 impacts cellular responses in a cell-type , stimulus- and disease context-specific manner . Since we found TREM-2 to suppress C1q secretion , it is interesting that C1q has been shown to dampen TLR mediated cytokine synthesis , although the exact mechanism behind this is unknown [66] , [67] . It is tempting to speculate that the suppressive effects of TREM-2 on C1q production in AM not only modulate bacterial phagocytosis but also dampen TLR mediated inflammation . This is the first report demonstrating a function for TREM-2 in the pulmonary compartment . Importantly , we clearly show that the ability of TREM-2 to confer bacterial phagocytosis is cell-type specific and that TREM-2 modulates C1q production by AM . It is tempting to speculate that targeting the TREM-2 pathway could be used as a novel strategy for modulating C1q production and pulmonary innate immune responses , which might be of relevance to other respiratory tract infections and possibly autoimmune diseases .
ELISA kits for mouse TNF , MCP-1 , IL-1β , IL-6 , KC and C5a were from R & D Systems and the mouse C3a ELISA was from Uscn Life Science Inc . All ELISAs were performed according to the manufacturer's instructions . GSK0660 was from Tocris Biochemicals . GW0742 and GSK3787 and recombinant C1q were purchased from Sigma . S . pneumoniae LTA was a kind gift from Sonja von Aulock ( University of Konstanz , Germany ) . C1q antibodies used for FACS and blocking experiments were clones 7H-8 ( in vitro and in vivo ) and JL-1 ( in vitro ) respectively , both purchased from Hycult Biotech as was the MARCO ( clone ED-31 ) antibody . Isotype control antibody IgG2bκ ( 559530 ) used for C1q blocking experiments was from BD Bioscience . CD36 ( clone 63 ) antibody was from Millipore . CD45-V500 ( clone 30-F11 ) was from BD Bioscience . Ly6G-PE ( clone 1A8 ) , CD11c-APC ( clone N418 ) , F4/80 ( clone BM8 ) antibodies were from Biolegend . CD11b-Alexa Fluor 700 ( clone M1/70 ) was from eBioscience . TREM-2 ( BAF1729 ) , PPAR-δ ab8937 ) and β-actin ( clone AC15 ) antibodies for western blotting were from R/D systems , AbCaM and Sigma respectively . Recombinant PPAR-δ was supplied in the PPAR-δ activity assay ( Abcam ab133106 ) . Ly6G ( clone 1A8 ) and active caspase 3 ( Asp175 – clone 5A1E ) antibodies used in IHC were from BD Bioscience and Cell Signaling respectively . As secondary reagents we used PE conjugated rat anti-mouse ( eBioscience ) , FITC conjugated goat anti-rat F ( ab' ) 2 ( Jackson Immunoresearch ) , anti-rabbit HRP ( Cell Signaling ) , streptavidin HRP ( R/D systems ) , anti-mouse HRP ( BioRad ) , biotinylated anti-rat IgG ( Vector Laboratories ) and biotinylated , swine anti-rabbit IgG ( Dako ) . Trem-2−/− mice were generated as previously described [10] and the TREM-2 mutation was backcrossed onto a >98% B6 C57BL/6 background facilitated by genome-wide SSLP typing at 10 cM intervals ( done by the Speed Congenics Facility of the Rheumatic Diseases Core Center ) . Wild type mice were purchased from Charles River and all mice were bred at the Medical University of Vienna Animal facility under pathogen free conditions . Age ( 8–10 week ) and sex matched mice were used in all experiments . All animal experiments were discussed and approved through the Animal Care and Use Committee of the Medical University of Vienna and the Austrian Ministry of Sciences and were carried out in strict accordance with Austrian law ( Tierversuchsgesetz; BMWF-66 . 009/0321-II/10b/2008 ) . S . pneumoniae serotype 3 was obtained from American Type Culture Collection ( ATCC 6303 ) . Serotype 19A S . pneumoniae was a clinical isolate from a patient suffering from severe invasive pneumococcal disease and confirmation of the serotype was provided by antibodies specific to the capsule ( Statens Serum Institute ) . Both strains were grown for 6 h to mid-logarithmic phase at 37°C in Todd-Hewitt broth ( Difco ) , harvested by centrifugation at 4000 rpm for 15 min at 4°C , and washed twice in sterile saline . S . pneumoniae serotype 3 , was used for all in vivo experiments and most in-vitro experiments ( except Fig . 3H ) . Bacteria were diluted in sterile saline to obtain an estimated concentration of 105 CFU per 50 µl for intranasal inoculation of mice . The true concentration was determined by growing serial 10-fold dilutions on sheep blood agar plates overnight . Pneumonia was induced by intranasal administration of a bacterial suspension containing 105 CFU S . pneumoniae ( ATCC 6303 ) as described earlier [19] , [58] , [68] . Six , 24 , or 48 h after infection , mice were anesthetized with ketamine ( Pfizer ) and sacrificed . Blood was collected in EDTA-containing tubes and plated on blood agar plates to determine bacterial counts , plasma was stored at −20°C . Whole lungs were homogenized at 4°C in 4 volumes of sterile saline using a tissue homogenizer ( Biospec Products ) , after which serial 10-fold dilutions in sterile saline were plated on blood agar plates , left at 37°C and CFU were counted 16 h later . Remaining lung homogenates were incubated for 30 min in lysis buffer ( containing 300 mM NaCl , 30 mM Tris , 2 mM MgCl2 , 2 mM CaCl2 , 1% Triton X-100 , and pepstatin A , leupeptin , and aprotinin ( all 20 ng/ml; pH 7 . 4; Sigma-Aldrich ) ) at 4°C , centrifuged at 1500 g at 4°C , and supernatants were stored at −20°C until cytokine measurements were performed . Lungs for histology were harvested 48 h post infection , fixed in 10% formalin , and embedded in paraffin . Four-µm sections were stained with H&E and analyzed by a pathologist blinded for groups , who scored lung inflammation and damage as previously described [19] , [58] . In separate experiments , bronchoalveolar lavage ( BAL ) was performed 6 h and 24 h after infection by exposing the trachea of mice through a midline incision , canulating it with a sterile 18-gauge venflon ( BD Biosciences ) and instilling two 500 µl aliquots of sterile saline . Approximately , 0 . 9 ml was retrieved per mouse . Total cell numbers were counted using a hemocytometer ( Türck chamber ) ; differential cell counts were done on cytospin preparations stained with Giemsa . BALF supernatant was stored at −20°C for cytokine measurements . In some experiments , to assess survival , mice were intranasally infected with S . pneumoniae and survival was monitored regularly for 10 days . In experiments involving C1q blocking , 125 µg of C1q blocking antibody ( kindly provided by Admar Verschoor , Technical University of Munich ) or isotype control were intranasally instilled prior to S . pneumoniae infection . Paraffin embedded lungs from WT and Trem-2−/− mice were deparaffinized in xylene and ethanol and subjected to antigen-retrieval using citrate buffer pH 6 . 0 ( Vector laboratories ) . Thereafter endogenous peroxidase activity was blocked with 1 . 6% H2O2 in PBS . Following washing and blocking steps using 10% swine or rabbit serum ( Vector laboratories ) in PBS for the active caspase 3 ( Cell signaling ) and Ly6G ( BD Biosciences ) stains respectively , sections were incubated either overnight at 4°C or 1 h room temperature with the active caspase 3 diluted to 1∶25 in swine serum overnight and the Ly6G antibody diluted 1 in 50 in rabbit serum . After washing , endogenous biotin and avidin sites were blocked using the avidin biotin blocking kit ( Vector laboratories ) . Sections were washed , and incubated with either biotinylated anti-rat IgG ( Vector Laboratories ) or biotinylated , swine anti-rabbit IgG ( Dako ) for the Ly6G and active caspase 3 stains respectively . Binding was visualized using the Vectastain ABC kit ( Vector laboratories ) followed by either a step where the sections were incubated with DAB conjugated to HRP ( active caspase 3 ) or Nova Red stain ( Vector laboratories , Ly6G stain ) . Sections were countered with hematoxylin , dehydrated and subjected to light microscopy . Tunnel stain was performed using the in situ cell death detection kit , according to the manufacturer's instructions ( Roche 1684809 ) . Briefly , paraffin embedded lungs from WT and Trem-2−/− mice were deparaffinized in xylene and ethanol and subjected to antigen-retrieval using Pronase E ( Sigma ) . Following washing and incubation with Tunnel reaction mix , nuclei were stained using DAPI ( Sigma ) . Slides were visualized under fluorescence illumination ( Zeiss , AxioImager . M2 ) . AM were obtained by BAL from healthy WT or Trem-2−/− mice . Cells were resuspended in RPMI 1640 containing 1% penicillin/streptomycin ( pen/strep ) and 10% FCS and plated for either phagocytosis assays or stimulations at the appropriate density . In some cases this was on C1q ( Sigma-Aldrich ) coated or BSA coated plates and cells were left to adhere for 3 h or overnight before stimulations . BMDMs were retrieved from the tibia and the femur of mice and differentiated in RPMI 1640 supplemented with 1% pen/strep , 10% FCS and 10% L929-conditioned medium for 7 days . Primary lung epithelial cells were isolated as previously described [69] and were cultured in in HITES media as were MLE-12 and 15 cells as previously described [31] . RAW 264 . 7 cells cultured in RPMI 1640 containing 1% pen/strep and 10% FCS . In all experiments related to cytokine production cells were stimulated with 2×107 CFU/ml S . pneumoniae ( MOI 100 ) or 10 µg/ml LTA for 6 h unless otherwise indicated . Retroviral transfection was used to generate RAW264 . 7 cells that were stably transfected with GFP or TREM-2 . Briefly , the packaging cell line GP-293 HEK ( Clontech ) was transfected with TREM-2 expression plasmid ( pORF-mTREM-2 originally purchased from InvivoGen ) or GFP control plasmids and VSV-G ( retroviral vector ) . RAW 264 . 7 cells were infected with the virus containing supernatants from HEK cells and successfully transfected , GFP expressing cells were sorted by flow cytometry . For reporter gene assays , HEK cells seeded at 1 . 5×105cells/ml , were transfected with PPRE luciferase promoter constructs ( provided by Nikolina Papac , Medical University of Vienna ) , expression vectors encoding PPAR-δ , RXR-α ( provided by Ajay Chawla , University of California , San Francisco ) , vector control ( pIRES , Stratagene ) or a combination of PPAR-δ , RXR-α together with TREM-2 and DAP-12 ( subcloned into the pIRES backbone ) using calcium phosphate . 24 h post transfection , cells were stimulated with 1 µM GW0742 and 48 h post transfection luciferase activity was determined using the Dual-Luciferase Reporter Assay System according to the manufacturer's instructions ( Promega ) . All transfectants contained the pRenilla luciferase gene vector ( Promega ) as an internal control for transfection efficiency and luciferase values were normalized to renilla . S . pneumoniae ( ATCC 6303 and serotype 19A ) or E . coli ( 018:K1 ) were grown in Todd-Hewitt broth ( Difco ) or Luria Bertani medium respectively , washed twice and resuspended in saline at a concentration in the range of 109 CFU/ml as determined by OD600 . Bacteria were heat killed at 65°C , for 30 min , washed once with 10 ml 0 . 1M NaHCO3 , resuspended in a solution containing 0 . 2 mg/ml FITC dissolved in 0 . 1M NaHCO3 and incubated under constant stirring in the dark at 37°C for 1 h . Bacteria were washed twice with PBS and the concentration was set to obtain 2×109 CFU/ml . Thymocytes were isolated from 4 week old C57BL/6 female mice and apoptosis was induced by culturing 6×106 freshly isolated thymocytes in RPMI/10% FCS with 1 µM dexamethasone for 16 h as previously described [70] . Apoptotic laddering was assayed using a commercially available kit ( Roche Diagnostics ) . 7-AAD/Annexin V staining was performed by resuspending the thymocytes in staining buffer ( 0 . 1M Hepes , 1 . 4M NaCl , 25 mM CaCl2 , pH 7 . 4 ) , after which Annexin V ( BD Biosciences ) was added at 1∶20 for 15 min , following which 7-AAD ( eBioscience ) was added at 1∶50 for 15 min . Cells were analyzed by flow cytometry ( BD LSR Fortessa ) . Following apoptosis , thymocytes were labeled with CFSE , according to the manufacturer's instructions ( Molecular Probes ) . AM or BMDM from WT and Trem-2−/− mice were plated at 0 . 5×106/ml in 12-well microtiter plates ( Greiner ) and allowed to adhere overnight . After washing steps , FITC-labeled heat-killed S . pneumoniae or E . coli ( O18:K1 ) was added in the presence of RPMI for 1 h ( MOI 100 ) at 37°C or 4°C ( as a negative control ) , respectively . Cells were treated with proteinase K at 50 µg/ml for 15 min at room temperature to remove adherent but not internalized bacteria and subsequently placed on ice and washed . In some experiments , bacteria were pre-opsonised with either 10% pooled WT mouse serum , C1qa−/− serum [47] or 10% Type III capsular antibody ( Statens Serum Institute , Denmark ) in RPMI for 30 min before addition to the cells . Uptake was analyzed using a flow cytometer ( Beckton Dickinson FACScalibur ) . The phagocytosis index of each sample was calculated: ( mean fluorescence x % positive cells at 37°C ) minus ( mean fluorescence x % positive cells at 4°C ) . Verification of phagocytosis results obtained via FACS was conducted using confocal microscopy as previously described [68] . Briefly , AM plated at 3×105 in 8 well chamberslides ( Lab-Tek Chamberslide system ) were incubated with FITC-labeled heat-killed S . pneumoniae at a MOI 100 for 1 h at 37°C . After washing steps , lysosomes were stained with Lysotracker red and nuclei with DAPI ( Invitrogen ) , followed by visualization using confocal laser scanning microscopy ( LSM 510 , Zeiss ) . The ratio of engulfed bacteria ( as determined by overlay of green bacteria and red lysosomes ) was quantified by an independent researcher from 300 counted cells per well and is expressed as percentage of cells that contain bacteria . For the in vivo phagocytosis assays WT and Trem-2−/− mice were inoculated intranasally with 5×106 CFU ( MOI 10 , assuming 5×105 cells in a naïve mouse ) FITC-labeled S . pneumoniae . BALF was collected 4 h later , cells were resuspended in a PBS supplemented with 1% FCS and antibodies against Ly6G , CD11b , F4/80 , CD11c and CD45 and incubated for 30 min . After a washing step , cells were resuspended in PBS and analyzed by flow cytometry ( BD LSR Fortessa ) . AMs were identified as F4/80+ , CD11c+ , Ly6G− , CD11b− cells . Neutrophils were identified as the Ly6G+ , CD11b+ , F4/80− and CD11c− population . To control for background FITC signals and non-specific binding of bacteria , the % of FITC-positive CD45− cells ( i . e . non-phagocytosing cells ) was subtracted from the % of FITC-positive CD45+ cells ( i . e . containing phagocytosing cells ) . For efferocytosis assays AM from WT and Trem-2−/− mice were incubated with CFSE labelled apoptotic thymocytes at the indicated MOI for 1 h at 37°C or 4°C ( as a negative control ) , respectively . AM were subsequently removed and stained using APC conjugated CD11c antibody ( clone N418 , eBioscience ) . Uptake was analyzed using a flow cytometer ( Beckton Dickinson FACScalibur ) . The efferocytosis index of each sample was calculated: ( mean fluorescence x % CD11c+ CFSE+ cells at 37°C ) minus ( mean fluorescence x % CD11c+ CFSE+ at 4°C ) . Trizol was used for RNA extraction from primary cells and cDNA was converted using the Superscript III first strand synthesis system as recommended by the supplier ( Invitrogen ) . RT-PCR was conducted according to the LightCycler FastStart DNA MasterPLUS SYBR Green I system using the Roche Light cycler II sequence detector ( Roche Applied Science ) . Mouse gene-specific primer sequences used are shown in supporting information Table S1 . All transcript levels studied were normalized to HPRT . 5×106 cells were treated as indicated in the figure legends and whole cell extracts were prepared . Cells were washed once with cold PBS , after which the cell pellet was solubilized and lysed in ice cold whole cell extract buffer ( 20 mM Hepes pH 7 . 6 , 400 mM NaCl , 1 mM EDTA , 5 mM NaF , 500 µM Na3VO4 , 25% glycerol , 0 . 1% NP-40 , 1 mM PMSF , 1 mM DTT , 0 . 1 mg/ml aprotonin ) . Lysates were centrifuged at 14 , 000 rpm for 15 minutes and stored at −80°C . Equal amounts of protein were separated by electrophoresis on a 10% SDS polyacrylamide gel and transferred to polyvinylidene difluoride ( PVDF ) membranes . Antibodies specific for TREM-2 and PPAR-δ were used at 1∶1000 and β-actin at 1∶500 . Immunoreactive proteins were detected by enhanced chemiluminescent protocol ( GE Healthcare ) . 5×106 cells were treated as indicated in the figure legends and nuclear extracts were prepared by washing the cells with ice cold PBS , followed by three washes in 1 ml of ice cold hypotonic buffer ( 10 mM HEPES pH 7 . 9 , 1 . 5 mM MgCl2 , 10 mM KCl ) to induce swelling . Thereafter , the pellet was resuspended in hypotonic buffer supplemented with 0 . 1% Nonidet P-40 and incubated on ice for 5 minutes to release nuclei . Subsequently , samples were centrifuged at 4°C to pellet nuclei , the cytoplasmic fraction was removed and the nuclear pellett resuspended in 50 µl of ice cold high salt buffer ( 20 mM HEPES pH 7 . 9 , 1 . 5 mM MgCl2 , 420 mM NaCl , 25% glycerol ) and incubated for 15 minutes at 4°C . Disrupted nuclei were centrifuged at full speed in a table top centrifuge for 15 minutes at 4°C , after which the supernatant ( nuclear fraction ) removed . PPAR-δ DNA binding activity in nuclear extracts was determined using the PPAR-δ transcription factor assay kit according to the manufacturer's instructions ( Abcam ) . In brief , nuclear extracts were incubated on a 96 well plate to which the PPRE consensus sequence was immobilized . After binding and washing , PPAR-δ activity within nuclear extracts was detected using a PPAR-δ specific antibody followed by incubation with a secondary HRP conjugated antibody and quantification using spectrophotometry . The specific OD was calculated by subtracting the non specific binding wells ( i . e . wells where nuclear extract was absent ) , and normalized to protein content . Isolated total RNA was purified using the RNeasy kit per manufacturer's instructions ( Qiagen ) . Total RNA ( 200 ng ) was then used for GeneChip analysis . Preparation of terminal-labeled cDNA , hybridization to genome-wide murine GeneLevel 1 . 0 ST GeneChips ( Affymetrix ) , and scanning of the arrays were carried out according to manufacturer's protocols . Affymetrix microarray cell intensity files were combined , and expression was normalized using the robust multi-array average algorithm [71] , generating an expression matrix . Identification of differentially regulated genes was performed with significance analysis of microarrays as described previously [72]; a false-discovery rate of 5% was imposed . All data are deposited at Gene Expression Omnibus ( http://www . ncbi . nlm . nih . gov/geo/ ) and the accession ID is GSE51378 . Data are presented as the mean ± SEM . Comparisons between groups was assessed using either T-test or ANOVA followed by Tukey's multiple comparisons analysis , where appropriate . Survival data was analyzed by Log rank ( Mantel-Cox ) test using GraphPad Prism Software . | Bacterial respiratory tract infections are a major cause of morbidity and mortality , and Streptococcus pneumoniae ( S . pneumoniae ) remains the main cause of community acquired pneumonia worldwide . The continued rise in antibiotic resistance stresses the need for better insights into the host defense mechanisms associated with pneumococcal pneumonia . The early innate immune response that constitutes bacterial phagocytosis , complement activation and inflammation is critical for the outcome during pneumonia . The triggering receptor expressed on myeloid cells 2 ( TREM-2 ) has recently been shown to be both a negative regulator of the inflammatory response and a promoter of phagocytosis , but its contribution to pneumonia remains unknown . In our study , we unexpectedly found that alveolar macrophage expressed TREM-2 is detrimental in bacterial phagocytosis and clearance during pneumococcal pneumonia . This occurred via the suppressive effects of TREM-2 on complement component 1q ( C1q ) , an important regulator of bacterial phagocytosis that is crucial for the host response during pneumonia . Thus , targeting the TREM-2 pathway could be used as a novel strategy for modulating C1q production and pulmonary innate immune responses , which could be of clinical relevance during pneumonia and other respiratory tract infections . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases",
"medicine",
"and",
"health",
"sciences",
"biology",
"and",
"life",
"sciences",
"immunology",
"microbiology",
"pulmonology"
] | 2014 | The Triggering Receptor Expressed on Myeloid Cells 2 Inhibits Complement Component 1q Effector Mechanisms and Exerts Detrimental Effects during Pneumococcal Pneumonia |
Plant guard cells gate CO2 uptake and transpirational water loss through stomatal pores . As a result of decades of experimental investigation , there is an abundance of information on the involvement of specific proteins and secondary messengers in the regulation of stomatal movements and on the pairwise relationships between guard cell components . We constructed a multi-level dynamic model of guard cell signal transduction during light-induced stomatal opening and of the effect of the plant hormone abscisic acid ( ABA ) on this process . The model integrates into a coherent network the direct and indirect biological evidence regarding the regulation of seventy components implicated in stomatal opening . Analysis of this signal transduction network identified robust cross-talk between blue light and ABA , in which [Ca2+]c plays a key role , and indicated an absence of cross-talk between red light and ABA . The dynamic model captured more than 1031 distinct states for the system and yielded outcomes that were in qualitative agreement with a wide variety of previous experimental results . We obtained novel model predictions by simulating single component knockout phenotypes . We found that under white light or blue light , over 60% , and under red light , over 90% of all simulated knockouts had similar opening responses as wild type , showing that the system is robust against single node loss . The model revealed an open question concerning the effect of ABA on red light-induced stomatal opening . We experimentally showed that ABA is able to inhibit red light-induced stomatal opening , and our model offers possible hypotheses for the underlying mechanism , which point to potential future experiments . Our modelling methodology combines simplicity and flexibility with dynamic richness , making it well suited for a wide class of biological regulatory systems .
Stomata are small pores located in the epidermes of plants that allow carbon dioxide ( CO2 ) uptake for photosynthesis as well as diffusion of O2 , produced by photosynthetic reactions , from the plant to the atmosphere . They are also the sites of water vapour loss through transpiration . Stomata are bordered by pairs of guard cells , the swelling of which leads to stomatal opening ( enlargement of the pore ) , while their shrinking leads to stomatal closure . The size and shape change of the guard cells is due to their uptake or loss of water , which is driven by changes in cellular osmotic potential as a result of the accumulation or depletion of solutes . Guard cells are sensitive to multiple external and internal stimuli , e . g . light , intercellular CO2 concentration ( Ci ) , the stress hormone abscisic acid ( ABA ) , and vapour pressure difference ( VPD ) between the leaf interior and the surrounding atmosphere [1]–[9] . Guard cells have photoreceptors for red and blue light , and guard cell responses to light of these wavelengths are the main focus of our work . As stomatal aperture regulation has a major impact on both the hydration status and the photosynthetic status of the plant , guard cells' sensitivity to stimuli is vital to the survival of vascular terrestrial plants . Plants' successful adaptation to the environment influences all life-forms on Earth . In particular , better understanding of the signalling and regulatory networks involved in stomatal responses is a necessary step toward improving the drought tolerance of crops . Guard cells have long been a popular system for dissecting the functions of individual genes and proteins within signalling cascades . The most studied signals are blue light and ABA [10] . There has been extensive experimentation carried out to elucidate the roles of key signaling proteins , enzymes , and small molecules in these signal transduction pathways , and to identify the relationships between diverse components in the system . Numerous experiments have addressed the roles of light quality [3] , [5] , [6] , Ci [11] , and VPD [5] , [12] . A synergistic action between blue and red light in the formation of malate , a major intracellular osmoticum , was discovered [13] . Phototropins were identified as blue light-specific photoreceptors of guard cells [14]–[16] , mediating blue light-specific stomatal opening . New evidence constantly adds to our knowledge on guard cell functioning , e . g . the recently discovered inhibition by phosphatidic acid ( PA ) of blue light-induced stomatal opening via type 1 protein phosphatase ( PP1 ) [17] and the relationship between the activation of the H+-ATPase and light quality [18] . Much less has been done , however , on a systems level to synthesize all existing evidence into a network model of light-regulated stomatal opening , or to elucidate the crosstalk between different signal transduction cascades , such as those triggered by light and ABA . One such pioneering work was done by Li et al . on modeling the ABA signal transduction network leading to stomatal closure [19] . That work synthesized the published evidence for direct interactions and indirect causal effects between cellular components into a consistent network of ABA-induced closure and formulated a Boolean dynamic model that recapitulated or predicted a large number of knockout phenotypes . Another recent systems level advance is the development of the OnGuard software that incorporates ion transporters at the guard cell plasma and vacuolar membrane , the salient features of osmolyte metabolism , and the major controls of cytosolic Ca2+ concentration and pH [20] . In this software , and models that use it [21] , [22] , the light signal transduction pathways are approximated by a pre-defined , light-dependent increase in the activities of all ion-translocating ATPases at the plasma and vacuolar membrane , and in sucrose and malate synthesis . That work does not consider light of different wavelengths nor the specific mechanisms through which the different types of light signals are perceived and transduced . Given the abundance of experimental results regarding stomatal opening and its regulation , dynamic modelling of the full light-stimulated stomatal opening process and its inhibition by ABA is now tenable , and is the focus of this work . We synthesize more than 85 articles describing experimental observations into a comprehensive network of 70 components , of which 4 are signals ( blue light , red light , CO2 , and ABA ) , and stomatal opening is the sole output . The network incorporates in a parsimonious manner more than 150 interactions or causal relationships between components . We develop a dynamic model based on the network by characterizing each component with discrete activity levels and by describing its regulation with a combination of logic and algebraic functions . The multiple activity levels of the components and the detailed updating functions offer a biologically more accurate representation of the system than Boolean models; for example , the output node , stomatal opening , has more than 20 levels in the model , ranging from 0 to 14 . 2 . The model has a repertoire of more than 1031 distinct states ( see Text S1 ) , which gives it substantial dynamic richness and makes it one of the most complex dynamic models of biological systems ( see also [23]–[26] ) . At the same time the discreteness of the states maintains the computational simplicity of the model . The model recapitulates a comprehensive array of known behaviours and phenotypes . Since the model is made up of node-level information ( i . e . the regulatory function of each component ) , this agreement serves as validation . The model enables an unprecedented understanding of the regulation of stomatal opening and predicts new phenotypes caused by the disruption of components . Moreover , the model reveals aspects of the system , particularly in the interplay between red light and ABA , where critical experimental evidence is lacking .
The first step in building the model is to construct the regulatory network that represents the system . A network is an abstraction of a system in which each element is represented as a node , and each pairwise interaction or regulatory relationship is represented by an edge . Edges in signal transduction networks are generally directed ( meaning that the interaction has a source and a target ) and signed ( positive or negative ) . The majority of the known components involved in stomatal opening are proteins , including receptors , enzymes , channels , protein kinases and phosphatases , thus most of the nodes of the network represent proteins . To be able to incorporate the metabolic processes and ion fluxes also involved in stomatal opening , we also include important inorganic compounds , ions , certain biological processes ( i . e . photophosphorylation , carbon fixation , stomatal opening ) and entities ( e . g . mitochondria ) as nodes . In some cases , the subcellular localization of a molecule or enzyme can change , making a key difference in the modulation of stomatal opening . In these cases we use multiple nodes , one for each location . Positive edges in our network correspond to activation , up-regulation , or biochemical synthesis , and are represented with a terminating arrowhead , while negative edges indicate deactivation , inhibition , or consumption , and are shown as terminating in a solid circle . The translocation of a protein or the transport of solutes through channels or carriers is also represented by an edge . A relationship stimulated by another component of the network is represented by an edge starting from the stimulus node and incident on the stimulated edge . For instance , malate exits the cytosol and enters the apoplast through active anion efflux channels ( AnionCh ) ; this is represented by an edge from AnionCh incident on the edge that starts from cytosolic malate and ends in apoplastic malate . Certain causal regulatory relationships may be mediated by other nodes; a path ( a sequence of nodes and edges ) is a better representation of such indirect relationships between nodes . We used logical inference to incorporate the components suggested by the totality of relevant experiments to mediate such indirect causal relationships; this process has been formalized previously [19] , [27] . We distilled more than 85 articles from the literature into 153 edges among 70 nodes , summarized in Table S1 . Text S2 provides an illustration of the process of network construction based on the literature , in which the pathway that starts from blue light and ends at the H+-ATPase is used as an example . The plant hormone ABA , produced in response to environmental stresses such as drought , opposes the effect of light on guard cells and reduces stomatal apertures [7] , [10] , [28] , [29] . To maintain our focus on stomatal response to light , yet to be able to investigate the cross-talk between different signals , the ABA-response section of the model is a condensed representation of the relevant pathways . We followed two contraction principles to achieve a simpler yet dynamically equivalent representation of the system [30] . Functionally redundant pathways in this section are merged; for instance , the two mechanisms by which NO can elicit calcium release from intracellular stores ( CaR ) ( by cyclic ADP-ribose or by 8-nitro-cyclic guanosine monophosphate [31] , [32] ) are compressed into a single edge from NO to CaR . Further , if the sole known function of an element is to pass on the signal it received , i . e . it has a single incoming activation edge and a single outgoing activation edge , the element is not shown in our model and its upstream regulator is directly connected to its downstream target . Figure 1 represents the resulting network of 70 nodes and 153 edges . The colour coding of the nodes signifies the functional connectivity of each node to the four signals , which is based on the existence of paths between a signal and the respective node but is also informed by the specific combinatorial regulation of the node ( described in detail in the section “Elements of the dynamic model” ) . A brief description of the biology represented by the network is as follows; Text S3 provides a detailed description of the network . Both red and blue light activate guard cell photophosphorylation , providing adenosine triphosphate ( ATP ) , the primary chemical energy transporter within the cell , for metabolic processes [33] . Subsequent carbon fixation provides sugars , primarily sucrose , as osmotica for guard cell swelling and stomatal opening [34] , [35] . This pathway is formed by purple coloured nodes in the left side of the network . A blue light-specific pathway ( blue coloured symbols ) leads to the activation of the plasma membrane H+-ATPase [4] , [36] . H+-ATPase activity hyperpolarizes the plasma membrane [4] , with subsequent uptake of K+ [37] , [38] and accumulation of its counterions , Cl- , NO3- , and malate2- [13] , [33] . These ions also function as osmotica during light-induced stomatal opening [6] , [39] , [40] . The stress hormone ABA initiates a signal transduction network ( yellow nodes ) which ultimately inhibits the plasma membrane H+-ATPase , inhibits malate synthesis , and induces malate breakdown and release [2] , [41]–[45] . Thus the majority of the nodes in the network ( the green-coloured nodes ) are regulated by blue light and ABA . The twenty-three nodes that have more than two levels in our model are highlighted with a red shadow . Representing a system with a network reveals important characteristics and interrelationships that have been hidden previously , and enables researchers to test prevailing theories and to identify new hypotheses [46] . We started by looking at the node degree , defined as the number of edges to which the node is connected , of the 70 nodes . The degree can be broken into the in-degree , i . e . the number of incoming edges ( and therefore , of direct upstream regulators ) , and the out-degree , i . e . the number of outgoing edges ( and therefore , of direct downstream targets ) . The four signal nodes , blue light , red light , CO2 , and ABA , have an in-degree of zero . The node stomatal opening is the only node in the system with an out-degree of zero . Table 1 lists the 10% of nodes with the highest in-degree , out-degree , and total degree , respectively . Most nodes in this list are known key mediators or regulators of light-induced opening . For instance , the node that represents the cytosolic malate2- concentration has the highest in-degree and also the highest total degree in the network . Malate , the major counterion for K+ in guard cells and a common organic metabolite , is indeed involved in multiple metabolic pathways . The node that represents the 14-3-3 protein-bound H+ ATPase , H+-ATPasecomplex , is also among the nodes with highest in-degree and total degree , indicating its multi-tiered regulation and its important role in determining the membrane potential and hence the flow of multiple ions . The ion channels Kin , Kout , and anion efflux channels are also among the highly regulated nodes in the system . The stress hormone ABA has the highest out-degree , due to its targeting of multiple nodes in the pathway of blue light-induced stomatal opening and in the ABA signalling network . Cytosolic Ca2+ concentration ( [Ca2+]c ) is an important secondary messenger for both blue light and ABA signalling , as reflected by its high out-degree and total degree . The node PMV , which denotes the potential across the plasma membrane , also has high out-degree and total degree , reflecting its control of channel activities . Next , we identified the strongly connected components of the system . A strongly connected component is a group of nodes wherein any node is reachable from any other node through a path ( a series of consecutive nodes and edges ) . Intuitively , a strongly connected component is a closely-knit group of nodes with interwoven feedback that usually forms an important functional module of a network . The stomatal opening network contains three strongly connected components ( SCCs ) , comprising 31 nodes ( SCC1 ) , 3 nodes ( SCC2 ) and 2 nodes ( SCC3 ) , respectively ( Figure 2A ) . The 3-node SCC2 represents the interplay amongst Ci and carbon fixation processes in guard cells and mesophyll cells: Ci is required by photosynthesis and photosynthesis lowers Ci in turn . The 2-node SCC3 represents the two directions of transport between apoplastic and cytosolic NO3- . The largest SCC signifies the crosstalk between the different signals of the system , since all four signals of our model connect to it . Eight of the thirteen high-degree nodes listed in Table 1 are in the largest SCC . Most of the remaining high degree nodes have only outgoing or incoming edges and thus cannot be strongly connected . Twenty-seven nodes , including the nodes of SCC2 , can reach the nodes of SCC1 through directed paths . Eleven nodes , including SCC3 , can be reached from SCC1 through directed paths . Only a single node , CHL1 , is not connected to SCC1 by a directed path . There are 19 , 436 simple paths ( i . e . , paths with no repeated nodes ) between the four signal nodes and stomatal opening . The vast majority of these paths pass through SCC1 . Remarkably , five important paths bypass SCC1 ( see Figure 2A ) . The five paths start from blue light , red light and CO2 , respectively , pass through photophosphorylation and SCC2 , and then through sucrose , and finish at stomatal opening . Photophosphorylation and SCC2 together represent the photosynthetic carbon fixation processes in guard cells and mesophyll cells . Two conclusions can be drawn based on the existence of these SCC1-bypassing paths: i ) The five paths represent photosynthetic carbon reduction pathways . ABA does not inhibit photosynthesis in mesophyll cells [47] , and there is no indication that ABA would inhibit guard cell photosynthesis . Thus there is no current wet bench evidence that ABA would be able to affect the accumulation of sucrose via guard cell photosynthesis ( see Figure 2A ) . ii ) Based on current knowledge , sucrose accumulation does not introduce any feedback into the system . We next ranked the 60 edges of SCC1 according to their importance to SCC1 integrity . If a strongly connected component is densely connected , the loss of a single edge should not affect the reachability of node pairs in it . However , if after the loss of an edge certain nodes can no longer reach other nodes , or cannot be reached from other nodes , they are no longer part of the strongly connected component and thus the number of nodes in the strongly connected component decreases . Table S2 provides information on the effects of removal of each edge . Removal of any one of 26 edges led to no change in the composition of SCC1 . Loss of any one of 19 edges led to minimal changes , i . e . the loss of a single node from SCC1 . Among the edge removals that do induce significant breakdown , the four listed in Figure 2B lead to loss of more than 40% of the nodes in SCC1 . All four edges are closely related to [Ca2+]c , indicating the critical role of [Ca2+]c in the formation of SCC1 . A closer examination revealed that SCC1 contains two smaller groups of strongly connected nodes ( Figure 2C ) . Group 1 contains 12 nodes from the ABA signalling pathways . Group 2 contains 17 nodes , including the H+-ATPase , its four regulators , nodes denoting major ions , and PMV , which are major mediators of blue light signalling . Seven edges connect group 1 to group 2 , which is the reason why the nodes in group 2 are coloured green in Figure 1 . However , there is a single path from group 2 back to group 1 , mediated by CaIC and [Ca2+]c . The loss of any of the nodes or edges involved in this path results in a major breakdown of SCC1 ( Figure 2B ) . The fact that both groups are strongly connected with [Ca2+]c indicates that [Ca2+]c bridges the signalling between blue light and ABA . This conclusion is corroborated experimentally [48]–[53] . Indeed , it is known that [Ca2+]c is an important secondary messenger in both blue light [53] , [54] and ABA signalling [55] , [56] . Our strongly connected component analysis offers additional insight into the role of [Ca2+]c and reveals that it is a key participant in a feedback loop formed by these pathways . The signal transduction network described in Figure 1 forms the basis of our dynamic model of light-induced stomatal opening . The dynamic model characterizes each node with a state variable ( which we will also refer to as “level” ) and with a regulatory function ( also called update function ) that indicates the future state of the node as a function of the current state of its regulators . Iterative determination of each node's state from a suitable initial condition yields the dynamic behaviour of the whole system . Importantly , the global dynamics of the whole system is an emergent property that is not predetermined by the modeller but arises from the local dynamics ( the regulation of each component ) . We developed a discrete dynamic model in which the nodes are assigned two or more qualitative levels . We aimed to employ the minimal number of levels that was sufficient to describe the experimentally observed relative outcomes for various conditions ( e . g . combinations of signals and manipulations of node states ) . The two possible levels of binary nodes ( 0 and 1 ) can be interpreted as “OFF” , “low” or “inactive” , versus “ON” , “high” , or “active” . Three levels can be interpreted as “low” , “medium” and “high” . The benefit of having three-level nodes is most evident when three qualitatively different categories of values are observed under three or more different experimental conditions , e . g . stomatal opening under red light alone , blue light alone or under dual beam . For such scenarios , having nodes with only two levels would force the grouping of qualitatively different values , and therefore lead to information loss . Among the observations that necessitated the use of more than two levels are the synergistic ( stronger than additive ) effect between red light and blue light in malate formation [13] and stomatal opening [3] , [5] , [9] , [39] , [57]–[59] , and the complex behaviour of [Ca2+]c as a secondary messenger during blue light-induced stomatal opening [50] , [51] , [60] and ABA-induced stomatal closure [43] , [48] , [49] , [61] . In addition , the osmotic potential difference across the plasma membrane that leads to stomatal movement results from the totality of all solutes , whose effect is biophysically additive . In our model , 47 nodes have two levels , nine nodes have three levels ( including photophosphorylation , carbon fixation , [Ca2+]c , CO2 ) , two nodes have four levels ( ATP , Ci ) , three nodes have five levels ( Protein Kinase , H+-ATPasecomplex , PMV ) and nine nodes have more than five levels ( including protein kinase , H+-ATPasecomplex , [K+]c , [malate2-]c , stomatal opening ) . Stomatal opening , in particular , has more than 20 reachable levels , ranging from 0 to 14 . 2 . The numerical values of these levels are not meaningful in isolation; rather , their relationships are reflective of the experimentally observed relative outcomes . Text S1 provides a listing of all node levels . The four signals of the model were assigned a set of levels that represent a particular experimental condition ( light condition , CO2 concentration and ABA presence or absence ) . The possible levels of blue light , red light and ABA are ON ( 1 ) or OFF ( 0 ) , indicating their presence or absence . CO2 has three levels , 0 ( reduced CO2 ) , 1 ( ambient atmospheric CO2 ) , and 2 ( high CO2 ) . The signal levels can be externally changed , e . g . to simulate a light pulse experiment . The 64 internal nodes were chosen to have an initial state of 1 ( 7 nodes ) or 0 ( 57 nodes ) based on experimental information . Text S1 describes these initial states and their justification . Time is discretized into steps in our model . In one time step , the state of each node is updated according to the update function assigned to it [19] . We followed random order asynchronous update [62] . A random permutation of the nodes ( except the node stomatal opening ) is first established at the beginning of each time step , and then all nodes are updated according to this sequence . Stomatal opening , as the sole output of our model , is always updated last within each time step . This algorithm effectively implements a random sampling of process durations . We have chosen this random sampling due to the scarcity of experimental data on relative reaction speeds of signalling pathways and on the timing of specific intracellular events . The degree of randomness can be reduced as timing information becomes available . A delay of 10 time steps is implemented for the node sucrose ( see the update function for the node sucrose in Text S1 ) . We determined empirically that for our network a total of 18 time steps in each simulation is sufficient for all components to reach a time-invariant state ( steady state or , for a minority of nodes such as [Ca2+]c , sustained oscillation ) . The update function of a target node indicates the future state of the target node as a function of the current states of the nodes that have a directed edge impinging on the target . The update functions were developed with information from the literature , such as the state of the target node when one of its regulators is knocked out , and basic biochemical or physical principles when applicable . We aimed to construct the simplest update functions to minimize the number of unknown parameters in the model . The update functions combine logic clauses ( using the Boolean operators NOT , AND , OR ) with addition , subtraction , and multiplication . This approach enables a more detailed and accurate representation than traditional Boolean models , while maintaining simplicity and using few parameters . Two examples of update functions are given in the Materials and Methods section , and Text S1 provides a full list of the update functions and their justifications . We tested different numbers of replicate simulations , and found that 2 , 000 replicate simulations were sufficient for a high reproducibility of our results ( see Text S4 ) . We also demonstrated that our model is robust against uncertainty in the update rules without losing its sensitivity to new information on critical nodes ( see Text S4 ) . For each experimental condition studied , a total of 2 , 000 simulations were performed , and for each node , the activity level averaged over all simulations is reported . Experimental condition refers to the level of the four signals and/or any other elements of the system that might be silenced to represent knockout ( KO ) experiments or made constitutively active; these factors are then invariant across all 2 , 000 runs . Importantly , since the input to our model is local ( the relationships among pairs of nodes , see Table S1 ) , an agreement between the global dynamic results of simulations from the model and wet bench results is not an inherent property of the model . As shown below , however , the model does in fact successfully reproduce known dynamic features exhibited by stomatal opening under various conditions , providing strong support for the validity of the model . We started by comparing the model's results to experiments under different qualities of light in ambient air . In signature experiments that investigated the roles of red and blue light in stimulating stomatal opening , leaves were illuminated with constant background red light upon which a short blue light pulse was superimposed . The stomatal conductance increased slightly in response to the red light , then displayed dramatic transient increase in response to the blue light pulse [5] , [9] , [57] , [58] . As depicted in Figure 3 , our model successfully reproduces this temporal pattern of stomatal opening . We simulated wild type stomatal responses to sustained light in ambient air , as illustrated in Figure 4A . The specific combination of signals for each curve ( blue light , red light , or dual beam ) is initiated at time step 0 and maintained throughout the simulation . All three time courses of average stomatal opening levels ( over the 2000 simulations ) have similar sigmoidal shapes . We consistently observed sigmoidal timecourses for stomatal opening and other variables and in the following summarize them by three parameters ( Figure 4B ) : the maximal ( steady state ) value of the mean level , the number of time steps at which 50% of simulations reach 50% of the maximal ( steady state ) level ( t50% ) , and the number of time steps at which 95% of all simulations reach 95% of the maximal level ( t95% ) . In the presence of both blue and red light , the average stomatal opening level reaches a maximum of ∼11 . 28 in ∼10 steps , whereas red light alone only generates an opening level of 1 . 00 . Notably , blue light , with an opening level of 4 . 15 , is more effective than red light in inducing opening , which is consistent with experimental observations of stomatal apertures [3] , [5] . A synergistic action of red and blue light on stomatal opening , which has been observed experimentally [5] , [9] , [39] , [57] , is reproduced in Figure 4A: the stomatal opening level under both blue and red light ( dual beam ) is larger than the sum of opening levels under each type of light alone . Malate , a common organic compound found in plants , is one of the major counterions for K+ , causing guard cell swelling and stomatal opening . The action spectrum of malate formation shows a synergistic action between red and blue light , i . e . the malate synthesis level under blue light with a red light background ( dual beam ) is higher than the sum of levels under each type of light ( red or blue ) alone [13] . This provides a valuable criterion to evaluate our model . Simulation results presented in Figure 4B clearly indicate that the maximal malate level under dual beam illumination is higher than the sum of maximal levels accumulated under individual light qualities . The result that malate has no observable accumulation under red light alone is also in accordance with experiments [35] . Also listed in Figure 4B is the maximum activation level of the H+-ATPasecomplex obtained in our model under each light condition with ambient air . The proton pump , H+-ATPase , is responsible for the plasma membrane polarization status and for concurrent ion flows . Our model indicates that the H+-ATPasecomplex is activated to the highest degree under a dual beam , to a significant degree under blue light alone , and is inactive under red light alone . Experimental evidence on the activation of the H+-ATPase under red light alone is mixed ( see Discussion ) . Our model supports the conclusion that in ambient air the H+-ATPase is not significantly activated under red light . Our result that blue light alone can activate the proton pump is consistent with multiple experiments [4] , [36] , [63]–[65] . Our model predicts synergy between red and blue light in the activation of the proton pump , and it suggests that this synergy is one of the mechanisms that underlies the synergy between red and blue light in stomatal opening and malate accumulation . Figure 4B also presents our model's prediction of the relative contribution of the two major types of osmotica , ions ( K+ and its counterions ) and sucrose , to the osmotic potential under different light qualities in ambient air . These relative contributions are normalized such that their sum is 100%; see Text S1 for the detailed definition of the contribution of each osmoticum to osmotic potential in our model . The model indicates that ions are the predominant osmoticum being accumulated in response to dual beam or blue light ( 82 . 5% and 75 . 9% , respectively ) , whereas sucrose is the sole osmoticum responsible for red light-induced stomatal opening . These results agree with experimental findings: ion accumulation was observed to take place predominantly under white light or blue light , and is nearly non-observable under red light under ambient CO2 conditions [6] , [35] , [39]; sucrose accumulation takes place under either blue or red light [35] . We investigated the effect that different levels of CO2 , another input signal to our model , has on stomatal opening induced by different qualities of light . The CO2 content in the ambient atmosphere affects light-induced stomatal opening . Air with lower CO2 concentration or CO2-free air was shown to promote white light-induced stomatal opening [1] , blue light-induced stomatal opening [5] , [58] , and red light-induced stomatal opening [40] . Our model captures the enhancement of stomatal opening levels by low CO2 under all light conditions ( Figure 5A ) . Our simulations also indicate that the pattern of the maximal H+-ATPase activity in response to different light and CO2 conditions parallels that of stomatal opening ( Figure 5B ) . Our model thus predicts that the H+-ATPase activity level is promoted by CO2-free air compared to ambient air under all light conditions , and suggests that the promotion of H+-ATPase activity level may contribute to the enhancement of stomatal opening levels by CO2-free air . In order to further test the validity of the model , we next investigated a number of perturbation scenarios . DCMU , a photosynthetic inhibitor , completely inhibits red light-induced stomatal opening , but only partially inhibits blue light-induced stomatal opening [3] . Our simulation of the DCMU effect ( via maintaining the node photophosphorylation at level 0 ) is consistent with these experimental observations: the stomatal opening level drops from 1 to 0 under red light , indicating a total inhibition , while the same disruption has a partial effect on stomatal opening induced by a dual beam or by blue light ( see Table 2 ) . Fusicoccin is a fungal toxin that stimulates K+ uptake in guard cells and causes stomatal opening in darkness [66] , [67] . Fusicoccin has been widely used as a physiological tool to investigate guard cell signaling [68] . Fusicoccin activates the plasma membrane H+-ATPase via a mechanism that involves inactivation of an autoinhibitory domain [69] , [70] . We simulated the effect of fusicoccin on the H+-ATPase by fixing the state of the H+-ATPase at its maximum activation level , nine . Our simulation indicated that without fusicoccin , stomatal opening and K+ levels remain 0 in darkness ( see Table 2 ) . This is due to the absence of H+-ATPase activation in the dark . When fusicoccin is present , stomata open despite the absence of light , and K+ increases in the darkness as well . Our simulation suggests that the H+-ATPase , when activated by fusicoccin in the dark , leads to the hyperpolarization of the plasma membrane and the subsequent activation of K+ uptake channels . The accumulation of K+ and its counterions in the guard cell is the cause for stomatal opening in the dark in the presence of fusicoccin . Both behaviours are consistent with experimental findings [66] , [67] . K+ and its counterions , and sugars , mostly sucrose , are the two types of primary osmotica contributing to stomatal opening . As presented earlier , light quality is one of the determining factors of osmotic composition ( Figure 4B ) . In addition , varying the environmental CO2 concentration was shown to have an effect on osmotic composition as well: CO2-free air or air with low CO2 concentration was observed to induce stomatal opening accompanied by K+ uptake in response to red light [40] . We systematically investigated the effect of different combinations of light qualities and CO2 concentrations on the contribution of each type of osmoticum during stomatal opening ( Table 3 ) . Our results indicated that in ambient air , ion accumulation is the predominant mechanism leading to stomatal opening under dual beam or blue light , while sucrose is the major osmoticum during red light-induced stomatal opening . In the case of CO2-free air or air with reduced CO2 concentration , our model corroborates Olsen et al . ( 2002 ) on the importance of K+ uptake during red light-induced stomatal opening [40] , and predicts the absence of sucrose and the predominance of ion accumulation as an osmoticum under all light qualities . Our model predicts that ion accumulation is severely suppressed in air with elevated CO2 concentration , and that under high CO2 concentration , sucrose is the primary osmoticum responsible for stomatal opening under all light conditions . We further probed the importance of different types of osmotica by computationally imposing an inhibition of sucrose accumulation ( sucrose = 0 ) or by virtually knocking out the H+-ATPase ( H+-ATPasecomplex = 0 ) . Our model indicated that in ambient air ( Table 3 , top ) , the H+-ATPase plays a more important role in dual beam- or blue light-induced stomatal opening than in red light-induced stomatal opening . Conversely , we found that sucrose is more important for red light-induced stomatal opening than for stomatal opening under dual beam or blue light , since its knockout completely inhibits red light-induced stomatal opening while the inhibition is partial for dual beam and blue light . Under CO2-free air ( Table 3 , middle ) , H+-ATPase activity is more critical than sucrose accumulation for stomatal opening under all light conditions . In fact , keeping sucrose at value 0 computationally has no effect on stomatal opening in CO2-free air . In air with elevated CO2 ( Table 3 , bottom ) , the proton pump and henceforth ion accumulation are suppressed , making sucrose the predominant osmoticum for stomatal opening under all light conditions . These results also confirm that the activity of the proton pump is the primary driving force for ion accumulation during stomatal opening [2] , [71] , [72] . Consistent with the difference in the types of osmotica mediating blue light or red light-induced stomatal opening [6] , [35] , a Kin knockout displayed a more severe reduction of opening level in white light and blue light than in red light [73] . This phenomenon is also captured by our model ( Table 4 ) : the simulated white or blue light-induced stomatal opening level decreases dramatically when Kin is forced to be 0 , but this disruption does not affect the red light-induced stomatal opening level . We performed a systematic compilation and comparison of available experimental observations with results generated by our model in a simulation of the experimental conditions . These conditions included different light and/or CO2 and/or ABA stimuli and the manipulation of node states by genetic modifications or pharmacological interventions . Sixty-six comparisons were made in total ( see Table S4 ) , out of which 64 instances exhibited qualitative consistence between experimental observations and simulation results—a successful validation rate of 97% . Our model's consistency with known experimental evidence enables confident prediction of new phenotypes . It takes a significant amount of time and effort for experimentalists to investigate the effect of the genetic knockout of even a single element in vivo . In contrast , a compilation of the phenotypes of all the single-node knockout phenotypes can be readily obtained in silico , and can then be used to inform and prioritize experiments . In vivo , a null phenotype is realized by creating knockout mutants , or by introducing a pharmacological suppressor of a certain element . In silico , this is achieved by keeping the level of the ‘knocked-out’ node at 0 . We systematically investigated the effect of the knockout of a single node from the system in the following three light conditions: dual beam ( blue light = red light = 1 ) , blue light alone ( blue light = 1 , red light = 0 ) , and red light alone ( blue light = 0 , red light = 1 ) , and three atmospheric conditions: normal air with moderate CO2 concentration ( CO2 = 1 ) , CO2-free air ( CO2 = 0 ) , and high CO2 air ( CO2 = 2 ) . ABA was set as absent ( ABA = 0 ) in these simulated knockouts , in which each of the 64 internal nodes was individually eliminated in silico . Table 5 lists the distribution of stomatal opening levels of the knockout phenotypes as a percentage of the wild type opening , which was equated to 100% . In all three ambient air cases , the majority of the knockouts ( 67 . 2% for dual beam , 68 . 8% for blue light , and 95 . 3% for red light ) maintained an opening level within 5% deviation from wild type opening , demonstrating the robustness of the system against single node loss . The knockout of single nodes has a larger impact on stomatal opening under CO2-free air , predominantly due to the inactivity of photosynthetic carbon fixation pathways in this condition , making H+-ATPase activation and the accumulation of ions crucial to stomatal opening . A smaller fraction of cases maintained an opening within 5% of wild type opening ( 68 . 7% for dual beam , 68 . 7% for blue light , and 70 . 3% for red light ) , and 17 . 2% of all phenotypes resulted in an inhibition of 95% or more of wild type stomatal opening level in all light conditions under CO2-free air . Under the high CO2 condition , since the proton pump H+-ATPase activity is greatly suppressed , stomatal opening under all three light conditions is solely dependent on photosynthesis and sucrose accumulation . Therefore , stomatal opening of knockout phenotypes under the three light conditions have an identical pattern: 95 . 3% stay close to wild type opening level , and 4 . 7% display total inhibition of opening . Interestingly , knocking out the small G protein ROP2 or RIC7 induced a stomatal opening level higher than wild type opening under three different light and CO2 conditions ( see Table 5 , 100%–105% ) . This model result recapitulates the experimental observation that ROP2 and recruited RIC7 inhibit stomatal opening in wild-type plants , thus providing a protection mechanism against excessive opening [74] . Table S3 provides a full list of stomatal opening levels for each simulated node knockout in each of the nine conditions . It is known that under simultaneous presence of white light and ABA , the latter functions through several secondary messengers , e . g . [Ca2+]c [75] , [76] and pHc [77] , to inhibit light-induced stomatal opening . Our model reproduced this effect as shown in Table 6 . ABA decreased stomatal opening under combined blue and red light from 11 . 28 to 2 , and stomatal opening under blue light decreased from 4 . 15 to 1 . Unexpectedly , however , our model predicted that ABA had no inhibitory effect on red light-induced opening , which remained at level 1 regardless of ABA . In the course of the construction of our network of guard cell secondary messengers of light and ABA signaling , we found 30 components ( nodes ) wherein regulation of the node by both blue light and ABA had been reported or could be inferred , consistent with experimental evidence that ABA inhibits blue light-stimulated stomatal opening [17] , and our model clearly indicated ABA-inhibition of blue light stimulated stomatal opening ( Table 6 , blue light ) . By contrast , we found no nodes for which regulation of the node by both red light and ABA had been reported . Accordingly , in our model , ABA is predicted to have no effect on red light-induced stomatal opening ( Table 6 , red light ) . This prediction led us to extensively peruse the literature for experiments in which ABA inhibition of red light-induced stomatal opening in isolated epidermes had been explicitly assessed , but no such reports were found . This absence of studies perhaps reflects the general ( but untested ) belief that ABA is able to inhibit light-induced stomatal opening regardless of the wavelength of light . Our identification of this unaddressed question exemplifies how codification of extant knowledge into network models can suggest key new experiments . Accordingly , we experimentally assessed the effect of ABA on red light-induced stomatal opening in Vicia faba epidermal peels ( see Materials and Methods ) . Significant inhibition by ABA of red light-induced stomatal opening was found ( Figure 6A ) . We also observed inhibition of stomatal opening by the photosynthetic inhibitor , DCMU , consistent with a previous report [3] . Qualitatively , the average stomatal aperture values from the wet bench experiments can be divided into two groups: those that have a high value , of which red light treatment is the only instance , and those that have a low value , which contains all the other experimental conditions . Figure 6B shows the stomatal opening levels obtained from our model for the same conditions as those studied experimentally . As our model , constructed based on available knowledge , lacks a mechanism through which ABA can inhibit red light-induced opening , combined action of red light and ABA results in a high opening level equal to that of red light alone , while all other cases have low opening levels . The model predictions are qualitatively consistent with the experimental findings with the exception of the case of combined red light and ABA input . The discrepancy between the model and experimental results in the latter case points out a missing piece of the current knowledge base of the model: a mechanism through which ABA can inhibit red light-induced stomatal opening . The question ‘how does ABA inhibit red light-induced stomatal opening ? ’ remains open . Since sucrose is the major osmoticum accumulated under red light in ambient CO2 , a natural first hypothesis is that ABA inhibits sucrose accumulation . This hypothesis is supported by our result that there are only three nodes whose individual knockout abolishes red light-induced stomatal opening: light reaction , carbon fixation and sucrose ( see Table S3 ) . These nodes form a linear path from ABA to sucrose ( see Figure 2 ) . ABA could disrupt the reaction cascade through which sucrose is generated , or cause the conversion of sucrose into starch , or promote sucrose catabolism within the guard cell or its efflux from the cell . To explore the explanatory power of a putative ABA inhibition of sucrose accumulation , we modified our model by adding an inhibitory edge from ABA to sucrose and adding the Boolean clause “And Not ABA” to the existing rule for sucrose . The simulation result of this modified model is shown in Figure 6B . Notably , after this modification , ABA is able to inhibit red light-induced stomatal opening . The qualitative response pattern across all treatments matches the experimental results . Further , we assessed the impact of the putative inhibitory edge from ABA to sucrose on the overall performance of our model by comparing all the simulation results obtained from the modified model with those obtained from the original model . We were able to confirm that all results stay either identical ( e . g . , for conditions where the ABA signal is absent ) or qualitatively consistent ( for conditions where the ABA signal is present , see Table S5 ) . Importantly , this exemplifies how discrete models , even in the absence of complete knowledge of all interactions and of temporal signaling dynamics , can be readily employed to test new hypotheses and putative pairwise relationships between components of the system .
Our model offers a comprehensive and systematic description of the process of light signal transduction in guard cells and its crosstalk with CO2 and ABA . The network representation we employ reveals the regulatory connections between seemingly remote components . For example , a recent publication , which was not included during our construction of the model , studied the indirect relationship between the SLAC1 anion channel and the regulation of K+ uptake [22] . It reported that anion accumulation in the slac1 anion channel knockout mutant induced the hyperpolarization of the plasma membrane which in turn promoted Ca2+ influx . Ca2+ influx led to an increase in the free cytosolic Ca2+ concentration , which then downregulated the inward K+ channels . This relationship is supported by a path in our network , namely AnionCh→PMV–•CaIC→[Ca2+]c–•Kin . Structural analysis of the network provided significant biological insight . The node degree offers a measure of node importance ( Table 1 ) and path analysis reveals the robustness of the crosstalk between ABA and blue light in regulating stomatal opening . Our strongly connected component analysis identified key components that mediate the cross-talk between blue light and ABA , such as [Ca2+]c ( Figure 2C ) , and highlighted the absence of cross-talk between red light and ABA ( Figure 2A ) . The latter observation , recapitulated by our dynamic model , revealed the absence of experimental investigations of regulatory effects of ABA on red light-induced stomatal opening . Experiments performed in this study fill this knowledge gap and reveal that ABA does in fact inhibit red light-induced stomatal opening . We formulate the novel hypothesis that ABA inhibits sucrose accumulation , and demonstrate that integration of this hypothesis into the model restores the agreement between model and experiments . Our model reveals additional questions where further experimental investigation would be especially fruitful . We discuss a few such examples below . The model can offer a prediction for outcomes based on integration of our current knowledge , but it is up to future experiments to answer these questions definitively . There has been a long debate on whether the plasma membrane H+-ATPase is active under red light alone . The evidence regarding the status of the H+-ATPase during red light-induced stomatal opening is mixed: activation of the proton pump by red light has been observed [78] or inferred [40] , but the results in [78] were not reproduced [64] , [79] and the experiment in [40] was done under reduced CO2 concentration . Little or no activation of the proton pump by red light has been observed in other experiments [18] , [80] . Our model predicts the inactivity of the H+-ATPase under red light in ambient air and predicts a moderate H+-ATPase activity under red light in a reduced CO2 condition ( see Figure 5B ) . Experiments dedicated to measuring the H+-ATPase activity under red light with varying CO2 concentrations will greatly improve our understanding of this matter . A remaining question about sucrose as an osmoticum is the relative contribution of different sources of guard cell sucrose accumulation under different light and CO2 conditions . Sugar accumulation , predominantly sucrose [35] , [81] , can in theory result from photosynthetic carbon fixation , degradation of stored starch [6] , [81] , [82] , or import from the apoplast [34] , [83] , [84] . These three processes exhibit different responsiveness to light and ABA . Photosynthesis is activated by either blue or red light , requires CO2 , and was not observed to be inhibited by ABA [47] ( see Figure 2 ) . Starch content was shown to be constant under red light but decreasing in time under blue light [6] . The H+/sucrose symporter function requires apoplast acidification by the H+-ATPase [34] , [83] , which is not effective under red light in ambient air . Since ABA inhibits the plasma membrane H+-ATPase [51] , [63] , [85] and apoplast acidification by the H+-ATPase is required by the H+/sucrose symporter , it can be inferred that the symporter activity can be inhibited by ABA . In our model we assumed that carbon fixation is the primary source of sucrose accumulation ( see Text S1 for a detailed justification ) . Experiments that dissect the contribution of each source of guard cell sucrose accumulation will not only help improve the model , but also provide insight into the interaction between blue and red light and between light and ABA in the regulation of stomatal movement . Our simulations showed that the inactivation of photophosphorylation , e . g . by DCMU , induces a significant reduction in stomatal opening under all three light quality conditions ( Table 2A ) . Our model predicts that the inactivation of photophosphorylation i ) reduces carbon fixation and hence the amount of sucrose accumulated via photosynthesis , and ii ) reduces the amount of ATP available for H+-ATPase activity . Since the H+-ATPase is not activated by red light in ambient air in our model , our simulations suggest that DCMU inhibits red light-induced stomatal opening through mechanism i ) only . Since the H+-ATPase is activated by blue light in ambient air , our simulations suggest that DCMU inhibits blue light-induced stomatal opening through both mechanism i ) and ii ) . Experiments showed that DCMU partially inhibited blue light-induced stomatal opening and it completely inhibited red light-induced stomatal opening [3] , [80]; it would be informative to investigate the effect of DCMU on dual beam- or white light-induced stomatal opening as well . Further , it would also be interesting to explore the effect of DCMU and the respiratory inhibitor potassium cyanide ( KCN ) on light-induced stomatal opening in CO2-free air , a condition under which photosynthetic carbon fixation is absent , as CO2 , the substrate for carbon fixation , is unavailable . Our model implements a brake-like effect of high Ci on H+-ATPase activity based on the observation that an enhanced level of CO2 depolarizes the plasma membrane [86] and the consequent hypothesis that elevated CO2 inhibits the proton pump at the plasma membrane . This inhibitory effect of Ci on the H+-ATPase activity helps to explain the activation of the H+-ATPase under low CO2 conditions [40] . Understanding the mechanism underlying this effect of Ci on the H+-ATPase and in particular , whether it is a direct or an indirect effect , would provide valuable information . Such data would not only clarify the relationship between light ( especially red light ) and the H+-ATPase activity level , but also offer insight into the synergy between blue and red light . According to our model , red light as a background to blue light can not only provide additional ATP through photophosphorylation for the H+-ATPase activity , but also lower Ci via stimulation of mesophyll photosynthesis and thus raise the activity of the H+-ATPase ( Figure 5B ) . These two mechanisms could be critical in explaining the synergy between blue and red light in the intact leaf . Our model also predicts that ion uptake/accumulation , which hinges upon H+-ATPase activity , is the primary mechanism for stomatal opening in response to red light under CO2-free air ( Table 3 ) . Therefore , stomatal opening of a Kin knockout phenotype in response to red light with CO2-free air should be severely impaired , in contrast to the minimal effect of Kin knockout on red light-induced stomatal opening under normal air ( Table 4 and [73] ) . Experimental verification of this prediction would also support the model's predictions regarding the osmotic composition during stomatal opening in CO2-free air ( Table 3 ) , and provide further evidence for the activation of the H+-ATPase by red light in CO2-free air as proposed in [40] . The current model offers a qualitatively accurate and quantitatively close depiction of short term stomatal movement in response to a light signal . There have been investigations , however , which demonstrate that under natural conditions ( white light ) sucrose accumulates in guard cells in the afternoon and replaces K+ as the dominant osmoticum to maintain stomatal apertures [6] . An interesting potential future direction for our model is the incorporation of emerging knowledge concerning the cross-talk of guard cell circadian rhythms , light and ABA responses ( e . g . [87] ) . The recent successful Boolean model of circadian clocks [88] makes the construction of such an integrated gene regulatory and signal transduction network model feasible . Our modelling framework characterizes each component with two or more levels and expresses the relationships between components as a mixture of logical rules and algebraic operations . Thus our model offers a parsimonious , computationally efficient yet quantitative description of the system's dynamics , making it a step forward from traditional Boolean models and an enhanced modelling tool for systems biology . Our choices of the update functions of the nodes are a simplified and abstracted representation of the best available knowledge . Assuming discrete node levels is an approximation of reality ( e . g . the concentration of a substance or the potential across the plasma membrane is continuous in reality ) ; there is , however , ample evidence of nonlinear regulation wherein not the concentration but rather its relationship with certain thresholds matters . Network-based discrete dynamic modelling has been successfully applied in a great variety of biological systems ( reviewed in [89]–[92] ) . These models enabled the understanding of the systems and generated insightful predictions that were subsequently validated experimentally; recent examples include [24] , [93]–[97] . In the absence of detailed knowledge on relative reaction speeds we deliberately sampled different timescales by implementing random order asynchronous update . While this is not a fully accurate representation of reality , the averaged results of a large number of replicate simulations are representative of behaviors that are not sensitive to small changes in kinetic rates . Future observations of relative temporal patterns of multiple components or measurements of time delays among components can be incorporated by imposing restrictions on the update sequence ( e . g . updating one group of nodes before another group [91] ) . Having established the biological validity of our model , follow-up work in several directions is now possible , linking to recent advances in discrete and continuous dynamic modelling . Translating the model into a polynomial discrete dynamic system [98] or logical discrete model [99] would allow the use of software tools such as ADAM [100] or GINsim [99] , and may yield further insights into the dynamic repertoire of the system . Our model could also be translated into a Boolean model of an expanded network , where multi-level components are represented by multiple nodes in such a way that the group of binary nodes representing the same component allows the recapitulation of the same number of relative outcomes as the original multi-level node ( see e . g . [97] ) . This transformation would allow the application of Boolean network analyses such as elementary signalling mode analysis and attractor analysis [91] . Through network simplification methods [30] , [101] , a core network with fewer nodes and edges could be distilled , which may be amenable for continuous modelling wherein differential equations replace the update functions . For instance , a model that integrates a simplified version of our model with OnGuard [20] would include the various signal transduction pathways , their cross-talk , and the quantitative description of ion flows in guard cells . Our model is generically adaptable , allowing one to incorporate emerging new pieces of information with ease . This modeling methodology can be readily applied to other systems where interaction and relative state information is available . The number of multi-level nodes can be minimized by identifying the node ( s ) for which more than three relative outcomes have been observed ( indicative of a need for more than two levels ) , tracing upstream in the network , and inferring a minimal set of nodes whose multi-level nature could cause all the other nodes' multiple levels . The states of this set of nodes can then be defined as fundamental states ( see Text S1 ) , and the states of other nodes subsequently derived through their updating rules . As demonstrated here , network-based dynamic models of biological systems can serve as a virtual control to test the coherence between experimental results generated in separate experiments , to generate predictions that inform and help prioritize future experiments , and to reveal new questions that deserve attention .
Plants of Vicia faba L . , cv . ‘Long pod’ were grown from seeds on Metro Mix 300 soil media ( Griffin Greenhouse Supplies , Inc . , Morgantown , PA ) in a reach-in growth chamber with 120 µmol m−2 s−1 light , with 8 h day/16 h night cycles and 22°C/20°C day/night temperatures . The plants were watered with deionized water once a week and fertilized once a week ( on a different day ) with half-strength Hoagland's solution . Approximately 20 healthy and young fully-expanded leaflets from 3 week old plants were selected for stomatal aperture measurements . Epidermal peels from the abaxial side were stripped from interveinal regions using forceps and blended in a buffer ( 10 mM MES , 5 mM CaCl2 , 0 . 1% polyvinylpyrrolidone PVP40 , adjusted to pH 6 using 1 M KOH ) for 5 seconds . Blended peels of uniform size were distributed among the wells of each of two multi-well Petri dishes , each well containing 0 . 5 mM CaCl2 , pH 6 . 15 , and placed in the dark for 1 hour to close the stomata initially . Then , 0 . 1% ethanol ( solvent control ) , 50 µM ABA in ethanol , or 20 µM DCMU in ethanol ( final concentrations ) was added to individual wells . One Petri dish containing peels with these treatments was transferred to red light ( 125 µmol m−2 s−1 ) for 3 hours . The other Petri dish was wrapped in foil and kept in darkness for 3 hours as a dark control . The red light source was an InFocus LP500 projector ( InFocus Corp . , Portland , OR ) combined with a red filter transmitting light above 600 nm ( Ridout Plastics Co . Inc . ) . Seventy to two hundred stomatal apertures were measured per replicate . Table S6 contains all the experimentally measured stomatal apertures; the values in Figure 6A are means ± standard errors from three independent replicates . All images were taken at 400x total magnification using a Nikon DIPHOT 300 microscope ( Nikon , Japan ) connected to a Nikon E990 camera ( Nikon , Japan ) . All stomatal apertures were measured using the free access software ImageJ ( http://rsb . info . nih . gov/ij/ ) , version 1 . 39u . Treatments were blinded during image acquisition and analysis . When there is only one upstream regulator of the target , the “Equal” rule is used for positive regulation and the “Not” rule is used for negative regulation . The “AND” operator is used when multiple regulators are required to activate the target . If each regulator is able to independently activate the target , they are connected with the “OR” operator . For inhibition , the “AND NOT” operator is used , thereby requiring a low level or inactivity of the inhibitor in order for the target node to activate . One example of a Boolean update function is shown below: phot1complex* = phot1 And 14-3-3 proteinphot1 Phototropin 1 binds reversibly to a 14-3-3 protein ( 14-3-3 proteinphot1 ) upon the auto-phosphorylation of phototropin 1 in guard cells [102] , [103] . The 14-3-3 protein-phototropin complex ( phot1complex ) is thought to confer an active state to phototropin 1 , which then transmits the light signal to downstream elements . The And function connects phot1 and the 14-3-3 protein , indicating that the formation of the complex requires both of them . The update functions combine Boolean clauses with addition , subtraction , and multiplication . One example of a complex update function is given below: PMV* = PMV – H+-ATPasecomplex + ( AnionCh And ( PMV = –2 ) ) + ( ( [Ca2+]c = 2 ) Or KEV ) PMV is the difference of electric potential across the plasma membrane , i . e . the “membrane potential” . Computationally , we use the value 0 to represent the resting potential of the plasma membrane . Negative values denote the further hyperpolarization of the plasma membrane , and positive values denote depolarization . We assume that five levels ( −2 , −1 , 0 , 1 , 2 ) are sufficient for a qualitative description , and require that the PMV value stays bounded , i . e . the value will not further decrease ( or increase ) when it reaches −2 ( or 2 ) . The future PMV value ( PMV* ) can be shifted away from or stay the same as its current value ( PMV ) , depending on the hyperpolarizing and depolarizing forces . Factors that cause hyperpolarization will decrease the PMV value ( e . g . from 0 to −1 ) , and factors that cause depolarization will increase the PMV value ( e . g . from 0 to 1 ) . An active 14-3-3 protein-bound H+-ATPase causes the extrusion of H+ from guard cell cytosol to the apoplast , hyperpolarizing the plasma membrane . Anion efflux at the plasma membrane causes the plasma membrane to depolarize . A steady anion efflux requires AnionCh to be active and it also requires plasma membrane depolarization . To avoid an oscillation in anion efflux and PMV induced by the discrete nature of the model , we require that PMV be -2 ( most hyperpolarized ) for effective anion efflux . A high [Ca2+]c concentration ( value 2 ) or K+ release into the cytosol from vacuole ( KEV ) are modelled as independent factors causing the plasma membrane to depolarize . Text S5 indicates the pseudo-code for the implementation of two representative update functions .
The network in Figure 1 was drawn with the software yED ( http://www . yworks . com/en/products_yed_about . html ) . The network analyses ( strongly connected component identification , calculation of the number of simple paths between two nodes ) were implemented by custom MATLAB code . Similar analyses can also be done with one of the following tools: NetworkX , a Python graph software library [104] , Cytoscape , a network integration , visualization and analysis tool [105] , or MATLAB's graph theory toolbox , grTheory . The dynamic model was implemented by custom MATLAB code . The pseudo-code of the simulations is indicated in Text S5 . | Stomata are microscopic pores surrounded and regulated by pairs of guard cells located on the surface of plant leaves . Stomata participate in CO2 uptake , O2 release and water vapor loss . Blue and red light induce stomatal opening ( enlargement of the pores ) , which allows the uptake of CO2 , providing the raw material for photosynthesis , and the release of O2 into the atmosphere . The stress hormone abscisic acid induces minimization of pore width , decreasing water loss through transpiration . During drought conditions these counteracting stimuli jointly determine the overall stomatal movement through an integrated guard cell signalling cascade . We synthesized this interaction network between blue light , red light , and abscisic acid by aggregating and interpreting the abundant biological evidence that has been accumulated to date . We used the resulting network as the basis of a multi-level dynamic model of stomatal opening regulation in response to multiple stimuli . The model is validated by comparing its results to a large number of published experimental observations . Our model , and our experiments inspired by it , reveal an unexplored facet of the interplay between light and abscisic acid in guard cell signalling . The model directs future experiments , and its methodology can readily be applied to other systems . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods",
"Computational",
"implementation",
"and",
"tools"
] | [
"plant",
"anatomy",
"computer",
"and",
"information",
"sciences",
"plant",
"science",
"plant",
"physiology",
"simulation",
"and",
"modeling",
"network",
"analysis",
"biology",
"and",
"life",
"sciences",
"signaling",
"networks",
"leaves",
"stomata",
"mathematical",
"mo... | 2014 | Multi-level Modeling of Light-Induced Stomatal Opening Offers New Insights into Its Regulation by Drought |
Dengue is a mosquito-borne viral disease endemic in many countries in the tropics and sub-tropics . The disease affects mainly children , but in recent years it is becoming more of an adult disease . Malaysia experienced a large dengue outbreak in 2006 to 2007 , involving mostly adults , with a high number of deaths . We undertook a retrospective study to examine dengue death cases in our hospital from June 2006 to October 2007 with a view to determine if there have been changes in the presentation of severe to fatal dengue . Nine of ten fatal cases involved adult females with a median age of 32 years . All had secondary dengue infection . The mean duration of illness prior to hospitalization was 4 . 7 days and deaths occurred at an average of 2 . 4 days post-admission . Gastrointestinal pain , vomiting , diarrhea , intravascular leakages and bleeding occurred in the majority of cases . DSS complicated with severe bleeding , multi-organ failure and coagulopathy were the primary causes of deaths . Seven patients presented with thrombocytopenia and hypoalbuminemia , five of which had hemoconcentration and increased ALT and AST indicative of liver damage . Co-morbidities particularly diabetes mellitus was common in our cohort . Prominent unusual presentations included acute renal failure , acute respiratory distress syndrome , myocarditis with pericarditis , and hemorrhages over the brain and heart . In our cohort , dengue fatalities are seen primarily in adult females with secondary dengue infection . The majority of the patients presented with common clinical and laboratory warning signs of severe dengue . Underlying co-morbidities may contribute to the rapid clinical deterioration in severe dengue . The uncommon presentations of dengue are likely a reflection of the changing demographics where adults are now more likely to contract dengue in dengue endemic regions .
Dengue virus ( DENV ) infection is a global health threat affecting at least 3 . 6 billion people living in more than 125 countries in the tropics and subtropics [1] . It is among the most important arthropod-borne diseases . All four dengue virus serotypes ( DENV-1 , DENV-2 , DENV-3 and DENV-4 ) can cause dengue . The disease can present as a mild self-limiting illness , dengue fever ( DF ) , or as the more severe forms of the disease , dengue hemorrhagic fever ( DHF ) and dengue shock syndrome ( DSS ) [2] . The World Health Organization ( WHO ) 2009 guidelines classify patients into three groups; dengue without warning signs , dengue with warning signs and severe dengue [3] . Clinical manifestation of severe dengue includes severe bleeding , severe organ involvement and severe plasma leakage . Most dengue deaths are associated with DHF/DSS ( WHO 1997 guidelines ) and severe dengue ( WHO 2009 guidelines ) . DF and DHF were first documented in Malaysia in 1902 and 1962 , respectively [4] , [5] . A major dengue epidemic was recorded in 1973 , and since then dengue has become endemic in Malaysia with major outbreaks occurring every 3–4 years [6] , [7] . There were a number of reports describing the clinical features and risk factors associated with the severe manifestations of dengue and dengue-related deaths during the first two decades following the 1973 epidemic . During this period , children were the most predominant group affected , hence contributed substantially to the clinical description of severe dengue [8] , [9] , [10] . In the last two decades , however , the number of dengue cases had escalated exponentially . There were 48 , 846 cases and 98 deaths in 2007 in Malaysia with those aged 15–35 years old contributing to at least 48% of the total number of dengue cases [11] . The trend of higher percentage of adults contracting dengue has also been reported in other dengue endemic countries [12] . This review of fatal cases of dengue infection was undertaken in light of this changing epidemiology of dengue in Malaysia and in this region .
The study was approved by the University Malaya Medical Center ( UMMC ) Medical Ethics Committee ( ethics committee/IRB reference number: 611 . 10 ) . Informed consents were not obtained from the patients as this was a retrospective study . The medical records of ten patients at UMMC with dengue-related deaths during the period from June 2006–October 2007 were reviewed and notes were transcribed into standardized data entry forms . Disease severity was classified following the WHO 1997 guideline [2] . This was done as the clinical notes were all in accordance to the WHO 1997 guidelines . The acute-phase serum samples were obtained from the UMMC Diagnostic Virology Laboratory Repository . Convalescent serum sample was available for only one of the patients . Serum samples were respectively tested for dengue-specific IgM and IgG antibodies using SD Dengue IgM and IgG Capture ELISA kits ( Standard Diagnostics , Korea ) [13] . Serum samples were also tested for dengue-specific NS1 antigen using both pan-E Early Dengue ELISA kit ( Panbio , Australia ) and Platelia Dengue NS1 Ag assay ( Bio-Rad Laboratories , USA ) . Virus isolation was performed by inoculating the serum samples onto monolayer of Aedes albopictus C6/36 cells in 24-well plate . The cells were maintained in EMEM supplemented with 2% fetal bovine serum at 28°C for one week . RNA was extracted from cell culture supernatant using QIAamp Viral RNA Mini Kit ( Qiagen , Germany ) . Genotyping was done using the in-house-developed multiplex RT-PCR genotyping kit which amplified a portion of viral NS3 gene . Amplification was performed as previously described [14] , [15] .
There was a dramatic increase in the number of dengue cases seen at the UMMC in Malaysia from 200 cases in the year 2000 to 1 , 826 and 2 , 096 dengue cases in 2006 and 2007 , respectively . There were ten fatal dengue cases between the periods of June 2006 to October 2007 . The demographics of all fatal cases are illustrated in Table 1 . Nine cases were female . The age range was between 11 to 59 years with a median age of 32 years . Adults over 18 years old comprised 80% of all the fatal cases . Of the 10 fatal cases , four were Malay , three Indian and two Chinese . One patient was a foreigner from Bangladesh . Three of the patients were known diabetics . Two patients ( Patient 1 & Patient 2 ) were brought in dead to the hospital ( dead on arrival , DOA ) ; whereas the remaining patients succumbed to the infection within 1–5 days of admission ( an average of 2 . 4 days ) . Four out of the eight admitted patients had rapid deterioration of their clinical features and died within 24 hours of admission . The mean duration of illness prior to hospitalization was 4 . 7 days . There was no clinical history related to dengue fever present for Patient 1 . She had collapsed and became unresponsive while playing at home . Patient 2 was seen at the outpatient clinic four days before she died with a diagnosis of probable dengue fever . Clinical presentations of nine patients ( Patients 2–10 ) are summarized in Table 2 . Persistent vomiting ( n = 9 ) , body ache ( n = 8 ) , bleeding ( n = 7 ) , plasma leakage ( n = 7 ) , abdominal pain ( n = 6 ) , diarrhea ( n = 6 ) and dehydration ( n = 6 ) were the commonest clinical presentations . All nine patients had a history of fever with or without chills and rigors prior to admission . However , only five had a recorded temperature of >37 . 5°C on admission , with only three recording a temperature of >38°C . One patient was hypothermic ( 35 . 5°C ) on admission . Plasma leakage ( presence of pleural effusion , ascites or pericardial effusion ) was noted in seven patients , including in Patient 6 where the presence of effusion was seen on postmortem ( Table 1 ) . Hematological tests were not available for Patient 1 and 2 , and limited results were available for Patient 6 . Severe thrombocytopenia ( <50×109/L ) and hypoalbuminemia ( ≤16 g/L ) were seen in seven patients . An increased alanine aminotransferase ( ALT ) and aspartate aminotransferase ( AST ) of more than 1000 IU/L were found in five patients , and hemoconcentration in five cases ( Table 3 ) . Coagulation profiles of seven patients showed prolonged activated partial thromboplastin time ( APTT ) , ranging from 45 . 1 to >200 seconds . All , except Patient 1 and Patient 2 , experienced rapid clinical deterioration following admission and were transferred to the intensive care units ( ICU ) . The diagnosis of dengue was not apparent in Patient 1 even after a postmortem . A postmortem was not performed for Patient 2 . All eight remaining fatal cases had a clear diagnosis of DSS ( Table 1 ) . Five patients had severe bleeding , either from gastrointestinal tract or from per vaginal bleeding . The postmortem showed hemorrhage in the brain and heart as well . One patient had myocarditis and pericarditis , which led to cardiogenic shock refractory to fluid resuscitation . One patient developed acute pulmonary oedema during the reabsorption phase and subsequently developed septiceamic shock secondary to hospital acquired pneumonia ( Table 1 ) . Her blood and bronchoalveolar lavage culture was negative . The patients who presented with DSS were resuscitated with intravenous fluid . Blood products used included whole blood , platelet concentrate , fresh frozen plasma , and cryoprecipitate . Six patients received blood product transfusion . Post-mortem examination was performed in three patients ( Patient 1 , 4 and 6 ) ( Table 1 ) . The postmortem results in Patient 4 and Patient 6 were supportive of DSS . The cause of death for Patient 1 remained uncertain despite postmortem examination . The postmortem suggested pulmonary hypertension secondary to chronic obstructive airway disease and obesity as the possible cause of death . Dengue infection was confirmed in all patients by dengue serological tests . Seven were both IgM and IgG positive while three were IgM negative and IgG positive . Eight of the patients ( two were unavailable ) tested positive for dengue NSI antigen with DENV-1 isolated from one case . The presence of anti-dengue IgG antibody concurrent with positive detection of dengue NS1 antigen confirmed secondary infections in eight patients . Of the eight patients , anti-dengue IgG titres in five patients were higher than IgM titres . Dengue NS1 antigen test was not performed in two cases ( Patient 2 and 7 ) . However , the presence of IgG in their serum within a period of less than two weeks since the onset of illness suggests secondary infection .
The present study reviewed the clinical features of ten fatal cases of DHF/DSS seen at UMMC , a major teaching and referral hospital , during the period when Malaysia experienced dramatic increase in dengue cases . The dengue deaths were seen primarily in adult females and were associated with secondary dengue infection . Majority of the patients presented with common clinical and laboratory warning signs of severe dengue before death . Underlying co-morbidities may be the contributing factors towards the rapid deterioration in severe dengue . Other complications included involvement of other organs including the brain , heart , liver and kidneys . This is reflective of the shift in the demographics of dengue cases in Malaysia where more adults are affected . In Southeast Asia , severe and fatal dengue has been primarily described among children . Similar pattern was also observed in Malaysia until early 1982 [16] , [17] , [18] , [19] where the percentage of dengue cases became most common among adults of 13 to 35 years old . In UMMC , the median age of laboratory-confirmed dengue cases between the year 2006 to 2007 was 25 years ( age range 1 month to 88 years ) . The majority of cases were adults of 21 to 25 years and >35 years old , with mean percentages of 20 . 5% and 23% , respectively ( unpublished data ) . This trend is similar in most dengue endemic countries in Southeast Asia [20] , [21] . With this changing demography , it is possible that there are features of severe dengue leading to death that could be different from those seen in children . In the present review , fatal cases comprised mainly of adults of >18 years old . There is a higher preponderance of fatal DHF/DSS amongst females . This is despite >55% of dengue cases seen at UMMC between the year 2006–2007 occurred in males ( unpublished data ) . This observation is similar to that reported earlier where there was higher tendency of females to develop DHF/DSS [9] , [19] with higher mortality rate in females [22] even though males consistently comprised of the larger proportion of both DF and DHF , especially in the ≥15 years age group [22] , [23] . More deaths among girls , especially those among the pediatric group , was also reported in Vietnam in 1996–2009 , despite the predominance of boys in dengue cases [24] , [25] . Currently , there is no satisfactory explanation to this phenomena but there are suggestions that this may be due to the more robust immune response in females , resulting in females to be more prone to develop greater inflammatory response or higher susceptibility to capillary permeability [26] , [27] . There was no evidence of differences in the susceptibility and severity to dengue between the different ethnic groups in Malaysia as the percentages of deaths paralleled the ethnic composition of patients who visited UMMC ( unpublished data ) . This is similar to earlier findings done among the fatal cases in Singapore [28] , [29] . On average the patients in this review were admitted on day five of illness and most of them had defervesce . This was followed by rapid deterioration of clinical condition . Four patients died within 24 hours of admission and the remaining four died within 5 days of admission . Our observation is consistent with an earlier study done on seven dengue deaths in Singapore where the reported mean period of illness prior to hospitalization was 4 . 8 days . The mean duration of hospitalization before deaths , however , was longer at 13 . 7 days [29] . A study in Cuba with 12 fatal cases also reported worsening clinical condition and death occurring at an average of 3 . 75 and 7 . 5 days post-hospitalization , respectively [30] . However , the study reported hospitalization of patients at an average of 2 . 9 days post-onset of illness . This rapid deterioration in the clinical condition is consistent with those presenting with advance stage of disease . Late hospitalization may also be a possible contributing factor to increased risk of mortality and rapid deterioration in severe dengue [24] . Mortality in dengue may also be due to the presence of acquired co-morbidities such as obesity , alcoholism , smoking and the presence of other chronic illnesses such as diabetes mellitus ( DM ) [31] , [32] . Worsening of co-morbidities , rather than directly from dengue infection [29] , [32] could be the reason for death seen especially in adults . The presence of DM is especially prominent among our dengue death cases . This observation is consistent with several earlier reports [29] , [31] , [32] implicating DM , the most common chronic disease in Malaysia especially among females , as a possible contributing factor to death in severe dengue [33] . Asthma , hypertension , chronic obstructive pulmonary disease and chronic renal insufficiency , were other important reported co-morbidities contributing to dengue fatalities [29] , [31] , [32] , [34] , [35] but these were not explicitly seen in our study . Common clinical features of dengue seen in our study include general body ache , abdominal pain , plasma leakage , diarrhea , vomiting , dehydration and bleeding manifestations . Some of the symptoms are consistent with warning signs for severe dengue [3] . Similar findings , especially gastrointestinal symptoms have been reported in other studies [29] , [30] , [32] emphasizing the importance of warning signs as a tool to recognize patients at risk of severe dengue . Hepatomegaly , a common and important clinical feature associated with DHF/DSS or severe dengue [36] , [37] , was seen in four patients; two from physical examination and another two at postmortem . Hepatomegaly may have been an under-diagnosed clinical feature in our study possibly due to the insufficient documentation . The frequency of hepatomegaly may be higher as it was seen in two out of the three postmortem conducted . Hepatosplenomegaly is a clinical feature associated with macrophage activation syndrome ( MAS ) seen in many autoimmune diseases [38] , [39] . It may also be associated with DENV infection [40] . However , splenomegaly was not identified as a clinical feature in our series of patients and was only seen in two patients from their autopsy studies . At least eight of the fatal cases in our study had evidence of secondary dengue infection , which has been associated with severe outcome of dengue via antibody-dependent enhancement ( ADE ) [41] and T cell original antigenic sin [42] . All our patients with results available had thrombocytopenia but only five had high hematocrit levels . Platelet count of less than 50×109/L concurrent with hemoconcentration has been shown to increase dengue mortality by six-fold [43] . Three patients ( Patient 3 , 4 and 7 ) had normal hematocrit levels despite clear evidence of severe plasma leakage . Patient 4 and 7 had clinical evidence of severe bleeding and Patient 3 may have occult bleeding . This may be the explanation for their ‘normal’ hematocrit levels at presentation . Therefore , hematocrit may not be a sensitive marker of plasma leakage in dengue with severe bleeding . In our study , elevated liver enzymes of >1000 IU/L was common . Equally common was hypoalbuminemia concurrent with elevated liver transaminases . Elevated liver transaminases and hypoalbuminemia could be a good indicator of vascular leakage or hepatic dysfunction in DHF and could be used as a significant marker in identifying cases of severe dengue [44] . Other clinical presentation and severe manifestation of dengue fever in our cohort of patients included involvement of other organs , leading to multi-organ failure . Many patients had acute hepatitis leading to liver impairment and coagulopathy . There was also evidence of tubular necrosis of the kidney , inflammation of the heart , and pulmonary hypertension on postmortem . One patient had cardiogenic shock with evidence of myocarditis , pericarditis , pericardial effusion and global left ventricular hypokinesia . The myocardial injuries could be secondary to the cellular immune response and the production of inflammatory cytokines , or as direct result of DENV infection of the myocardial tissue [45] , [46] . These uncommon presentations are often related to poor prognosis and associated with high mortality in severe dengue [47] , [48] , [49] . In our study , decompensated DSS with evidence of massive plasma leakage , massive bleeding , MOF and coagulopathy were the primary cause of deaths . Per rectum bleeding , vaginal bleeding and gastrointestinal bleeding were the commonest sites of severe bleeding . These observations are consistent with those reported in a number of other studies [30] , [50] . Here we also report hemorrhage over the brain hemispheres , the endocardium and septum of the heart and intra-alveolar demonstrated from autopsy . It has been suggested that several of these rare hemorrhagic manifestations has become more apparent and significant in DHF in the past 30 years and carries a higher risk of mortality [49] , [50] . The increasing reports of uncommon manifestations in dengue may be reflective of the shift in demographics where there is increasing incidence of DHF or severe dengue among the older age group of patients . The present study is a retrospective descriptive review . Study limitations include limited documentation of clinical information especially for patients who were brought in dead and those who died within 24 hours . This is also a single-center review involving a relatively small number of fatal cases , which may introduce bias in sample selection . A case control study is needed to determine if those common clinical and laboratory findings seen in our case series are exclusive to fatal cases of severe dengue or seen equally in the non-fatal cases . In conclusion , our study demonstrates a case series of severe dengue leading to death seen primarily in adult females with secondary dengue infection in Malaysia . The possible contributing role of underlying co-morbidities commonly seen in adults especially diabetes mellitus is highlighted . While most patients with fatal outcomes presented with common clinical and laboratory warning signs of severe dengue seen in all ages , the manifestation of uncommon clinical presentations of dengue is most likely a reflection of a change in the demographic pattern of the population being infected with dengue . | Dengue continues to be a major mosquito-borne disease of serious public health concern . Children are usually the most affected group , but in recent decades , dengue and severe dengue have become more common among adults . Here we reviewed ten fatal dengue cases with a view to determine if there have been changes in the presentation of severe to fatal dengue . Our findings revealed high dengue mortality among adult females , associated with secondary dengue infection . Underlying co-morbidities particularly diabetes mellitus were common in our cohort . This may contribute to the rapid deterioration of clinical condition seen in severe dengue . Most patients presented with common clinical and laboratory warning signs for severe dengue . Uncommon presentations seen among our fatal cases are likely a reflection of the changing demography of dengue from children to more of an adult disease in dengue endemic countries . Our findings , hence , emphasize the importance of healthcare awareness of clinical warning signs for severe dengue , especially in adult females with underlying co-morbidities and secondary dengue infection . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"infectious",
"diseases",
"clinical",
"research",
"design",
"dengue",
"fever",
"neglected",
"tropical",
"diseases",
"retrospective",
"studies",
"dengue",
"viral",
"diseases"
] | 2013 | Review of Dengue Hemorrhagic Fever Fatal Cases Seen Among Adults: A Retrospective Study |
Despite having only begun ∼10 , 000 years ago , the process of domestication has resulted in a degree of phenotypic variation within individual species normally associated with much deeper evolutionary time scales . Though many variable traits found in domestic animals are the result of relatively recent human-mediated selection , uncertainty remains as to whether the modern ubiquity of long-standing variable traits such as coat color results from selection or drift , and whether the underlying alleles were present in the wild ancestor or appeared after domestication began . Here , through an investigation of sequence diversity at the porcine melanocortin receptor 1 ( MC1R ) locus , we provide evidence that wild and domestic pig ( Sus scrofa ) haplotypes from China and Europe are the result of strikingly different selection pressures , and that coat color variation is the result of intentional selection for alleles that appeared after the advent of domestication . Asian and European wild boar ( evolutionarily distinct subspecies ) differed only by synonymous substitutions , demonstrating that camouflage coat color is maintained by purifying selection . In domestic pigs , however , each of nine unique mutations altered the amino acid sequence thus generating coat color diversity . Most domestic MC1R alleles differed by more than one mutation from the wild-type , implying a long history of strong positive selection for coat color variants , during which time humans have cherry-picked rare mutations that would be quickly eliminated in wild contexts . This pattern demonstrates that coat color phenotypes result from direct human selection and not via a simple relaxation of natural selective pressures .
The sizes , shapes , and colors among domestic animals vary significantly more than that of their wild counterparts , often reflecting variation normally associated with genus or family level divergence [1] . Domestication therefore provides an ideal model to test numerous evolutionary questions including the relationship between molecular and morphological change , how the intensification of the relationship between humans and wild plants and animals have altered both players' genetic and phenotypic constitutions , and whether changes associated with domestication resulted primarily from a release of natural selection pressure , selection on standing genetic variation present in the wild ancestor , or positive selection on novel mutations that have occurred subsequent to domestication . Coat color variation in domestic animals is of considerable interest in this respect considering that it can be traced back to at least 5 , 000 years before present when it was documented by administrative officers who recorded the coat color of livestock during the UR III dynasty in Mesopotamia [2] . Modern domestic animal species display a bewildering diversity in coat color , and the melanocortin receptor 1 ( MC1R ) locus is most consistently polymorphic , having been previously documented and associated with coat color variation in horses , cattle , foxes , pigs , sheep , dogs , and chickens [3]–[10] . MC1R is a G protein-coupled receptor that is primarily expressed in melanocytes and plays a key role in melanogenesis by determining the switch between production of red/yellow pheomelanin and dark eumelanin [11] . The binding of melanocyte stimulating hormone ( MSH ) to MC1R induces synthesis of eumelanin , whereas in the absence of MC1R signaling , melanocytes produce only pheomelanin . Loss-of-function mutations are therefore associated with recessive red coat color , whereas dominant black coloring is linked with mutations causing constitutive activation of MC1R signaling . We have previously described the molecular basis for an allelic series at the classical Extension locus ( equivalent to MC1R ) in pigs [6] , [7] , [12] . The wild-type ( E+ ) allele allows full expression of both pheomelanin and eumelanin . The dominant black color results from two different mutations , each of which evolved independently in Asia and Europe . The ED1 allele is Asian in origin and is associated with an L102P missense mutation , and ED2 is European and associated with a D124N substitution . The recessive red allele ( e ) possesses two missense mutations A164V and A243T , though it is not clear if one or both of these are responsible for the phenotype . The most interesting allele is EP which causes black spotting on a red or white background . This allele evolved from ED2 and possesses , in addition to the D124N substitution for dominant black coloring , a two base pair insertion at codon 22 . Two C nucleotides have been inserted in a stretch of six Cs , thus extending a short mononucleotide repeat . The resulting frameshift is expected to cause a uniform red pigmentation due to the complete loss of MC1R signaling , but because the mononucleotide repeat is somatically unstable , MC1R function in some melanocyte lineages is occasionally restored , resulting in the appearance of black spots [7] . All of these alleles were described from a limited subset of wild and domestic pigs . In order to more fully understand the range of MC1R variations within Sus scrofa , we determined the entire MC1R coding sequence ( 963 bp ) within 68 domestic pigs ( representing 51 Asian and European breeds ) and a total of 15 Chinese and European wild boar , the results of which are described in Tables S1 , S2 , S3 , S4 . In addition , one previously published Japanese wild boar sequence [13] was incorporated into the analysis .
The sampling strategy outlined above resulted in a near doubling of the number of described MC1R alleles from seven to 13 , all of which belonged to the five previously described allelic groups . A large proportion of the domestic pigs ( 60 out of 68 ) were homozygous at MC1R despite the fact that eight different alleles were identified within domestic pigs ( Table S3 ) . This high degree of homozygosity is not surprising given that coat colors have been used as a specific breed characteristic over the past 200 years , and within most populations there has been strong selection for uniformity in coat color as this trait often defines the breed . This screen revealed three new missense mutations in pig MC1R , Val122Ile in the Asian *0202 allele , Ala21Thr in the European *0502 allele and finally Arg166Trp in the European *0503 ( Table S2 ) . Since these variants were detected in a few pigs that may carry other coat color mutations we cannot judge if they have an impact on the coat color phenotype and this needs to be further investigated . Until such data become available we assume that they are associated with dominant black color ( *0202 ) and black spotting ( *0502 and *0503 ) . All domestic breeds in Europe and China carried mutant MC1R alleles except the Hungarian Mangalica . Though this breed is homozygous for the European wild type MC1R allele , it still possesses a variable coat color phenotype , a fact at least partially explained by allelic segregation at the agouti ( ASIP ) locus [14] . The EP ( black spotting ) MC1R allele dominates among European domestic pigs , particularly amongst commercial populations ( Table S4 ) . The allele for dominant black color ( ED2 ) is more common in local European breeds whereas the allele ( e ) for recessive red color was only found in the Duroc and Leicoma breeds . One Creole pig carried the Asian ED1 allele for dominant black color , reflecting the introgression of Asian pigs into European stocks that took place during the 18th and 19th century [13] . All tested Chinese pigs carried the ED1 allele for dominant black color . Lastly , three Chinese domestic pigs were heterozygous for the European e or EP alleles , an observation that is most certainly caused by recent introgression of European germplasm into local Chinese breeds . The European , Chinese , and Japanese wild boars all carried wild-type alleles that differ only by synonymous substitutions ( Tables S2 , S3 ) . All 12 European wild boar carried identical MC1R sequences which were also identical to a previously described European wild boar sequence [13] and to the one present in Mangalica domestic pigs . The complete absence of synonymous substitutions between European wild boar and European domestic pig MC1R sequences supports the notion that the European wild boar experienced a population bottleneck prior to domestication [15] , [16] . The higher MC1R diversity within Chinese wild boar relative to European wild boar is also consistent with recent microsatellite data indicating that Asian domestic pigs were derived from a more diverse wild boar population [17] . One Chinese wild boar was heterozygous for the EP allele but this must be the result of gene flow from domestic pigs since the allele is of European origin and differs from native Chinese wild boar alleles by at least two synonymous and two non-synonymous substitutions ( Table S2 ) . The amino acid sequence conservation within wild boar is sharply contrasted by the sequence diversity amongst MC1R alleles found within domestic pigs . This is illustrated in Figures 1A and 1B which depict the completely synonymous nature of all seven nucleotide substitutions found among European and Asian wild boar . The figure also shows that all mutations except one detected amongst European and Chinese domestic pigs altered the amino acid sequence , eight are non-synonymous substitutions and one is the frameshift at codon 22 that is widespread among European domestic pigs . The same synonymous substitution was found in both the Chinese domestic allele *0203 and in the Asian wild boar allele *0104 ( Figure 1B ) . The existence of radically different selective pressures acting on MC1R in wild and domestic pigs was demonstrated by an analysis of the relative frequencies of non-synonymous ( dN ) and synonymous substitutions ( dS ) ( Table 1 ) ; a dN/dS ratio higher than 1 . 0 indicates positive selection while a dN/dS ratio significantly lower than 1 . 0 provides evidence for purifying selection . The most informative contrasts were those between Asian and Europan wild boars and between Asian and European domestic pigs . The former comparison provided evidence for purifying selection in wild boars since no non-synonymous substitution was observed and the frequency of synonymous substitutions was significantly different from 0 ( Z = 2 . 55 , P<0 . 01 ) . In contrast , the dN/dS ratio was as high as 23 . 5 for the comparison between Asian and European domestic pigs , ignoring the three synonymous substitutions that distinguish Asian from European alleles but which must have occurred in wild boars prior to domestication ( Figure 1B ) . The excess of non-synonymous substitutions between European and Asian domestic pigs was statistically significant ( Z = 2 . 05; P<0 . 01 ) and provided strong support for positive selection . This analysis in fact underestimates the evidence for diversifying selection since the dN/dS analysis did not take into account the frameshift mutation that is widespread among European domestic pigs . Furthermore , the large number of non-synonymous substitutions cannot be explained by relaxed selection subsequent to pig domestication . The reason is that the time since domestication ( about 10 , 000 years ) is not sufficiently long to accumulate as many mutations that have gone to fixation in domestic populations or the existence of several alleles that differ by multiple steps from the ancestral form . We have previously estimated the time since divergence of European and Asian wild boar populations to about 900 , 000 years before present based on a 1 . 2% sequence difference across the entire mitochondrial genome [18] . This is very similar to the recently estimated sequence divergence of about 1 . 3% for the entire mitochondrial genome between modern humans and Neandertals and the time since divergence of these two mtDNA lineages were estimated at 660 , 000±140 , 000 years [19] . Thus , the seven synonymous substitutions observed among Asian and European wild boars have accumulated during hundreds of thousands of years whereas the nine mutations changing the MC1R protein sequence among domestic pigs have accumulated within the last 10 , 000 years . Thus , we refute relaxed selection as a possible explanation for the fast evolution of MC1R diversity in domestic pigs and conclude that the only reasonable explanation for this plethora of diversity is that humans have cherry-picked novel mutations with favorable phenotypic effects during the course of domestication . A reconstructed evolutionary history of porcine MC1R sequences ( Figure 1B ) demonstrates not only the pattern of synonymous substitutions ( Table S2 ) that differentiate Asian and European wild boar sequences , but also the fact that domestic pig MC1R alleles are directly derived from wild boar from the same region , a conclusion supported by previous publications that investigated other loci [13] , [18] , [20] .
The MC1R diversity within Europe and China presented here reveals divergent selection pressures in wild and domestic pigs . A previous study of mtDNA genome sequences derived from European and Asian wild boars indicated that these subspecies diverged well before the advent of domestication [18] . In this study , we identified a total of seven synonymous nucleotide substitutions within MC1R sequences from Asian and European wild boar . The temporally and geographically conserved nature of Sus MC1R amino acid sequence demonstrates that purifying selection is the dominant mode of evolution at this locus in the wild boar because normal MC1R function , which generates a mixture of dark eumelanin and red/yellow pheomelanin , is necessary for a camouflaged coat . MC1R underlies coat color diversity in many natural and domestic populations . This study demonstrates that this diversity cannot be explained by an exceptionally high substitution rate at this locus . Instead , a more likely explanation of this pattern is that though MC1R is a master regulator for melanogenesis , its primary role is restricted to melanocytes . Thus , MC1R mutations can have major effects on coat color phenotypes without causing severe pleiotropic effects on other tissues , a phenomenon that often results from mutations at other coat color loci [21] . Because domestication necessarily involves the separation of animals from their natural environment , the alterations in coat color during animal domestication could have been the result of a relaxation of the selection pressure against non-camouflaged coat . The evidence presented here , however , demonstrates that relaxed selection alone cannot explain the observed MC1R diversity in domestic pigs . The complete lack of non-synonymous substitutions within European and Chinese wild boar ( who diverged in the Pleistocene ) is contrasted sharply by the nine separate amino acid-altering mutations present within domestic pigs . This pattern implies that naturally occurring mutations that altered camouflaged coat colors were quickly eliminated in the wild , but within a domesticated context , these mutations were prized and positively selected . Having been allowed to proliferate , the new mutated alleles served as templates on which additional mutations occurred ( Figure 1B ) . This conclusion is underscored by the fact that seven of the eight MC1R alleles found in domestic pigs differed by two or more non-synonymous substitutions from the wild-type . In fact , three of the alleles ( *0202 , *0502 and *0503 ) differed by as many as three non-synonymous substitutions from the wild-type . We propose that an important component of the selection at coat color loci in domestic animals has been direct selection for non-camouflaged patterns , since the mutated coat colors may have facilitated animal husbandry . The traditional practice of keeping domestic pigs involved allowing them to forage beyond the boundaries of the farm . Possessing a coat color significantly different from the wild type would have made it easier to recognize and keep track of domestic stocks . Furthermore , variant coat color phenotypes may also have been used as markers associated with improved domesticated forms . Finally , the ubiquity of the black spotting caused by the EP allele , may be a result of the universal human penchant to select for and propagate attractive patterns present within the floral and fauna with which humans have a close relationship . Three recent illustrative examples of human preference for novelty include selection for white color in horses [22] , the human-mediated spread of white Polynesian tree snails between Polynesian islands [23] , and the global proliferation of white grape cultivars , which , like some domestic coat colors , results from the inactivation of a gene necessary for wild type ( red ) grape coloring [24] . Black coloring is common in both European and Asian domestic pigs but because different allelic substitutions underlie this indistinguishable phenotype , it is clear that there have been independent selective sweeps for separate alleles causing dominant black color in both China and Europe . In China , selection for the L102P substitution has been strong enough to penetrate all 23 breeds representing the six traditional types of Chinese pigs . Similarly , 90% of MC1R sequences found in European domestic pigs carried the D124N missense mutation causing dominant black color . Despite dramatic phenotypic differences between wild and domestic animals , no fixed mutation has yet been identified that distinguishes domesticated and wild forms of a given species . The MC1R locus in pigs approaches such a diagnostic status since only one breed ( the Mangalica ) out of 51 included in this study carried the wild-type allele . Interestingly , to the best of our knowledge , Mangalica pigs are also the only pigs included in this study that give birth to striped piglets that resemble striped wild boar piglets ( Figure 1C ) . This indicates that normal MC1R signaling is required for the development of stripes in piglets and it may therefore also be important for the development of stripes in other species like the Tiger and Zebra . The striping that occurs in wild boar piglets is most certainly a camouflage pattern ( Figure 1C ) and it is plausible that selection against this pattern facilitated early animal husbandry . This study provides an interesting insight into a long-standing evolutionary question . The dramatic phenotypic differences between wild and domestic taxa could either be the result of altered selection upon pre-existing variation in the wild ancestor once under human control , or the result of selection for new mutations with major effects that arise after the domestication process has begun . The MC1R locus in pigs provides a clear example of the latter case . Specifically , the EP allele provides particularly strong evidence for this since it involves two mutations with distinct phenotypic effects: the D124N missense mutation causes dominant black color , and the two base pair insertion at codon 22 causes a frameshift that results in inactivation of the MC1R locus . The combination of these two mutations leads to a unique allele and resulting coat color , neither of which could have evolved in a wild context since the appearance of either of the two mutations would have been eliminated from the population by purifying selection . The additive nature of mutations found within domestic MC1R alleles proves that shifts in selection pressure can lead to rapid phenotypic change as newly favored alleles form the templates on which additional mutations can be added to create novel phenotypic effects . It is common knowledge that there has been selection acting on coat color phenotypes in domestic animals since color has been used as breed characteristics . However , breed formation is a very recent phenomenon ( within the last few hundred years ) and the possibility that coat color variants have been accumulating due to relaxed selection since domestication and then been under strong selection during breed formation has not been excluded . This study demonstrates that selection on coat color phenotypes has a much older history and must trace back to the early period of domestication .
DNA samples from 68 unrelated domestic pigs ( no common grandparents ) were included in the study . The material comprised 45 European pigs representing 31 populations and 23 Chinese pigs representing 19 breeds . The Chinese pigs included members from each of the six traditional types of Chinese indigenous pig breeds described by Zhang [25] . In addition , three Chinese wild boars from two different regions and 12 European wild boars from Poland were included . Genomic DNA was extracted from blood by standard methods or from hair roots by Chelex extraction . An MC1R fragment including the entire coding region plus 43 bp of 5′-UTR and 208 bp of 3′-UTR was amplified with the forward EPIG16 primer [7] and the reverse primer PCR2 ( 5′-CGCCGTCTCTCCAGCCTCCCCCACTC-3′ ) . PCR was carried out on a DNA Engine Gradient Cycler ( Perkin Elmer , Norwalk , Connecticut , USA ) in a total volume of 35 µl containing 70 ng genomic DNA , 0 . 5 mM of both primers , 0 . 3 mM dNTPs , 1× PCR Gold Buffer , 2 U AmpliTaq Gold enzyme ( Perkin Elmer , Norwalk , Connecticut , USA ) , 1 M Betaine ( SIGMA , Missouri , USA ) and 2 . 5 mM MgCl2 . The PCR profile was 45 cycles of 45 s each at 94°C , 63 . 5°C and 72°C; the cycles were preceded by 7 min at 94°C and terminated with 10 min at 72°C . Three other sequencing primers ( Seq1: 5′-GTCCATCTTCTACGCGCT-3′; Seq2: 5′-GGTGGTAGTAGGCGATGA-3′ and Seq3: 5′-GGT TCT TGG CGA TGG CGG –3′ ) were designed to produce a final 963/965 bp sequence including the entire MC1R coding region . The sequences from one sample were compiled into a single contiguous fragment using SEQUENCHER ( Gene Codes , Ann Arbor , Michigan , USA ) . Ambiguous positions were verified by resequencing . The haplotypes carried by heterozygous animals were deduced by sequencing cloned PCR products . Heterozygosity for the Ep mutation comprising a two bp insertion was further verified by fragment analysis as previously described [7] . A network tree of the MC1R sequences was constructed by hand . The relative frequencies of non-synonymous ( dN ) and synonymous ( dS ) substitutions were calculated using the Nei-Gojobori method in MEGA4 [26] , [27] . Standard errors were estimated with a bootstrap procedure ( 500 replicates ) and the statistical significance of differences in dN and dS values was calcuated with a Z test [27] , [28] . We propose a new more informative nomenclature for porcine MC1R alleles ( Table S1 ) . The nomenclature is composed of four digits where the first two digits are used to distinguish alleles that show documented phenotypic differences and corresponds to the five previously described alleles . The two last digits are used to distinguish alleles presumed to be associated with the same phenotype . All novel sequences were submitted to GenBank with reference numbers EU443644-EU443726 . | This study addresses why coat colors of domestic animals are so variable , while those of their wild ancestors are so uniform . Specifically , we asked whether this was the result of ( i ) relaxed purifying selection , ( ii ) that some mutations affect both coat color and another trait under strong selection ( for instance behavior ) , or ( iii ) direct human selection for altered coat color phenotypes . We investigated genetic variation in the melanocortin receptor 1 ( MC1R ) gene among wild and domestic pigs from both Europe and Asia . Though we found a similar number of mutations in wild and domestic pigs , the nature of the mutations was strikingly different . All mutations found among wild boars were silent , i . e . , they did not change the protein sequence . This implies strong purifying selection in the wild that maintains camouflage coat color . In contrast , nine out of ten mutations found in domestic pigs altered the protein sequence , thereby drastically transforming the resulting coat color . These results demonstrate that early farmers intentionally selected pigs with novel coat coloring . Their motivations could have been as simple as a preference for the exotic or selection for reduced camouflage to facilitate animal husbandry and/or to make the domesticated forms distinct from their wild ancestor . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics",
"evolutionary",
"biology/evolutionary",
"and",
"comparative",
"genetics"
] | 2009 | Contrasting Mode of Evolution at a Coat Color Locus in Wild and Domestic Pigs |
The role of Type I interferon ( IFN ) during pathogenic HIV and SIV infections remains unclear , with conflicting observations suggesting protective versus immunopathological effects . We therefore examined the effect of IFNα/β on T cell death and viremia in HIV infection . Ex vivo analysis of eight pro- and anti-apoptotic molecules in chronic HIV-1 infection revealed that pro-apoptotic Bak was increased in CD4+ T cells and correlated directly with sensitivity to CD95/Fas-mediated apoptosis and inversely with CD4+ T cell counts . Apoptosis sensitivity and Bak expression were primarily increased in effector memory T cells . Knockdown of Bak by RNA interference inhibited CD95/Fas-induced death of T cells from HIV-1-infected individuals . In HIV-1-infected patients , IFNα-stimulated gene expression correlated positively with ex vivo T cell Bak levels , CD95/Fas-mediated apoptosis and viremia and negatively with CD4+ T cell counts . In vitro IFNα/β stimulation enhanced Bak expression , CD95/Fas expression and CD95/Fas-mediated apoptosis in healthy donor T cells and induced death of HIV-specific CD8+ T cells from HIV-1-infected patients . HIV-1 in vitro sensitized T cells to CD95/Fas-induced apoptosis and this was Toll-like receptor ( TLR ) 7/9- and Type I IFN-dependent . This sensitization by HIV-1 was due to an indirect effect on T cells , as it occurred in peripheral blood mononuclear cell cultures but not purified CD4+ T cells . Finally , peak IFNα levels and viral loads correlated negatively during acute SIV infection suggesting a potential antiviral effect , but positively during chronic SIV infection indicating that either the virus drives IFNα production or IFNα may facilitate loss of viral control . The above findings indicate stage-specific opposing effects of Type I IFNs during HIV-1 infection and suggest a novel mechanism by which these cytokines contribute to T cell depletion , dysregulation of cellular immunity and disease progression .
Pathogenic HIV-1 infections are characterized by a generalized immune activation with concomitant CD4+ T cell depletion and the failure to effectively control viral replication . Increased apoptosis of uninfected T cells is observed in HIV-1-infected individuals and positively correlates with disease progression [1] . CD4+ T cells and CD8+ T cells from HIV-1-infected patients undergo elevated spontaneous apoptosis , activation-induced cell death ( AICD ) , and CD95/Fas-mediated apoptosis [2] , [3] , [4] , [5] , [6] , [7] . Although the exact mechanisms underlying this apoptosis are largely unknown , HIV-1-induced immune activation may contribute to the destruction of T cells and acquired immunodeficiency syndrome ( AIDS ) progression [8] , implicating a role for cytokines in sensitizing T cells for apoptosis . Type I IFNs ( IFNα/β ) are antiviral cytokines that are synthesized in response to the activation of molecular pattern recognition receptors by virus-specific molecules . Plasmacytoid dendritic cells ( pDC ) produce the majority of IFNα in response to Toll-like receptor ( TLR ) 7 and TLR9 activation by HIV-1 [9] . Accordingly , IFNα is detected at elevated levels in the sera of HIV-1-infected and AIDS patients [10] , [11] , [12] . Type I IFN modulates innate and adaptive immune responses by decreasing viral replication [13] in a cell type-specific manner [14] , regulating the differentiation of antigen-presenting cells [15] , and promoting the proliferation or death of T cells [16] . Despite its well-characterized antiviral activity , the role of IFNα/β in HIV-1 infection remains controversial , with conflicting studies suggesting protective versus deleterious effects on host immunity . Although the level of immune activation in the early stages of HIV-1 infection is predictive of disease outcome [17] , [18] , the kinetics of Type I IFN production in relation to CD4+ T cell loss and viral control over time is largely unknown . IFNα production may be beneficial , as it has been demonstrated to directly suppress HIV-1 replication in vitro [19] . The administration of recombinant IFNα to HIV-1-infected individuals provides a modest therapeutic benefit [20] , [21] , but decreases CD4+ T cell counts during more advanced stages of HIV-1 disease [22] . In individuals infected with HIV-1 and rhesus macaques infected with simian immunodeficiency virus ( SIV ) , plasma levels of IFNα remain elevated over time [10] , as opposed to natural hosts of SIV [9] , [23] and this may contribute to immunosuppression . In addition , persistent elevated IFNα production may impair thymopoiesis , bias T cell selection and induce generalized immune activation [24] . Furthermore , the expression of IFNα , Fas/FasL , and tumor necrosis factor–related apoptosis-inducing ligand ( TRAIL ) /death receptor ( DR ) 5 is increased in the lymphoid tonsillar tissue of individuals with progressive HIV-1 disease , as compared to non-progressors who do not exhibit CD4+ T cell depletion [25] . It is therefore possible that Type I IFN exerts pathogenic effects during HIV-1 infection by facilitating the apoptotic death of T cells through mechanisms involving the tumor necrosis factor ( TNF ) receptor family . To understand the role of Type I IFN in priming T cells for apoptosis , we investigated the effect of IFNα/β on pro- and anti-apoptotic molecules and the apoptosis sensitivity of T cells in HIV-1-infected patients . We found that ex vivo CD4+ T cells and CD8+ T cells , as well as HIV-specific CD8+ T cells from individuals with chronic HIV-1 infection are primed to undergo apoptosis through CD95/Fas but not TRAIL or TNFα signaling . Although Bak , Bax and Bim levels were significantly increased in T cells from HIV-1-infected patients , only Bak expression correlated inversely with CD4+ T cell counts and positively with CD95/Fas-mediated death . In addition , Bak knockdown by RNA interference prevented this apoptosis . Upon investigating what drives this apoptotic phenotype , we found that IFNα/β increased Bak expression and the sensitivity of healthy donor T cells , as well as HIV-specific CD8+ T cells to CD95/Fas-mediated death . In support of these findings , non-infectious HIV-1 sensitized healthy T cells to CD95/Fas-induced death and this apoptosis induction was mediated by TLR7/9 and Type I IFN . Further evidence for the pro-apoptotic role of Type I IFN and Bak in HIV-1 infection was revealed by our finding that ex vivo expression of the IFNα-induced genes [26] in PBMC from HIV-1-infected patients was increased and positively associated with Bak expression levels , CD95/Fas-mediated T cell apoptosis sensitivity and high viral loads , while it correlated inversely with CD4+ T cell counts . An inverse correlation between plasma IFNα levels and CD4+ T cell counts was also observed during chronic SIV infection . Finally , we show that although peak levels of plasma IFNα in SIV-infected rhesus macaques appear to contribute to viral control during the early stages of infection , persistent IFNα production was ultimately associated with loss of viral control and disease progression . Our findings indicate that although in acute infection Type I IFN may exert an important antiviral effect , during chronic HIV-1 infection , IFNα/β may contribute to the loss of CD4+ T cells and the failure of HIV-specific CD8+ T cells to control viral replication by upregulating Bak and sensitizing T cells to CD95/Fas-induced apoptosis . This study also suggests that Type I IFN has both beneficial and deleterious effects on the host immune response to pathogenic HIV-1 and SIV , depending on the stage of infection .
We and others have previously reported that T cells from HIV-1-infected individuals are highly susceptible to CD95/Fas-induced apoptosis [5] , [6] , [7] , [27] , [28] , [29] . However , the role of other TNF receptor family members in mediating increased T cell death in HIV-1-infected individuals is not well characterized . To address this , we examined the effects of TNFα and TRAIL on the survival of total CD4+ T cells , CD8+ T cells , as well as HIV- , CMV- and EBV-specific CD8+ T cells from HIV-1-infected patients . PBMC from HIV-1-infected subjects were incubated with media alone or with anti-CD95 antibody , TNFα or TRAIL for 14 hours and subsequently stained for apoptosis . While total CD4+ T cells , CD8+ T cells , HIV-specific and CMV/EBV-specific CD8+ T cells were susceptible to CD95/Fas-induced apoptosis ( Figure 1A and B ) , neither TNFα nor TRAIL increased apoptosis above levels of spontaneous death ( Figure 1A and B ) . In contrast , treatment of Jurkat T cells with anti-CD95/Fas antibody , SuperKiller TRAIL alone or TNFα with cyclohexamide induced similar apoptosis levels ( data not shown ) . Furthermore , the enhanced susceptibility of CD4+ T cells to CD95/Fas-mediated death was strongly associated with CD4+ T cell loss in patients ( Figure 1C ) . These in vitro data suggest that CD95/Fas is a predominant apoptotic pathway of the TNF receptor family in HIV-1 disease which contributes to the depletion of CD4+ T cells and impairs cytotoxic T cells . We have previously demonstrated that Bcl-2 and Bcl-xL levels are significantly reduced in HIV-specific CD8+ T cells and correlate with high sensitivity to apoptosis [30] . Additionally , pro-apoptotic Bim and Bak are reported to be upregulated in CD4+ T cells from SIV-infected rhesus macaques [31] . However , the involvement of other Bcl-2-related proteins in T cell apoptosis during chronic HIV-1 infection is largely unknown . In order to address this , we examined ex vivo levels of pro-apoptotic members of the Bcl-2 family and other signaling molecules of the CD95/Fas pathway in T cells from HIV-1-infected subjects and uninfected donors . The expression of Bak , Bax and Bim was significantly increased in CD4+ T cells from HIV-1-infected patients , relative to healthy donors ( Figure 2A; Figure S1; Table 1 ) . Bak and Bax were also upregulated in CD8+ T cells from HIV-1-infected individuals , while Bim levels , although higher , did not significantly differ between HIV-1-infected and uninfected controls ( Figure 2A , Table 1 ) . A significant inverse correlation was observed between Bak expression and the absolute numbers of CD4+ T cells in HIV-1-infected patients ( Figure 2B ) , whereas Bax and Bim were not associated with CD4+ T cell loss in vivo ( Figure 2B ) . As previously reported in HIV-1-infected patients [28] , [30] , [32] , Bcl-2 was decreased in CD8+ T cells ( Table 1 ) . Ex vivo levels of Bid and FADD in CD4+ T cells and CD8+ T cells did not differ between healthy donors and HIV-1-infected individuals ( Table 1 ) . These data suggest that while CD8+ T cells in chronic HIV-1 infection exhibit low levels of anti-apoptotic Bcl-2 , pro-apoptotic Bak is increased and may be related to CD4+ T cell depletion . In order to determine whether ex vivo expression of Bak is predicative of cell death , we correlated Bak expression and T cell apoptosis sensitivity . We found a significant positive correlation between Bak expression and CD95/Fas-induced apoptosis of CD4+ T cells and CD8+ T cells , but not with spontaneous apoptosis or AICD ( Figure 2C ) . To confirm the involvement of Bak in CD95/Fas-induced death of T cells , we performed Bak knockdown using siRNA ( Figure 3A–B ) . Targeting of Bak with siRNA resulted in a marked reduction of Bak expression ( Figure 3B ) , with minimal off-target effects ( Figure 3C ) . The introduction of siRNA directed against Bak into T cells from chronically HIV-1-infected patients caused a marked reduction in apoptosis induced by CD95/Fas cross-linking , compared to a control siRNA pool ( Figure 3D–E ) . Therefore , reducing Bak expression drastically decreases the susceptibility of T cells in chronic HIV-1 infection to CD95/Fas-mediated death . These findings demonstrate that Bak is increased in T cells during chronic HIV-1 infection , correlates with CD4+ T cell depletion and is directly involved in CD95/Fas apoptosis . HIV-1 disease progression is characterized by well-defined changes in the composition of circulating lymphocyte subpopulations , including the gradual loss of naïve and memory T cells [33] . We therefore wanted to determine whether the level of Fas apoptosis susceptibility varies between T cell differentiation states . Indeed , Fas apoptosis sensitivity was higher in CD45RA−CD62L− effector memory populations and in terminally differentiated CD45RA+CD62L− effector memory T cells ( Figure S2A ) . In contrast , CD45RA+CD62L+ naïve T cells and CD45RA-CD62L+ central memory populations were resistant to apoptosis ( Figure S2A ) . In line with these data , Bak levels were significantly elevated in effector memory subpopulations , as compared to naïve T cells ( Figure S2B ) . Increased levels of Bak were also observed in central memory T cells , relative to naïve T cells ( Figure S2B ) . However , increased Bak in central memory T cells may be counterbalanced by elevated Bcl-xL expression in this specific subset , thus potentially accounting for the relative resistance of these cells to apoptosis ( Figure S2C ) . Therefore , CD95/Fas-mediated death and Bak expression in CD4+ T cells and CD8+ T cells is associated with the differentiation stage of these cells . We next examined whether the magnitude of plasma viremia in HIV-1-infected subjects affects the expression levels of Bak in peripheral CD4+ T cells . Total ex vivo levels of Bak were significantly increased in CD4+ T cells from ART-treated patients with high-level viremia ( ≥1000 HIV-1 RNA copies/ml of plasma; Bak MFI = 375±44 ) , as compared to patients with low-level viremia ( <1000 HIV-1 RNA copies/ml of plasma; Bak MFI = 253±30 ) ( Figure S3A ) . In contrast , CD4+ T cells from HIV-1-infected individuals with high viral loads did not exhibit elevated levels of Bax ( MFI = 271±56 vs . 296±52 ) or Bim ( MFI = 767±125 vs . 673±29 ) ( Figure S3B–C ) . Therefore , high levels of HIV-1 in the peripheral blood plasma may result in increased pro-apoptotic Bak expression and this could influence the sensitivity of CD4+ T cells to CD95/Fas-induced death . In order to address the question of what drives Bak overexpression in HIV-1 infection , we examined whether IFNα/β alters Bak levels and increases T cell apoptosis . PBMC from healthy individuals were treated with Type I IFN for 3 days in the absence or presence of CD3-induced activation . Type I IFN significantly increased the expression of Bak in both resting and activated CD4+ T cells and CD8+ T cells ( Figure 4A; Figure S4 ) . In addition , both CD4+ T cells and CD8+ T cells cultured in the presence of IFNα exhibited significant upregulation of CD95 expression ( Figure S5A–B ) . Type I IFN did not affect the expression levels of Bcl-2 , Bcl-xL , FADD , Bid , Bim or Bax in CD4+ T and CD8+ T cells ( Table 2 ) . When we tested T cell apoptosis sensitivity after 3 days of pre-incubation with Type I IFN , we found a 3-fold increase in CD95/Fas-induced apoptosis of resting and activated CD4+ T cells treated with IFNα/β compared to untreated cells ( Figure 4B ) . A similar effect was seen in CD8+ T cells ( Figure 4B ) . While Type I IFN induced slight increases in AICD , it did not enhance the spontaneous death rate of T cells ( Figure 4B ) . As anticipated , a direct correlation was observed between Type I IFN-mediated Bak elevation in CD4+ T cells and increased sensitivity to CD95/Fas-induced death ( Figure S5C ) , but not spontaneous apoptosis or AICD ( data not shown ) . IFN treatment also increased CD4+ T cell and CD8+ T cell apoptosis in HIV-1-infected patients ( Figure 5A ) . In contrast , neither TRAIL- nor TNFα-mediated apoptosis were increased by IFNα treatment ( Figure 5A ) . Additionally , Type I IFN increased the susceptibility of HIV-specific CD8+ T cells , but not CMV/EBV-specific from HIV-1-infected individuals to CD95/Fas-induced death ( Figure 5B ) . However , Type I IFN did not sensitize HIV-specific CD8+ T cells to TRAIL or TNFα-induced apoptosis ( Figure S6 ) . Taken together , the above findings indicate that IFNα/β enhances the sensitivity of T cells in HIV-1 infection to CD95/Fas-induced apoptosis , possibly through upregulation of Bak . Type I IFN may also be involved in the survival defect of HIV-specific CD8+ T cells . Because TLR7 and TLR9 recognition of HIV-1 and SIV in pDC triggers Type I IFN production [9] , [34] , [35] , we next evaluated whether HIV-1 exposure of PBMC from healthy individuals would lead to Type I IFN production and a subsequent increase in T cell apoptosis sensitivity . Although pDC are present at very low frequency in the blood , they produce 100 times more Type I IFN per cell than monocytes [36] . We exposed healthy PBMC to virus in the presence or absence of a TLR7/9-specific antagonist and IFNα/β receptor blocking antibodies and analyzed CD95/Fas-mediated apoptosis . PBMC exposed to 105 TCID50/ml of HIV-1Ba-L for 24 hours secreted IFNα , and this production was significantly reduced in the presence of a phosphorthioate-based TLR7/9 antagonist ( Figure 6A ) . The TLR antagonistic properties of the phosphorothioate deoxyribose compound that was used does not depend on a specific immunoregulatory DNA sequence [37] and can selectively suppress TLR7/9 activation , with no apparent cross-reactivity involving other TLRs [38] . The mechanism of TLR inhibition may be attributed to the ability of the inhibitor to bind TLR7 and TLR9 specifically [38] , or its capacity to physically interact with cognate TLR7/9 ligands [37] . This ultimately prevents receptor triggering , MyD88 activation [39] and downstream signaling . HIV-1-exposed CD4+ T cells and CD8+ T cells showed significantly increased CD95/Fas-mediated apoptosis compared to unexposed cells ( Figure 6B ) . This death was inhibited by a TLR7/9 antagonist or by blocking the IFNα/β receptor ( Figure 6B ) , suggesting that the enhanced apoptosis sensitivity observed in these cells was Type I IFN-dependent . To exclude the effect of productive viral infection on T cell death in these assays , we used AT-2-inactivated HIV-1 . The non-infectious AT-2 HIV-1Ba-L significantly augmented CD95/Fas-induced apoptosis of CD4+ T cells and this effect was abrogated in the presence of either anti-IFNα/β receptor blocking antibodies , or a TLR7/9-specific antagonist ( Figure 6C ) . Thus , triggering of TLR7/9 by HIV-1 elicits the production of Type I IFN which sensitizes T cells to undergo CD95/Fas-mediated apoptosis . To determine whether HIV-1-mediated induction of CD95/Fas apoptosis in CD4+ T cells is largely attributed to productive infection or bystander effects of the virus , we cultured purified CD4+ T cells and PBMC from the same healthy donors in parallel with HIV-1 for 72 hours , followed by a 14 hour stimulation with immobilized anti-CD95 antibodies . Prior to induction of CD95/Fas-induced apoptosis , Bak levels were quantified and the frequency of CD95-expressing lymphocytes was determined in the above cell populations . Notably , HIV-1 did not directly induce Fas apoptosis sensitivity , Bak upregulation or an increase in the frequency of CD95-expressing CD4+ T cells in purified CD4+ T cell cultures ( Figure S7A–C ) . However , CD95/Fas apoptosis sensitivity , Bak levels and the proportion of CD95-expressing lymphocytes were markedly elevated in the CD4+ T cell population present in HIV-1-exposed PBMC ( Figure S7A–C ) . These data suggest that increased CD95/Fas-mediated death of CD4+ T cells is an indirect consequence of HIV-1 exposure and not a direct viral cytopathic effect . To directly examine the relationship between Type I IFN , apoptosis sensitivity and CD4+ T cell counts in HIV-1-infected patients , we analyzed mRNA expression levels of the Type I IFN-induced genes , IFI6-16 and ISG56 in ART-naïve chronic HIV-1-infected patients . Both IFI6-16 and ISG56 mRNA levels were increased in PBMC from HIV-1-infected patients relative to healthy donors ( Figure 7A–7B ) . IFI6-16 levels directly correlated with ex vivo Bak expression in CD4+ T cells and CD8+ T cells ( Figure 7C ) and with CD95/Fas-mediated apoptosis sensitivity ( Figure 7F; Figure 7K ) . Similar correlations were found with ISG56 ( Figure 7J and data not shown ) . IFI6-16 levels did not correlate with Bax ( Figure 7D ) or Bim ( Figure 7E ) expression in ex vivo CD4+ T cells and CD8+ T cells from ART-naïve HIV-1-infected patients . Furthermore , IFI6-16 expression was negatively correlated with CD4+ T cell counts in HIV-1-infected patients with detectable viremia ( Figure 7G ) . Both IFI6-16 and ISG56 expression directly correlated with plasma viral load in HIV-1-infected ART-naïve patients ( Figure 7H–7I ) . As expected , plasma HIV-1 viremia strongly correlated with the specific death of CD4+ T cells triggered by CD95/Fas cross-linking ( Figure 7L ) . These results suggest that viremia drives Type I IFN production in chronic HIV-1 disease , which in turn increases Bak expression and apoptosis sensitivity , leading to accelerated rates of T cell depletion . In order to investigate the protective or pathogenic roles of Type I IFN in an animal model of HIV-1 infection , we examined the relationship between plasma levels of IFNα ( Figure 8A ) , viral loads ( Figure 8B ) and peripheral CD4+ T cell counts ( Figure 8C ) during acute and chronic SIV infection in rhesus macaques . We found that IFNα peaks early ( day 7 ) during acute infection ( Figure 8A ) and precedes the peak plasma viral load ( day 14; Figure 8B ) . These SIV-infected animals exhibited a typical viral load profile ( Figure 8B ) and CD4+ T cell depletion kinetics ( Figure 8C ) . Importantly , we found that the association of IFNα production with viremia was different during acute and chronic infection . Whereas peak IFNα levels were negatively associated with peak viremia during acute SIV infection ( Figure 8D ) , IFNα levels and plasma viral loads were positively correlated during chronic infection in the same animals ( Figure 8E ) . Similar to HIV-1-infected patients , we found that plasma IFNα levels were inversely correlated with CD4+ T cell counts during chronic SIV infection ( Figure 8F ) . These data indicate that although elevated IFNα secretion during acute infection may suppress peak viral replication , during chronic SIV infection , viremia-induced IFNα at low levels drives CD4+ T cell decline and the possible loss of viral control .
A hallmark of HIV-1 infection is the gradual loss of both CD4+ T cells and CD8+ T cells that has been associated with disease progression [40] . Although multiple factors may contribute to T cell depletion , exaggerated apoptosis of T cells is most closely related to CD4+ T cell loss [41] . Despite the large body of literature that has emerged concerning the death pathways responsible for T cell destruction during HIV-1 infection , the relative contributions of TNF receptor family members are a matter of debate . The increased susceptibility of CD4+ T cells from HIV-1-infected patients to CD95/Fas-mediated apoptosis has been previously established [5] , [6] , [7] , [27] , [28] , [29] , [42] . The importance of the CD95/Fas pathway is further supported by blocking FasL in SIV-infected macaques , which inhibits CD4+ T cell apoptosis and preserves CD4+ T cell counts following acute SIV infection [43] . Some studies have suggested a role for TRAIL [44] , [45] , [46] , [47] and TNF [48] , [49] , whereas other reports indicate that direct engagement of their receptors has a much weaker effect or does not increase T cell apoptosis [48] , [50] , [51] . Apoptosis may also impair anti-HIV immunity . HIV-specific CD8+ T cells are susceptible to CD95/Fas-induced apoptosis [7] , but their sensitivity to TRAIL and TNFα is currently unknown . We find that ex vivo T cells from HIV-1-infected patients are very sensitive to CD95/Fas-mediated apoptosis , but not to TNFα or TRAIL-induced death . Furthermore , consistent with our previous findings [5] , [7] , CD95/Fas stimulation markedly enhanced the susceptibility of HIV-specific CD8+ T cells to undergo apoptosis . In contrast , HIV-specific CD8+ T cells were not primed for apoptosis in response to stimulation with TRAIL or TNFα . These findings further suggest that CD95/Fas is the TNF receptor family member that is the critical mediator of CD4+ T cell and HIV-specific CD8+ T cell depletion during HIV-1 infection . The susceptibility of T cells in HIV-1 disease to CD95/Fas-induced death and relative resistance to TRAIL and TNFα-mediated killing may be attributed to differences in signaling induction of apoptosis between these pathways , including death receptor and ligand expression or the extent of cross-linking , the presence of death domains [52] , as well as regulation of their adapter molecules [53] , and preferential usage of apoptotic mitochondrial mediators such as Bak and Bax [54] . Elucidation of these exact differences is the subject of ongoing studies . Because direct viral cytopathic effects alone cannot account for the increased death of peripheral T cells in HIV-1 infection , it is important to determine what factors control bystander CD95/Fas T cell apoptosis sensitivity . We and others have previously shown that the level of CD95 expression on primary T cells from HIV-1-infected patients is not predictive of apoptosis sensitivity [7] , [55] . This suggests that intracellular signaling pathways are altered in these cells . In this study , we found that other major signaling molecules that associate with the CD95/Fas pathway [56] , including c-FLIP and FADD were not increased in T cells from HIV-1-infected patients . Therefore , signaling proximal to CD95 alone is not likely to account for this enhanced T cell apoptosis sensitivity . Mitochondria act as an amplification loop for CD95/Fas-mediated apoptotic signaling and triggering of CD95 leads to activation of Bcl-2 homology 3 ( BH3 ) -only proteins such as Bid and Bim , which act to compromise mitochondrial integrity either directly by activating Bak/Bax or indirectly by neutralizing Bcl-2 and Bcl-xL [57] , [58] . We have previously demonstrated that co-localization between CD95/Fas and mitochondria occurs early in HIV-specific CD8+ T cells upon death-receptor triggering [59] . A role for mitochondria in this apoptotic process is further supported by the reduced expression of anti-apoptotic Bcl-2 family members in T cells during HIV-1 infection [28] , [30] and our current study confirmed this . However , little is known about the pro-apoptotic proteins of the Bcl-2 family . Our data indicated that ex vivo Bak , Bax and Bim were significantly increased in T cells from HIV-1 infected individuals . Interestingly , only Bak was found to be inversely correlated with the loss of CD4+ T cells in vivo . Bak levels were also directly correlated with CD95/Fas apoptosis of T cells in HIV-1-infected patients and inhibiting Bak abrogated this cell death . High levels of Bak expression and elevated CD95/Fas-induced apoptosis of T cells were predictive of CD4+ T cell decline in vivo suggesting a pathogenic role for this death pathway . We also observed that effector memory T cells expressed abnormally high levels of Bak and were highly sensitive to Fas apoptosis , as compared to naïve cells . The relative resistance of central memory cells to apoptosis , despite elevated Bak expression may be attributed to significantly increased levels of Bcl-xL expressed in this specific subset . Increased apoptosis has been proposed as a major defect that affects both the differentiation and effector function of T cells during chronic HIV-1 infection [60] . Therefore , understanding how the critical balance of specific pro- and anti-apoptotic factors changes as a function of differentiation status may potentially permit us to elucidate the molecular pathways that control the depletion of particular subpopulations in HIV-1 disease . Interestingly , in chronic SIV infection , higher Bak expression has been demonstrated in CD4+ T cells from the lymph nodes of moderate progressors as compared to slow progressors [61] . Our studies have thus revealed a novel role for Bak upregulation in CD95/Fas T cell apoptosis and peripheral T cell depletion during HIV-1 infection . The mechanisms underlying the upregulation of Bak in T cells and the consequent increase in the death rate of T cells during HIV-1 and SIV infection are currently unknown . We find that Type I IFN induced Bak overexpression in T cells derived from healthy individuals and this was accompanied by increased sensitivity to CD95/Fas-mediated apoptosis . This suggested that in HIV-1 seropositive patients , Type I IFN may be responsible for accelerated T cell apoptosis and depletion . To test this hypothesis , we used IFN-stimulated gene ( ISG ) expression in PBMC from patients as a marker of in vivo IFNα exposure . ISGs are excellent surrogate indicators of IFN activity because they can be induced by very low levels of IFNα [23] . In HIV-1-infected patient PBMC , IFNα-stimulated gene expression was elevated and positively correlated with high Bak levels in T cells . Furthermore , IFNα-induced gene expression correlated directly with the sensitivity of T cells to CD95/Fas-induced death and inversely with absolute CD4+ T cell counts . ISG expression directly correlated with viremia , suggesting that viral loads in HIV-1-infected individuals drive Type I IFN production and this leads to accelerated T cell apoptosis and loss . This is further supported by the finding that patients with higher viral loads had higher Bak expression in CD4+ T cells . Accordingly , we also show that HIV-1 can induce CD95/Fas-induced T cell apoptosis in a TLR7/9- and Type I IFN-dependent manner and this effect does not require virus replication . Although CD95 expression is not a predictive correlate of apoptosis sensitivity , it is a necessary component involved in the delivery of the Fas apoptosis signal . We found that Type I IFN upregulated the expression levels of CD95 on healthy donor T cells . These findings , coupled with our observations that HIV-1 exposure markedly increased the frequency of CD95-expressing CD4+ T cells in PBMC , but not purified CD4+ T cells further suggests that Fas apoptosis sensitivity of T cells in HIV-1 disease is a bystander effect of the virus which is likely mediated by IFN production by a cell population other than the T lymphocytes , most likely pDCs [9] , [34] , [35] , [36] . IFNα may have additional deleterious effects by inducing inhibitory receptors such as programmed cell death-1 ( PD-1 ) on T cells [62] . Subsequent studies are required to elucidate the direct or indirect role of IFNα-mediated PD-1 expression on the decline of T cells during chronic viral infections . Our results indicate that during chronic HIV-1 infection , the virus elicits Type I IFN production which upregulates CD95 levels and elevates Bak expression , thereby increasing the susceptibility of T cells to CD95/Fas apoptosis . This aberrant apoptosis induction ultimately results in CD4+ T cell loss and impairment of HIV-specific CD8+ T cells . The potential deleterious effect of Type I IFN is further supported by SIV studies which show that non-pathogenic SIV-infections exhibit reduced IFNα compared to pathogenic SIV-infections [9] , [23] , [63] , and this may be associated with reduced T cell apoptosis [64] , [65] . Previous studies of pathogenic SIV infection have shown that plasma IFNα peaks and returns to undetectable levels early during acute infection [23] , [66] . We show here that SIV disease is associated with persistent systemic IFNα generation that can be detected during the acute and chronic stages of infection . Our results demonstrate that SIV-infected rhesus macaques with higher levels of peak IFNα exhibited a significantly lower peak viral burden during acute infection , suggesting a potential beneficial anti-viral effect of the Type I IFN response . However , at the chronic stage of SIV infection in these same animals , higher viremia resulted in elevated plasma IFNα , suggesting that the virus drives IFNα production and the beneficial effect of IFN is largely lost . Our interpretations of these correlations are strongly supported by two very recent studies showing that persistent Type I IFN production in chronic ( Clone 13 ) lymphocytic choriomeningitis virus ( LCMV ) infection is causatively associated with the loss of viral control and this effect is CD4+ T cell-dependent [67] , [68] . In contrast , Type I IFN signaling blockade during acute ( Armstrong ) LCMV infection results in severely impaired viral clearance [67] . Although these reports suggest that Type I IFN may have stage-specific opposing effects during viral infections , the impact of the immunoprotective and immunopathological effects of these cytokines during the acute and chronic stages of single viral infection over time is currently unknown [69] . Our investigation addresses this question in part for pathogenic SIV infection through examination of the longitudinal relationships between Type I IFN production , CD4+ T cell decline and viral load . The finding that IFNα in chronic SIV infection is detrimental to the host was suggested by the observation that higher plasma IFNα was associated with loss of CD4+ T cells in these monkeys . One explanation for the contrasting effect of IFNα during acute and chronic SIV infection may be the different levels of IFN that are produced . Thus , the beneficial versus the harmful effect of IFNα may be determined by cytokine concentration , with high concentrations suppressing virus replication and persistent low levels leading to T cell loss . This is supported by studies which demonstrate that administration of high dose IFNα or pharmacologic induction of a robust Type I IFN response enhances antiviral immunity [70] , [71] , while continuous low-dose IFNα treatment drives strong lymphopenia in chronically SIV-infected monkeys [70] . HIV-1 can clearly induce Type I IFN and this could explain the positive correlation between IFNα and viral load during chronic infection . An alternative explanation , in light of the observation that Type I IFN primes HIV-specific CD8+ T cells for apoptosis , is that increased IFN impairs both the function and survival of these cells , contributing to the loss of viral control . In conclusion , our data support a scenario in which the induction of IFNα/β during chronic HIV-1 infection exerts pathogenic instead of protective effects on host immunity by upregulating CD95 and Bak expression and sensitizing CD4+ T cells and CD8+ T cells to Fas-mediated apoptosis , but not to TRAIL or TNFα-induced death . Our SIV studies suggest that although during acute SIV infection , elevated Type I IFN levels may contribute to lower peak viremia , this is not the case during chronic infection where viral loads seem to drive elevated IFNα levels . Continued production of Type I IFN during chronic HIV-1 disease may compromise CD4+ T cell survival , in addition to impairing HIV-specific CD8+ T cells . Therefore , blocking the activity of Type I IFN or its production with TLR antagonists may be a useful strategy to inhibit CD4+ T cell loss and enhance HIV-specific immunity .
Peripheral blood was collected from individuals following Drexel University College of Medicine Institutional Review Board ( IRB ) approval and obtaining written informed consent . Rhesus macaques ( Macaca mulatta ) were housed at BIOQUAL , Inc . ( Rockville , MD ) , in accordance with the standards of the American Association for Accreditation of Laboratory Animal Care . The protocol was approved by the BIOQUAL's Institutional Animal care and Use Committee under OLAW Assurance Number A-3086-01 . Bioqual is IAAALAC accredited and procedures were carried out in accordance with the recommendations of the Weatherall report . All of the patients were HIV-1 positive for at least 1 year ( range 1–30 years ) ; median CD4 count was 432 cells/µl ( range 10–1 , 551 cells/µl ) ; median CD8 count was 980 cells/µl ( range 224–2 , 565 cells/µl ) ; median viral load was 325 RNA copies/ml blood ( range <20–978 , 190 copies/ml blood ) ; 66 patients were on antiretroviral therapy ( ART ) . HIV-1 viral loads ( HIV RNA copies/ml plasma ) were determined using COBAS AmpliPrep/COBAS TaqMan HIV-1 Test , v2 . 0 ( Roche Diagnostics , Indianapolis , IN ) with an assay range from 20–10 , 000 , 000 HIV RNA copies/ml . IFNα-induced gene expression studies were conducted with ART-untreated HIV-1-infected patients ( ART naïve ) . Control samples were obtained from HIV-1 seronegative age-matched healthy individuals . All assays were performed on freshly isolated PBMC from HIV-1-infected and healthy individuals . PBMC were freshly isolated from heparinized venous blood by centrifugation over a Ficoll-Hypaque gradient ( Amersham Pharmacia Biotech , Uppsala , Sweden ) . HIV-specific , CMV-specific and EBV-specific CD8+ T cells were detected using tetramers of HLA class I A*0201 loaded with either HIV-Gag p17 77–85 ( SLYNTVATL ) , HIV-Pol 476–484 ( ILKEPVHGV ) , CMV p65 495–503 ( NLVPMVATV ) or EBV 280–288 ( GLCTLVAML ) peptide . To analyze the apoptosis sensitivity of PBMC , cells were stained with Annexin V Cy5 . 5/anti-CD8 PE-Texas Red ( Caltag , Burlingame , CA ) /anti-CD4 FITC/anti-CD3 Pacific Blue/HIV- or CMV-specific tetramer APC . Annexin V and all other antibodies were purchased from BD Biosciences ( San Diego , CA ) . Briefly , 106 cells were stained with tetramers and antibodies in FACS wash ( HBSS ( Cellgro , Herndon , VA ) , 3% horse serum ( Life Technologies , Carlsbad , CA ) , 0 . 02% NaN3 ) for 30 minutes on ice; washed with FACS wash and fixed with 1% paraformaldehyde . When Annexin V staining was performed , 2 . 5 mM CaCl2 was included in all steps . Intracellular levels of pro- and anti-apoptotic proteins were measured directly ex vivo in PBMC or following a 3 day incubation in the presence or absence of 1000 U/ml IFNα/β ± anti-CD3 antibody ( plates coated with 0 . 1 µg/ml ) . Following Live/Dead ( Invitrogen ) and surface staining with anti-CD8 PE-Texas Red/anti-CD4 FITC/anti-CD3 Pacific Blue , cells were permeabilized with cytotofix/cytoperm buffer ( BD Biosciences ) and intracellular staining was performed for 1 hour with fluorochrome conjugated antibodies ( anti-Bcl-2 and anti-Bcl-xL antibodies , BD Biosciences ) , or specific primary antibodies ( anti-FADD , Biovision , Mountain View , CA; anti-Bid , Epitomics , Burlingame , CA; anti-Bim , Millipore , Billerica , MA; anti-Bax , Abgent , San Diego , CA; anti-Bak , Epitomics ) and appropriate isotype controls , followed by a 1 hour incubation with a secondary antibody ( PE-conjugated anti-rat IgG or anti-rabbit IgG , Southern Biotechnologies ) . Samples were collected on a FACSAria ( BD Biosciences ) and analyzed using FlowJo software ( Treestar , San Carlos , CA ) . Mean fluorescence intensity ( MFI ) of intracellular protein expression as expressed as delta ( Δ ) MFI indicates: MFI of specific antibody – MFI of isotype control . To determine the sensitivity of T cells to apoptosis stimuli , PBMC ( 106 cells/ml ) from HIV-1-infected individuals or healthy donors were stimulated in complete RPMI ( RPMI 1640/10% heat-inactivated fetal bovine serum/2 mM L-glutamine/100 U/ml penicillin/100 µg/ml streptomycin sulfate , Cellgro , Manassas , VA ) at 37°C in 5% CO2 , in the presence or absence of plate-bound anti-CD95 monoclonal antibody ( plates coated with 5 µg/ml , CH11 , Millipore ) , soluble SuperKiller ( cross-linked ) TRAIL ( 10 and 100 ng/ml , Alexis Biochemicals , Lausen , Switzerland ) or TNFα ( 10 and 100 ng/ml , R & D Systems , Inc . , Minneapolis , MN ) for 14 hours , harvested and stained . For IFNα/β apoptosis sensitization studies , PBMC from healthy and HIV-1-infected individuals were first stimulated with human IFNα or IFNβ ( 1000 U/ml; PBL , Piscataway , NJ ) for 3 days in the presence or absence of plate-bound anti-CD3 antibody ( plates coated with 0 . 1 µg/ml OKT3 ) at 37°C in 5% CO2 , before apoptosis sensitivity was evaluated by re-stimulating the cells with anti-CD95 monoclonal antibody ( plates coated with 5 µg/ml ) , anti-CD3 monoclonal antibody ( plates coated with 5 µg/ml ) , TRAIL ( 10 ng/ml ) or TNFα ( 10 ng/ml ) . Plates were coated with antibodies as previously described [5] . Cells were then stained and fixed for flow cytometry . Specific apoptosis was calculated using the following formula: [ ( percentage of induced apoptosis - percentage of spontaneous apoptosis ) / ( 100 - percentage of spontaneous apoptosis ) ]×100 . PBMC from chronically HIV-1-infected patients were isolated by Ficoll-Hypaque gradient centrifugation . Knockdown of Bak expression was achieved through siRNA transfection by electroporation using a Gene Pulser XCell ( BioRad ) . PBMC ( 7 . 5–8×106 cells ) were washed and resuspended in 300 µl of Opti-MEM ( Life Technologies ) in a 2-mm cuvette and pulsed ( square wave , 500 V , 1 ms ) with 1 nmol of siRNA ( ON-TARGET Non-Targeting pool or Bak ON-TARGETplus SMARTpool , Dharmacon ) . The siRNA pools consist of a mixture of 4 siRNAs that are pre-designed to reduce off-target effects by up to 90% [72] . Knockdown efficiency in HIV-1 patient PBMC following siRNA transfection was determined by real-time quantitative PCR with Taqman gene expression assays ( Applied Biosystems , Foster City , CA ) for Bak ( assay Hs00832876_g1 ) , Bax ( assay Hs00180269_m1 ) which served as a negative control for knock-down specificity , and 18S rRNA ( assay Hs03928985_g1 ) which served as a loading and normalization control . Transfection efficiency was assessed with siGlo fluorescent oligonucleotides ( Dharmacon ) . Following a 72 hour incubation period , cells were left unstimulated or stimulated with plate-bound anti-CD95 monoclonal antibody for 14 hours and assayed for apoptosis as described above . CD95/Fas-induced specific apoptosis was calculated by subtracting the level of spontaneous cell death and apoptosis caused by electroporation alone . A high-titer virus stock was produced following a 3 day cell-associated infection of PM1 cells with HIV-1Ba-L . The infectious cell-free virus present in the culture supernatant was treated with 250 µM aldrithiol-2 ( AT-2; Sigma Aldrich , St . Louis , MO ) for 1 hour at 37°C [73] , concentrated and the AT-2 HIV-1 stock was tested for infection . CD4+ T cells were isolated from whole blood by negative selection using the RosetteSep system ( STEMCELL Technologies Inc . , Vancouver , BC ) . Purity was >96% , as determined by flow cytometry . Cells were cultured in complete RPMI at a density of 106 cells/ml/well and pre-treated for 1 hour with 5 µM of a TLR7/9-specific antagonist ( 13mer phosphorothioate deoxyribose inhibitor that we have previously characterized [74] ) ; Invitrogen Corporation , Carlsbad , CA ) , 10 µg/ml of an anti-IFNα/β receptor blocking antibody ( Millipore ) , or an isotype control ( eBioscience ) ; 7×104–105 TCID50/ml of infectious HIV-1Ba-L or an equivalent p24 amount of AT-2 HIV-1 was then added for 3 days before the cells were stimulated with anti-CD95/Fas-antibodies for 14 hours . Percentages of apoptotic cells were determined by flow cytometry as described above . IFNα present in supernatants was measured at 24 hours following virus exposure using the Human Interferon Alpha Multi-Subtype ELISA Kit ( PBL ) . Total RNA was extracted from the freshly isolated PBMC of HIV-1-infected individuals and healthy donors . The RNA was reverse transcribed into complementary DNA which was then subjected to real-time PCR using gene-specific primers for IFI6-16 ( 5′-cctgctgctcttcacttgca-3′ and 5′-ccgacggccatgaaggt-3 ) and ISG56 ( 5′-ctggactggcaatagcaagct-3′ and 5′-gagggtcaatggcgttctga-3′ ) as described previously [26] , [75] . IFNα-stimulated gene expression levels were normalized to β-actin controls and results were calculated for each HIV-1-infected patient as a fold change in gene expression , relative to healthy donors using the delta-delta Ct method of analysis . Macaques were inoculated intravenously or intrarectally with 100 ID50 of SIVmac251 and plasma IFNα levels were measured by Luminex as previously described [76] with monoclonals MMHA-11 and MMHA-2 ( PBL ) used as capture and detector antibodies , respectively . Viral load was determined by TaqMan RNA RT-PCR assay with a sensitivity of 200 SIV RNA copies/ml of blood ( Applied Biosystems ) . Absolute CD4+ T cell numbers were determined by flow cytometry . Statistical analysis was performed using the Shapiro-Wilk W test for normality , Student's t-test and nonparametric Wilcoxon signed-rank test for paired and unpaired samples . Parametric ( Pearson's r ) and nonparametric ( Spearman's rho ) statistics were used for measurements of correlation . Analyses were performed with the JMP program ( SAS , Cary , NC ) . P values<0 . 05 were considered to be statistically significant . | Type I interferons ( IFNα/β ) are innate immune mediators that are produced by cells in response to viral infections . Although the protective effects of IFNα/β are well-established , it is not clear whether these cytokines are beneficial or deleterious during HIV-1 infection . We report that HIV-1 infection renders T cells more prone to undergo programmed death and that this enhanced apoptosis susceptibility is associated with abnormal expression of pro- and anti-apoptotic molecules . Importantly , IFNα/β increases the expression level of the pro-apoptotic protein , Bak , a major gatekeeper of the mitochondrial apoptosis machinery . Exposure of healthy donor T cells to IFNα/β increased Bak expression and induced an apoptosis sensitivity that is similar to what is observed in HIV-1-infected patient T cells . Furthermore , elevated IFNα production and Bak expression correlated with heightened T cell apoptosis , low CD4+ T cell numbers and high viral loads in patients . In a primate model of HIV-1 infection , we observed that increased IFNα was associated with low peak viral loads in early infection , but high viremia and decreased CD4+ T cell counts during chronic infection . These studies identify a novel mechanism by which Type I IFN may induce immune dysfunction in HIV-1 infection . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Type I Interferon Upregulates Bak and Contributes to T Cell Loss during Human Immunodeficiency Virus (HIV) Infection |
Members of the Rhizobiales ( class of α-proteobacteria ) display zonal peptidoglycan cell wall growth at one cell pole , contrasting with the dispersed mode of cell wall growth along the sidewalls of many other rod-shaped bacteria . Here we show that the seven-transmembrane receptor ( 7TMR ) protein RgsP ( SMc00074 ) , together with the putative membrane-anchored peptidoglycan metallopeptidase RgsM ( SMc02432 ) , have key roles in unipolar peptidoglycan formation during growth and at mid-cell during cell division in Sinorhizobium meliloti . RgsP is composed of a periplasmic globular 7TMR-DISMED2 domain , a membrane-spanning region , and cytoplasmic PAS , GGDEF and EAL domains . The EAL domain confers phosphodiesterase activity towards the second messenger cyclic di-GMP , a key regulatory player in the transition between bacterial lifestyles . RgsP and RgsM localize to sites of zonal cell wall synthesis at the new cell pole and cell divison site , suggesting a role in cell wall biogenesis . The two proteins are essential for cell wall biogenesis and cell growth . Cells depleted of RgsP or RgsM had an altered muropeptide composition and RgsM binds to peptidoglycan . RgsP and RgsM orthologs are functional when interchanged between α-rhizobial species pointing to a conserved mechanism for cell wall biogenesis/remodeling within the Rhizobiales . Overall , our findings suggest that RgsP and RgsM contribute to the regulation of unipolar cell wall biogenesis in α-rhizobia .
Rod-shaped bacteria have evolved diverse modes of cell growth . In Bacillus subtilis , Escherichia coli and Caulobacter crescentus , cells grow by elongating along the lateral sidewall , incorporating peptidoglycan ( PG ) cell wall material in a dispersed pattern along the sidewall [1] . Other bacteria elongate by zonal growth in which PG is incorporated at one or both cell poles [2] . Unipolar growth is observed in α-proteobacterial Rhizobiales , such as Agrobacterium tumefaciens and Sinorhizobium meliloti [3] , while Gram-positive Actinomycetales including Mycobacterium tuberculosis grow bipolarly [4] . Later in the cell cycle , all species switch PG synthesis completely or partially to mid-cell for cell division [1] . During PG synthesis nascent material is incorporated into the pre-existing PG structure ( sacculus ) , by the activity of hydrolases which selectively cleave bonds in the stress-bearing sacculus [5] . Because the integrity of the PG sacculus is essential for maintaining cell shape and resisting turgor [6] , cell growth and PG synthesis require precise spatial and temporal regulation of the incorporation of new material into the PG sacculus [7] . Rod-shaped bacteria with a dispersed mode of PG incorporation along the lateral cell wall utilize the cytoplasmic actin-like MreB protein to direct PG synthesis . Dynamic filaments or patches of MreB are believed to serve as platforms for the intracellular and extracellular PG synthesis machineries [8 , 9] . By contrast , most rod-shaped bacteria with polar growth do not contain MreB homologs [10] and it is currently unknown how polar cell wall elongation is regulated in these bacteria . Bis- ( 3′-5′ ) -cyclic dimeric guanosine monophosphate ( c-di-GMP ) has a central role in the regulation of motility , adherence , biofilm formation and virulence and in several bacterial species , also for promoting cell cycle progression , growth and development [11–15] . c-di-GMP is synthesized by diguanylate cyclases ( DGCs ) with a conserved GGDEF domain and degraded by phosphodiesterases ( PDEs ) with either an EAL domain or a HD-GYP domain [16] . In the α-proteobacterium C . crescentus , the spatial organization of proteins involved in c-di-GMP metabolism contributes to cell polarity and cell cycle progression [14] , and we showed previously that strong overproduction of this second messenger inhibited growth and resulted in cell filamentation in S . meliloti [17] . Seven-transmembrane receptors ( 7TMRs ) form the largest , most ubiquitous and most versatile family of membrane receptors . In eukaryotes , they are involved in signaling via interaction with cytoplasmic G-proteins [18] . A distinct class of bacterial 7TMRs consists of so-called 7TMR-DISMs , which stands for 7TMR with diverse intracellular signaling modules [19] . 7TMR-DISM proteins contain seven transmembrane α-helices ( 7TMR-DISM_7TM domain ) fused to various cytoplasmic signaling and/or extracytoplasmic 7TMR-DISMED1 or 7TMR-DISMED2 domains [19] . These extracytoplasmic domains have been predicted to bind ligands such as carbohydrates [19] . Common cytoplasmic signaling modules in 7TMR-DISM proteins include histidine protein kinase domains , GGDEF and EAL domains involved in c-di-GMP homeostasis and sensory Per-Arnt-Sim ( PAS ) domains . PAS domains are known to sense small molecules , ions , gases , light or redox state [20] . With up to 14 paralogs per genome , for example in the spirochete Leptospira interrogans , 7TMR-DISMs are widely distributed in both Gram-negative and Gram-positive bacteria [19] . However , only a few of these proteins have been functionally characterized . These include the Pseudomonas aeruginosa 7TMR-DISM histidine kinases RetS and LadS , which are involved in regulation of biofilm formation , and the GGDEF domain containing protein NicD , which is involved in biofilm dispersal [21 , 22] . The 7TMR-DISM protein SMc00074 ( renamed RgsP for rhizobial growth and septation c-di-GMP phosphodiesterase ) was previously shown to be essential in S . meliloti [17 , 23 , 24] . Here , we provide evidence that RgsP is important for PG synthesis in certain α-rhizobial species . We identify SMc02432 ( a putative membrane-anchored periplasmic PG metallopeptidase , renamed RgsM for rhizobial growth and septation metallopeptidase ) as a RgsP interaction partner and show that both are required for unipolar cell growth in S . meliloti and related Rhizobiales . RgsP was also found to be an active PDE , substantially contributing to c-di-GMP homeostasis and hence possibly connecting c-di-GMP signaling with spatiotemporal control of PG synthesis .
RgsP is composed of 7TMR-DISMED2 , 7TMR-DISM_7TM , PAS , GGDEF and EAL domains . Previous systematic mutagenesis of c-di-GMP-related genes identified rgsP ( SMc00074 ) as a potentially essential gene in S . meliloti Rm2011 [17] . C-terminally 3×FLAG-tagged RgsP accumulated in growing cells and was only detected at very low levels in stationary phase cells ( S1A Fig ) , suggesting a role during cell growth . To further study the function of this protein , we constructed a RgsP depletion strain ( Rm2011 rgsPdpl ) by placing the native rgsP gene under the control of an IPTG-inducible promoter . Using the same promoter , we confirmed depletion of a C-terminally 3×FLAG-tagged RgsP variant in the absence of IPTG ( S1B Fig ) . Growth of Rm2011 rgsPdpl was strongly dependent on IPTG ( Fig 1A ) . Cells cultured in the presence of IPTG showed wild type-like growth and morphology , whereas RgsP-depleted cells lost the rod shape and had an irregular , uneven cell surface ( Fig 1B and 1C; S2 and S3A Figs ) . 4 . 7% of these cells were stained by the dead cell indicator propidium iodide ( S4A Fig ) . This suggests that the physiological effect of RgsP depletion probably is mostly bacteriostatic after 24 h of incubation in absence of IPTG . To estimate the capability of RgsP-depleted cells to resume growth , we analyzed colony formation and microscopically monitored growth on TY medium containing IPTG . In cultures of Rm2011 rgsPdpl grown without added IPTG for 12 or 24 h , the number of colony forming units ( CFU ml-1 OD600-1 ) was reduced by 67 . 8% or 97 . 4% , respectively , compared to IPTG-supplemented cultures ( S4B Fig ) . Following the fate of RgsP-depleted cells ( previously grown in the absence of IPTG for 24 h ) , 25 . 8% were able to recover and to give rise to microcolonies on TY agarose pads with IPTG . In contrast , only 3 . 0% of the cells were able to divide at least once during the 12 h observation period on pads lacking IPTG ( S4C Fig ) . Thus , RgsP depletion predominantly resulted in growth arrest , which was relieved upon IPTG addition in a minor fraction of the cells only . However , when interpreting these results , it has to be taken into account that the cells grown under depletion conditions may contain some residual amounts of RgsP . Zones of PG synthesis were visualized by HADA ( 7-hydroxycoumarin-3-carboxylic acid-D-alanine ) -labeling [25] . In IPTG-supplemented cultures with induced rgsP expression , 98% of the cells were stained at one of the cell poles and/or septal region . By contrast , only 33% of the RgsP-depleted cells incorporated HADA ( Fig 1D; S5A Fig ) . Moreover , upon RgsP depletion , the proportion of pre-divisional cell doublets with visible septum constriction increased to 42% compared to 13% in RgsP-replete cultures ( Fig 1D; S5B Fig ) . These results suggest that polar cell wall synthesis and late stages of cell division were impaired in the absence of RgsP . To determine the subcellular localization of RgsP , we replaced rgsP with rgsP-egfp at its native genomic location . In exponentially growing S . meliloti cells , RgsP-EGFP co-localized with HADA-labeled sites of PG synthesis at one of the cell poles and the division site ( Fig 2A and 2B ) . Time-lapse microscopy showed that RgsP-EGFP remained at the pole during the entire phase of cell elongation until it relocated to the mid-cell region ( Fig 2C ) . We next localized RgsP-EGFP simultaneously with ParB-mCherry . ParB binds to the parS sites close to the chromosomal origin of replication [26] . Early in the cell cycle , a single fluorescent ParB-mCherry focus localized at the old cell pole and RgsP-EGFP localized at the opposite , new pole ( Fig 2C ) . Later , a second ParB-mCherry focus moved to the new pole . Following a period of co-localization ( ~90 minutes ) of RgsP-EGFP and ParB-mCherry at the new pole , the polar RgsP-EGFP signal disappeared and appeared at mid-cell . Thus , RgsP-EGFP localizes at the new cell pole early in the cell cycle and later at the cell division site . Importantly , these are the sites of PG synthesis . To relate the temporal pattern of RgsP relocation to the mid-cell to known pre-divisional processes , we co-localized RgsP with the early divisome component FtsZ [7] . mCherry-FtsZ showed diffuse fluorescence in growing cells , and localized to the mid-cell in pre-divisional cells ( Fig 2A ) . Septal localization of RgsP-EGFP always coincided with the presence of a mCherry-FtsZ focus at mid-cell , but not vice versa ( Fig 2A and 2B ) , suggesting that RgsP accumulates at the septal site later than FtsZ . Time-lapse microscopy showed localization of RgsP-EGFP at mid-cell about 24 minutes after occurrence of the mCherry-FtsZ focus . This RgsP-EGFP relocalization was immediately followed by the onset of septum constriction , which was completed by cell division about 24 minutes later ( Fig 2D ) . This implies that RgsP may also have a function in septation . The dynamics of RgsP relocation to the mid-cell region was analyzed at a higher time resolution relative to PleD . We found previously that the DGC PleD localizes to the new cell pole within 20 minutes before cell division [17] . Simultaneous tracking of PleD-EGFP and RgsP-mCherry showed that the accumulation of PleD-EGFP at the growing cell pole temporally correlated with the RgsP-mCherry signal fading away at the pole and relocating to the division site ( S6 Fig ) , indicating mutually exclusive localization of RgsP and PleD at the new pole . To dissect the functionality of RgsP relative to its complex domain organization , we generated RgsP variants , each with or without C-terminally fused EGFP ( Fig 3A ) . The gene variants were ectopically expressed either from the native promoter P*rgsP ( S7A Fig ) on the single-copy plasmid pABCS2-mob ( native level of 3×FLAG-tagged RgsPwt ( S1C Fig ) ) or from a constitutive synthetic promoter ( Psyn ) on the low-copy plasmid pR_egfp ( elevated level of 3×FLAG-tagged RgsPwt ( S1D Fig ) ) in the RgsP-depleted strain Rm2011 rgsPdpl . All RgsP variants were assayed for their ability to complement the cell growth and morphology defects of the RgsP-depleted strain , and for subcellular localization of EGFP fusion proteins . Ectopically expressed rgsPwt or rgsPwt-egfp restored growth and cell morphology of RgsP-depleted cells ( Fig 3B and 3C ) , and RgsPwt-EGFP localized similarly when expressed from the ectopic and native gene locus ( Figs 2 and 3D ) . Point mutations targeting the conserved inhibitory site ( I site ) and the DGC active site motifs in the GGDEF domain ( RxxD to AxxA and GGDQF to GAAAF ) and the PDE active site in the EAL domain ( EAL to AAL ) , or removal of the EAL domain did not significantly affect protein localization or complementation of the cell growth and morphology defects ( Fig 3B , 3C and 3D ) . RgsP and RgsP-EGFP variants lacking the GGDEF or both GGDEF and EAL domains were unable to fully complement growth and morphology defects unless they were expressed at higher levels from Psyn ( Fig 3B and 3C ) . Nevertheless , the EGFP-tagged versions showed normal cellular localization ( Fig 3D ) . When gene expression was driven by P*rgsP on pABCS2-mob , levels of corresponding 3×FLAG-tagged wild type protein and variants were similar , except for RgsPΔGGDEF , which was detected at 67% of the corresponding wild type protein level ( S1C Fig ) . This implies that RgsP lacking the EAL domain fully supports cell growth , whereas lack of the GGDEF domain impairs protein stability or function . To investigate the possible role of c-di-GMP in RgsP function , we tested complementation of RgsP depletion by rgsP and rgsP-egfp variants in the c-di-GMP-deficient strain Rm2011 ΔXVI , which lacks all genes predicted to encode active DGCs and in which the level of c-di-GMP was below the detection limit [17] . The complementation properties of each of these RgsP versions were indistinguishable in strains with or without c-di-GMP , and RgsP-EGFP showed normal localization in Rm2011 ΔXVI ( S8 Fig ) . This suggests that c-di-GMP is not required for RgsP localization and the essential function of this protein in cell growth . Deletion of the N-terminal domains 7TMR-DISMED2 , 7TMR-DISM_7TM or PAS abolished the ability of RgsP to complement RgsP depletion phenotypes , irrespective of the expression vector ( Fig 3B and 3C ) . When gene expression was driven by P*rgsP on pABCS2-mob ( or Psyn on pR_egfp ) , 3×FLAG-tagged RgsPΔPAS , RgsPΔPASΔGGDEFΔEAL , RgsPΔ7TMR-DISMED2 , and RgsPΔ7TMR-DISMED2Δ7TMR-DISM_7TM variants accumulated to 58% ( 736% ) , 30% ( 592% ) , 6% ( 236% ) and 9% ( 19% ) , respectively , of the corresponding wild type protein levels ( S1C and S1D Fig ) . Furthermore , the fluorescence signal of these RgsP variants , tagged with EGFP , did not show the characteristic polar and septal localization ( Fig 3D ) . Taken together , these data suggest that the N-terminal part of RgsP determines the essentiality of this protein , whereas the EAL domain is dispensable for the growth-promoting function of RgsP . Since the GGDEF and EAL domains may have a regulatory role , we analyzed the enzymatic activities and ability to bind c-di-GMP in vitro . A purified His6-tagged variant of RgsP containing the PAS , GGDEF and EAL domains ( His6-RgsPPAS-GGDEF-EAL ) hydrolyzed [α-32P]-c-di-GMP , whereas no cleavage product was detected with the PDE active site mutant variant His6-RgsPPAS-GGDEF-EAL-AAL ( S9A Fig ) . In a DGC activity assay with [α-32P]-GTP as a substrate , His6-RgsPPAS-GGDEF-EAL-AAL did not synthesize c-di-GMP , in contrast to C . crescentus DgcA used as a positive control ( S9B Fig ) . This is in agreement with the degenerate active site GGDQF in the RgsP GGDEF domain and our previous in vitro DGC activity assay with a RgsP fragment comprising the PAS , GGDEF and EAL ( active site intact ) domains [17] . Since an intact I site RxxD is present in the GGDEF domain of RgsP , we assayed for the ability of RgsP to bind c-di-GMP in a differential radial capillary action of ligand assay ( DRaCALA ) using a His6-tagged RgsP variant containing only the PAS and GGDEF domains ( His6-RgsPPAS-GGDEF ) . In this assay , the positive control His6-DmxB from Myxococcus xanthus produced the characteristic DRaCALA pattern , whereas His6-RgsPPAS-GGDEF was not able to prevent the diffusion of [α-32P]-c-di-GMP ( S9C Fig ) . Thus , RgsPPAS-GGDEF-EAL has c-di-GMP PDE activity , presumably catalysed by the EAL domain , but the GGDEF domain does not bind or synthesize c-di-GMP . To evaluate c-di-GMP PDE activity of RgsP in vivo , we determined the c-di-GMP content of RgsP-depleted Rm2011 rgsPdpl complemented with rgsPwt , rgsPAAL and rgsPGAAAF expressed from P*rgsP . The c-di-GMP content of cells expressing rgsPwt and rgsPGAAAF was similar , whereas expression of the PDE active site mutant variant rgsPAAL resulted in a two-fold increase in c-di-GMP content ( S9D Fig ) . This data is consistent with the in vitro data and provides evidence that RgsP substantially contributes to c-di-GMP degradation in vivo . To identify protein interaction partners of RgsP , we performed co-immunoprecipitation ( Co-IP ) experiments with a C-terminally 3×FLAG-tagged variant of RgsP , encoded by rgsP-3×flag replacing the native rgsP at its chromosomal location . 27 RgsP interaction partner candidates were identified ( S1 Table ) . Among these , the hypothetical transmembrane protein RgsM ( SMc02432 ) was most abundant and identified with the highest number of unique peptides . Co-IP with RgsM-3×FLAG resulted in identification of RgsP , further supporting an interaction between the two proteins ( S2 Table ) . Analysis of RgsM amino acid sequence with transmembrane topology and signal peptide prediction tool PHOBIUS [27] suggested cytoplasmic localization of the amino acids 1–31 , a short hydrophobic transmembrane α-helical region ( amino acids 32–57 ) and periplasmic localization of the remaining C-terminal portion ( Fig 4A ) . Amino acids 508–606 of RgsM represent a conserved peptidase M23 domain ( pfam01551 ) often referred to as a LytM domain . This domain is predicted to have PG endopeptidase activity and is characteristic for zinc-dependent metallopeptidases . The LytM domain of RgsM contains a conserved HxxxD motif ( S3 Table ) , which is required for zinc ion coordination and hydrolysis of glycine-glycine bonds in the peptides of staphylococcal PG by Staphylococcus aureus LytM [28–30] . The remaining RgsM amino acid sequence did not provide any hint about its possible function . To verify the predicted membrane topology of RgsM , we fused this protein to a truncated E . coli alkaline phosphatase PhoA , which is only active in the periplasm , and is missing its own signal peptide . The phosphatase detection assay revealed that PhoA , following RgsM1-66 , indeed localized to the periplasm in both S . meliloti and E . coli . This strongly suggests the periplasmic localization of RgsM60-646 ( S10 Fig ) . Interaction between RgsP and RgsM was verified by a bacterial two-hybrid assay [31] . In this assay , putative interaction partners , fused to Bordetella pertussis adenylate cyclase fragments T18 and T25 , are produced in an E . coli adenylate cyclase-deficient strain . Interaction between the two fusion proteins results in reconstitution of a functional enzyme , which activates expression of the lacZ reporter gene . Simultaneous production of T18-RgsM and RgsP∆GGDEF∆EAL-T25 fusion proteins , as well as of T18-RgsM and T25-RgsM , resulted in increased β-galactosidase activity , indicating protein-protein interactions ( Fig 4B ) . Thus , RgsM was able to interact with RgsP and to homodimerize in a heterologous host . This is consistent with the Co-IP data that suggested RgsP-RgsM interaction in S . meliloti . Attempts to generate a rgsM knockout mutant failed , suggesting that rgsM is essential . Similar to the growth phase-dependent accumulation of RgsP , C-terminally 3×FLAG-tagged RgsM was only detected in growing cells but not in stationary phase cells ( S1A Fig ) . Next , we constructed a RgsM depletion strain Rm2011 rgsMdpl by placing the chromosomal rgsM gene under the control of an IPTG-inducible promoter . Depletion of a C-terminally 3×FLAG-tagged RgsM variant was confirmed using the same genetic setup ( S1B Fig ) . The growth of Rm2011 rgsMdpl was significantly impaired in the absence of IPTG ( Fig 5A ) . Strikingly similar to cells depleted of RgsP , RgsM-depleted cells lost the wild type rod shape ( Figs 5B and 1B; S2 Fig ) . Electron microscopy revealed regions of low electron density in RgsM-depleted cells ( Fig 5C; S3B Fig ) . 7 . 1% of these cells were stained by propidium iodide , indicating both lethal and bacteriostatic effects of RgsM depletion , similar to RgsP depletion ( S4A Fig ) . Following depletion of RgsM for 12 or 24 h , the number of colony forming units was reduced by 89 . 4% or 99 . 5% , respectively , relative to cultures pre-incubated under non-depletion conditions ( S4B Fig ) . We microscopically monitored growth of cells , previously grown under depletion conditions for 24 h . 10 . 9% of the previously RgsM-depleted cells gave rise to microcolonies on medium supplemented with IPTG , whereas 99 . 1% of these cells did not divide on medium lacking IPTG during the 12 h observation period ( S4C Fig ) . Labeling of RgsM-depleted cells with HADA revealed incorporation of new PG in only a minor part of pre-divisional cells at the septal region , whereas Rm2011 rgsMdpl grown in IPTG-containing medium displayed a wild type-like HADA staining ( Fig 5D; S5A Fig ) . Similar to RgsP depletion , RgsM depletion resulted in an increase in the proportion of pre-divisional cell doublets ( with visible septum constriction ) to 40% ( S5B Fig ) . Thus , like RgsP , RgsM is required for cell growth and division and PG biosynthesis . To analyze RgsM localization relative to RgsP , we constructed a strain with rgsP-mCherry and mVenus-rgsM replacing rgsP and rgsM at their native genomic locations . mVenus-RgsM accumulated at one cell pole or at mid-cell and showed co-localization with RgsP-mCherry throughout the cell cycle ( Fig 6A and 6B ) . This indicates that RgsM , like RgsP , localizes to sites of zonal cell wall synthesis . Localization of mVenus-RgsM was dependent on RgsP , since RgsP depletion resulted in diffuse mVenus-RgsM fluorescence ( Fig 6C ) . Likewise , in RgsM-depleted cells , only diffuse RgsP-mCherry signal was observed ( Fig 6D ) , indicating that polar and septal localization of RgsP and RgsM is mutually dependent . Taken together , these data imply a functional relation between RgsP and RgsM . To further characterize the role of RgsM , we analyzed effects of rgsM overexpression . When grown in TY medium , RgsM-overproducing cells were indistinguishable from the empty vector control ( S11 Fig ) . Contrastingly , in LB medium they grew very poorly and appeared enlarged and spherical ( Fig 7A and 7B ) , with dramatically enlarged periplasm and inner membrane invaginations ( Fig 7C; S3C Fig ) . Overexpression of rgsM in LB impaired PG biosynthesis , as judged from a very weak dispersed HADA staining , only visible after adjusting HADA fluorescence signal intensity , in contrast to cells carrying an empty vector , which showed polar and septal HADA signals ( Fig 7D; S5A and S11C Figs ) . Noteworthy , even in the absence of RgsM overexpression , the rod cell shape differed slightly in TY and LB , with broader and shorter cells grown in LB ( S11B Fig ) . Since the most prominent difference between the composition of TY and LB media is the content of NaCl ( 86 mM in LB , none in TY ) and CaCl2 ( 2 . 7 mM in TY , none in LB ) , we analyzed the effects of these salts on the growth of RgsM-overproducing cells . Increasing the NaCl concentration to 300 mM in LB alleviated the rgsM overexpression-associated morphology and growth defects , and addition of CaCl2 to 2 . 7 mM resulted in wild type-like growth ( S12 Fig ) . Replacement of Zn2+ with Ca2+ in RgsM may inactivate the protein and thereby mitigate the effect of rgsM overexpression . However , this explanation for the effect of CaCl2 is unlikely since the presence of Ca2+ in the growth medium did not negatively affect growth of S . meliloti , as RgsM depeletion did . Thus , the growth defect resulting from rgsM overexpression in LB may be a combined effect of an artificial increase in RgsM abundance and outer membrane destabilization in the absence of calcium [32] . This was probably counteracted by elevated medium osmolarity , which is supposed to reduce turgor . To analyze the putative RgsM metallopeptidase active site H510xxxD , we constructed the RgsMH510A variant . The H510A mutation mitigated the strong cell morphology defect caused by RgsM overproduction in LB . Notably , both in TY and LB , growth of the rgsMH510A overexpressing cells was inhibited and morphology was altered similar to RgsM-depleted cells ( S11 Fig ) , suggesting a possible dominant negative effect of RgsMH510A overproduction on the native RgsM function . Accumulation of overproduced RgsMwt and RgsMH510A in Rm2011 was confirmed by detecting the corresponding 3×FLAG-tagged variants ( S1E Fig ) . Overexpression of rgsM , but not rgsMH510A , compromised the cell envelope of E . coli and resulted in cell lysis in LB lacking NaCl . This was detected using the β-galactosidase substrate chlorophenol red-β-D-galactopyranoside ( CPRG ) as an indicator of cell lysis and increased membrane permeability [33] and by microscopy ( Fig 7E; S13A Fig ) . As in S . meliloti , overproduction of wild type RgsM and RgsMH510A inhibited growth of E . coli in LB medium ( S13B Fig ) . Both corresponding 3×FLAG-tagged RgsM variants accumulated in E . coli ( S1E Fig ) . To investigate the role of RgsP and RgsM in cell wall biogenesis , we determined the muropeptide composition of PG isolated from S . meliloti Rm2011 depleted and non-depleted of RgsP or RgsM . Muropeptide profiles of the depletion strains grown in TY under non-depletion conditions were similar to those obtained for the Rm2011 wild type ( S14A Fig ) . Depletion of RgsP or RgsM resulted in alterations in the relative abundance of specific muropeptides . In both strains , the uncross-linked pentapeptides and the 3-3-cross-linked TetraTri dimers accumulated , whereas the 4-3-cross-linked TetraTetra dimers and TetraTetraTetra trimers were less abundant than in non-depleted cells ( Fig 8A; S14 and S15 Figs; S4 Table ) . Rm2011 cells , overproducing RgsM in LB , accumulated more Penta and TetraPenta muropeptides compared to cells harboring the empty vector control , whereas overproduction of RgsMH510A resulted in similar but less pronounced changes in the muropeptide profiles ( S16A Fig ) . The accumulation of pentapeptides upon either depletion or overproduction of RgsP and RgsM points to altered incorporation and/or processing of new material into the growing PG sacculus , and the increase in 3–3 cross-links suggests increased LD-transpeptidase activity . Both might be indirect effects in cells with impaired PG growth , in response to stress . Muropeptides of E . coli S17-1 overproducing RgsM or RgsMH510A and grown in LB remained unchanged suggesting that the phenotypic changes of E . coli cells were independent of a putative PG hydrolase activity of RgsM ( S16B Fig ) . To gain further mechanistic insights into RgsP and RgsM functions in S . meliloti , purified His6-tagged variants of both proteins were assayed for PG binding in vitro . Whereas His6-RgsM57-646 was recovered in the insoluble fraction after incubation with PG , His6-RgsP7TMR-DISMED2 mostly remained in the soluble fraction ( Fig 8B ) . This suggested that RgsM bound tightly to PG and that this was not the case for the periplasmic domain of RgsP ( RgsP7TMR-DISMED2 ) . PG hydrolase activity of His6-RgsM57-646 alone or in combination with His6-RgsP7TMR-DISMED2 was assayed with S . meliloti PG as substrate . No evident changes in the muropeptide profiles were observed , indicating that His6-RgsM57-646 did not hydrolyze PG in vitro ( S17A Fig ) . Since we observed a prominent proteolysis product of RgsM in Western blot analysis ( S1E Fig ) , we asked if this might represent the active form of the enzyme . Therefore , N-terminally truncated version of His6-RgsM ( amino acids 260–646 ) was analyzed . Although this fragment retained the ability to bind PG , it did not show PG hydrolase activity in vitro ( S17B and S17C Fig ) . It cannot be excluded that the assay conditions were not optimal or that an additional factor is required to activate RgsM . Overall , the effects of RgsM and RgsP depletion or overproduction on the muropeptide profiles and strong PG binding by RgsM further support involvement of both proteins in PG biogenesis . Comparative protein sequence analysis using BLASTP revealed conservation of RgsP and RgsM in Rhizobiales , Rhodobacterales , and in a single γ-proteobacterial species , whereas no homologs were detected in the remaining eubacterial phyla ( S3 Table ) . Thus , we asked whether the homologous proteins from other species were also functionally conserved . Rhizobium etli and A . tumefaciens rgsP homologs RHE_CH00976 ( rgsPRe ) and Atu0784 ( rgsPAt ) were translationally fused to egfp at their native genomic locations . These tagged proteins displayed polar and septal localization corresponding to the HADA-staining zones in R . etli and A . tumefaciens , similar to the labeled RgsP variant in S . meliloti ( S18A Fig; Fig 2 ) . We also tested the ability of A . tumefaciens and R . etli rgsP and rgsM homologs , ectopically expressed from a taurine-inducible promoter , to complement S . meliloti RgsP and RgsM depletion strains . Without promoter induction ( 'leaky expression' ) , rgsPRe , rgsPAt and rgsMRe fully complemented the growth and morphology defects of the respective S . meliloti depletion strains , similar to the S . meliloti homologs , whereas induction with taurine was required for complementation of RgsM depletion with rgsMAt ( S18B and S18C Fig ) . Similar subcellular localization of RgsP homologs in S . meliloti , R . etli and A . tumefaciens , and cross-complementation of rgsP and rgsM between these species provide evidence for functional conservation of both proteins in the Rhizobiales .
Members of the Rhizobiales , including S . meliloti , show unipolar cell growth [3] . However , the molecular mechanisms governing polar growth of the PG sacculus are largely unknown in these bacteria . In this study , we provide evidence suggesting important roles for RgsP ( a member of the seven-transmembrane receptor family 7TMR-DISM ) and its interaction partner RgsM ( a putative PG metallopeptidase ) in polar cell wall growth . Proteins associated with the polar PG growth zones in A . tumefaciens include the division scaffold proteins FtsZ and FtsA , PG synthase PBP1a and LD-transpeptidase Atu0845 [34 , 35] , which are directly or indirectly involved in PG biosynthesis . RgsP does not show homology to known cell division or PG biosynthesis proteins; however , it localizes to sites of zonal cell wall synthesis in S . meliloti , R . etli and A . tumefaciens . This feature is likely to be conserved in α-rhizobia containing a RgsP homolog . The phenotypes caused by RgsP depletion in S . meliloti − i . e . growth inhibition , altered cell shape , reduced incorporation of HADA and accumulation of penta-muropeptides in the PG sacculus − imply a regulatory role of RgsP in PG biosynthesis . Although known protein components of the cell elongation and division machineries have not been detected in the RgsP and RgsM pull-down assays , indirect , transient or low affinity interactions of RgsP or RgsM with such proteins may have escaped this analysis . In E . coli , pentapeptides are specific for newly synthesized PG and are quickly processed as PG matures [36] by trans- , endo- and carboxypeptidase reactions [7] . In agreement with a previous report [3] , we did not detect pronounced pentapeptide peaks in the S . meliloti wild type muropeptide profile . Perhaps in the wild type these pentapeptides were quickly processed but accumulated in RgsP-depleted cells due to impaired PG maturation . In E . coli , lack of the PG carboxypeptidase PBP5 results in an increase in the PG pentapeptide content; an effect which is further augmented by eliminating endopeptidases PBP4 and PBP7 [37] . Along with accumulation of monomeric pentapeptides , tetrapeptide dimer and trimer levels decreased upon RgsP depletion , indicating impaired PG cross-linking or enhanced hydrolysis of the peptide bridges . The cell shape can be influenced by the degree of PG cross-linking [38–44] . Hence , the impaired growth and altered cell shape of RgsP-depleted cells might have resulted from dysregulation of PG remodeling . Cell wall biogenesis by necessity involves incorporation of new PG material into the existing PG sacculus , and this requires local hydrolysis of peptide bridges by PG endopeptidases . These include proteins with a M23 metallopeptidase ( or LytM ) domain . In E . coli and Vibrio cholerae , single genetic knockdowns of LytM domain proteins were not lethal because of genetic redundancy [45 , 46] . By contrast , the LytM domain protein RgsM is essential in S . meliloti , which is in agreement with a previous report [24] . The A . tumefaciens RgsM and RgsP orthologues were also shown to be essential [47] . Although we were not able to detect RgsM PG hydrolase activity in vitro , several findings are in agreement with an enzymatic activity in vivo . The RgsM LytM domain contains conserved histidine residues , which have been found to be essential for PG hydrolase activity [28 , 30 , 48] . Moreover , rgsM overexpression in S . meliloti destabilized the cell envelope and caused perturbations in the muropeptide composition . However , expression of rgsM in E . coli resulted in cell lysis but did not cause changes in the muropeptide profile . Structural studies of S . aureus LytM and Neisseria meningitidis LytM domain protein NMB0315 suggested the full-length proteins to adopt conformations interfering with enzymatic function , implying activation by proteolysis or protein-protein interactions in vivo [28 , 49] . Thus , we speculate that in vivo , RgsM may hydrolyze PG once activated by a yet-unknown factor . Alternatively , RgsM may have a regulatory role similar to E . coli and C . crescentus LytM domain proteins , which activate amidases to cleave septal PG [50–54] . Some LytM domain proteins also contain PG-binding domains [53 , 55] . PG binding by RgsM , demonstrated in vitro , is in agreement with an enzymatic or regulatory role of this protein . Based on the mutual pull-down of RgsP and RgsM , an interaction in a bacterial two-hydrid assay and similar phenotypes of RgsM- and RgsP-depleted cells , we suggest that both proteins are involved in the same regulatory pathway . We narrowed down the essential part of RgsP to the N-terminal section including the 7TMR-DISM and PAS domains , which both may have a sensory function . In bacteria , 7TMR-DISMs are best understood in P . aeruginosa . In this bacterium , ligand binding to 7TMR-DISMED2 domains and their homodimerization were described as regulatory cues [21 , 22 , 56 , 57] . We speculate that RgsP either homodimerizes or , once triggered by an unknown cue , interacts with RgsM , to modulate RgsM dimerization and activity . This model is in agreement with the similar phenotypes of RgsP- and RgsM-depleted cells and the dominant effect of RgsMH510A overproduction in S . meliloti . This enzymatically inactive protein variant may compete with native RgsM for the interaction sites on RgsP or other interacting proteins . Although abundance of RgsP lacking the PAS domain was only moderately reduced , the localization and growth-promoting function of this protein was dramatically affected . A role of PAS domains or PAS-like motifs in polar protein localization was previously reported for example in C . crescentus . These include the C . crescentus single PAS domain protein MopJ , which directly interacts with the polarly localized histidine protein kinases DivJ and CckA , which are both involved in cell cycle regulation [58 , 59] . We speculate that the PAS domain may be important for recruitment of RgsP to PG biosynthesis sites . The nature of the signal , perceived by the RgsP PAS domain , and its regulatory output remain to be investigated . RgsP PDE activity substantially contributes to c-di-GMP turnover in S . meliloti . To our knowledge , RgsP is the only c-di-GMP PDE which is polarly localized in S . meliloti . Polarly localized c-di-GMP PDEs in C . crescentus and P . aeruginosa contribute to heterogeneity in the cellular c-di-GMP content [60–62] . Lack of the RgsP PDE activity did not cause apparent phenotypic defects . Yet , we cannot exclude that RgsP PDE activity has a regulatory function in S . meliloti , which may be linked to its localization . A regulatory function may have escaped our phenotypic analyses or may not have been detected because of compensation by any of the twelve additional c-di-GMP PDEs encoded by the S . meliloti genome [17] . Whereas the RgsPSm GGDEF domain seems to be inactive because of a degenerate active site [17] , RgsPAt contains an intact GGDEF motif , suggesting enzymatic activity of this protein . Since RgsPAt complemented depletion of RgsPSm in S . meliloti , the DGC activity of RgsP is unlikely to be relevant for its growth-promoting function . However , the GGDEF domain may modulate PDE activity of the EAL domain , as for example in case of C . crescentus PdeA or P . aeruginosa BifA and RmcA [61 , 63 , 64] . Overall , our study suggests that compared to well-studied γ-proteobacterial models , α-rhizobia utilize different sets of proteins for PG metabolism during cell elongation and division , and for relocating the PG growth machinery from a pole to the cell division site . Identification of further enzymes involved in PG synthesis and remodeling , and understanding the regulatory roles of RgsP and RgsM will lead to the clarification of specific mechanisms for polar PG biogenesis in α-rhizobia .
Bacterial strains and plasmids used in this study are shown in S5 Table . S . meliloti was grown at 30°C in tryptone-yeast extract ( TY ) medium [65] , LB medium [66] , modified MOPS-buffered minimal medium [67] , and nutrient-depleted 30% minimal medium ( nitrogen , carbon , and phosphate sources reduced to 30% ) . R . etli was grown in TY medium and A . tumefaciens was grown in LB medium at 30°C . E . coli was grown in LB medium at 37°C . For S . meliloti , antibiotics were used at the following concentrations ( mg/l; liquid/solid medium ) : kanamycin , 100/200 , gentamicin , 15/40 , tetracycline , 5/10 , spectinomycin , 200/200 and streptomycin , 600/600 . For E . coli the following concentrations were used: kanamycin , 50/50 , gentamicin , 5/8 , tetracyclin , 5/10 , spectinomycin 100/100 , and ampicillin 100/100 . For A . tumefaciens kanamycin was used at 100/200 and for R . etli at 30/30 . Unless otherwise specified , the inducers isopropyl β-D-1-thiogalactopyranoside ( IPTG ) and taurine were added at 0 . 5 and 20 mM , respectively . Chlorophenol red-β-D-galactopyranoside ( CPRG ) , an almost membrane-impermeable β-galactosidase substrate previously used to detect compromised E . coli cell envelope and lysis by purple staining of agar cultures [33] , was added at 20 μg/ml . Alkaline phosphatase substrate 5-bromo-4-chloro-3-indolylphosphate ( BCIP ) was used at 50 μg/ml and β-galactosidase substrate 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside ( X-Gal ) was used at 40 μg/ml . Growth assays were performed using 100 μl cultures in flat-bottom 96-well plates ( Greiner ) , grown at 30 oC with shaking at 1 , 200 rpm . Three to six culture replicates were analyzed per strain . Optical density ( OD600 ) was recorded using Infinite M200 PRO fluorescence reader ( Tecan ) . For growth assays involving protein depletion , cultures with or without IPTG were inoculated with 0 . 15 μl of stationary TY preculture with IPTG and grown for 24 h . Relative growth was calculated as a ratio of OD600 of cells grown without IPTG and OD600 of cultures supplemented with 0 . 5 mM IPTG . For growth assays , involving taurine- or IPTG-induced gene overexpression , the cultures containing or not containing taurine or IPTG were inoculated with 0 . 15 μl of stationary TY preculture and the growth was recorded at the indicated time points . TY medium ( 5 g/l tryptone , 3 g/l yeast extract , 0 . 4 g CaCl2×2H2O ) . LB medium ( 10 g/l tryptone , 5 g/l yeast extract , 5 g/l NaCl ) . MOPS-buffered minimal medium ( MM ) ( 10 g/l MOPS , 10 g/l mannitol , 3 . 55 g/l sodium glutamate , 0 . 246 g/l MgSO4×7H2O , 0 . 25 mM CaCl2 , 2 mM K2HPO4 , 10 mg/l FeCl3×6H2O , 1 mg/l biotin , 3 mg/l H3BO3 , 2 . 23 mg/l MnSO4×H2O , 0 . 287 mg/l ZnSO4×7H2O , 0 . 125 mg/l CuSO4×5H2O , 0 . 065 mg/l CoCl2×6H2O , 0 . 12 mg/l NaMoO4×2H2O , pH 7 . 2 ) . Constructs used in this work were generated using standard cloning techniques and are listed in S5 Table . The primers used are listed in S6 Table . All constructs were verified by sequencing . Plasmids were transferred to S . meliloti by E . coli S17-1-mediated conjugation as previously described [68] . Electroporation was used to introduce plasmids to R . etli and A . tumefaciens following the protocol previously described [69] . To generate chromosomally integrated constructs encoding RgsP or RgsM with C-terminally fused enhanced green fluorescent protein ( EGFP ) or triple FLAG-tag ( 3×FLAG ) , the 700 to 800 bp 3' portion of the gene excluding the stop codon was cloned into suicide plasmids pK18mob2-egfp or pG18mob-3×flag yielding translational fusions of the C-terminal portion of the protein coding sequence to the corresponding tag . Integration of these gene fusion constructs into the S . meliloti genome by homologous recombination resulted in a replacement of the native gene copy with egfp- or 3xflag-tagged gene copy at the corresponding native chromosomal location . To construct markerless translational fusions of rgsP and rgsM to egfp or mCherry at the native chromosomal location , the 700–800 bp 3' portion of the gene fused to egfp or mCherry was cloned into suicide plasmid pK18mobsacB [70] together with 700–800 bp of adjacent downstream genomic region . The resulting constructs were introduced into S . meliloti and transconjugants were subjected to sucrose selection to obtain double recombinants that have lost the integrated vector [70] . Correct positions of chromosomally encoded gene fusions were verified by PCR . To obtain RgsP and RgsM depletion strains , plasmids designed to uncouple the native promoter from the coding sequence and to place the coding sequence under the control of IPTG-inducible promoters were constructed . The rgsP gene is most likely co-transcribed with the preceding rimJ gene , encoding a probable ribosomal-protein-alanine acetyltransferase ( S7A Fig ) . To uncouple transcription of rimJ and rgsP , and to place rgsP under the control of an IPTG-inducible promoter , the lac-T5 tandem promoter sequence was inserted between rgsP and rimJ without altering the rimJ open reading frame . To this end , a DNA fragment containing the IPTG-inducible T5 promoter and a Shine-Dalgarno sequence followed by the partial 5' rgsP coding sequence starting from the start codon ( 586 bp ) was PCR-amplified from rgsP expression plasmid pWBT-SMc00074 [17] . This fragment was inserted into pK18mob2 [70] downstream of the IPTG-inducible lac promoter , thus generating a lac-T5 tandem promoter . In the case of RgsM depletion , the partial 5' rgsM coding sequence starting from the start codon ( 406 bp ) was PCR-amplified from S . meliloti genomic DNA and inserted into pK18mob2 downstream of the IPTG-inducible lac promoter . Integration of these constructs into the genome placed the full-length protein coding sequence under the control of the corresponding IPTG-inducible promoter . Adjacent open reading frames were not affected by integration of these constructs . Conditional RgsP and RgsM depletion was then achieved by constitutively expressing the lacI repressor gene from vector pSRKGm in the strains Rm2011 rgsPdpl and Rm2011 rgsMdpl , respectively . Gene overexpression constructs were generated by insertion of the corresponding coding sequences downstream of either the IPTG-inducible lac promoter in medium-copy vectors pSRKKm and pSRKGm [71] , the IPTG-inducible T5 promoter in medium-copy vector pWBT , the taurine-inducible tauA promoter in low-copy vector pR-Ptau , or a constitutive synthetic promoter ( Psyn ) in low-copy vector pR_egfp [72] . In the case of pR_egfp , a 114 bp region upstream of the rgsP coding region was included . To generate single-copy ectopic rgsP or rgsP-egfp expression constructs , we first determined the native rgsP promoter and its activity levels . At its native chromosomal locus , rgsP is most likely in an operon with rimJ . Therefore , a 300 bp genomic region upstream of rgsP as well as a 944 bp region including the putative rimJ promoter and the whole rimJ coding sequence ( S7A Fig ) were tested for promoter activity using fusions to egfp ( S7B Fig ) . Since higher levels of egfp expression were observed in the case of the 944 bp fragment , we used this DNA region as native rgsP promoter . To exclude possible non-desirable effects of an additional rimJ copy , a nonsense mutation was introduced 64 nucleotides downstream of the rimJ start codon yielding promoter construct P*rgsP ( S7A Fig ) . The rgsP , rgsP-egfp or rgsP-3×flag coding sequences were inserted downstream of the P*rgsP sequence in single-copy plasmid pABC2S-mob [73] . For generation of amino acid substitutions or protein variants lacking specific domains splicing by overlap extension PCR was applied . The rgsP promoter-egfp fusions were generated by insertion of the upstream non-coding region ( either long ( 944 bp ) or short ( 300 bp ) ) and the three first codons of rgsP into replicative medium-copy number plasmid pSRKKm-egfp [17] . Constructs for purification of His6-tagged proteins were generated by insertion of the coding sequence excluding the start codon into expression vector pWH844 [74] . Fusions of rgsM to phoA were assembled from the full-length or partial rgsM coding sequence and the E . coli phoA coding sequence missing the first 26 codons . For promoter-egfp activity assays , TY overnight cultures were diluted 1:500 in 100 μl of TY medium or 30% MM and grown in 96-well plates at 30°C with shaking at 1 , 200 rpm . EGFP fluorescence ( excitation 488 ± 9 nm; emission 522 ± 20 nm , gain 82 ) and OD600 were determined using the Infinite 200 Pro multimode reader ( Tecan ) and calculated as relative fluorescence units ( RFU ) , which represent fluorescence values divided by OD600 . Background EGFP fluorescence was determined using a control strain harboring pSRKKm-egfp . Fluorescence of three independent transconjugants was measured as biological replicates . Microscopy of bacteria on 1% agarose pads was performed using the Nikon microscope Eclipse Ti-E equipped with a differential interference contrast ( DIC ) CFI Apochromat TIRF oil objective ( 100x; numerical aperture of 1 . 49 ) and a phase-contrast Plan Apo l oil objective ( 100x; numerical aperture , 1 . 45 ) with the AHF HC filter sets F36-513 DAPI ( excitation band pass [ex bp] 387/11 nm , beam splitter [bs] 409 nm , and emission [em] bp 447/60 nm ) , F36-504 mCherry ( ex bp 562/40 nm , bs 593 nm , and em bp 624/40 nm ) , F36-525 EGFP ( ex bp 472/30 nm , bs 495 nm , and em bp 520/35 nm ) and F36-528 YFP ( ex bp 500/24 nm , bs 520 nm , and em bp 542/27 nm ) . Images were acquired with an Andor iXon3 885 electron-multiplying charge-coupled device ( EMCCD ) camera . For time-lapse analysis , MM 1% agarose pads were used , and images were acquired every 2 , 4 or 15 min at 30°C . IPTG was added to the medium at 0 . 2 or 0 . 5 mM for microscopy of cells harboring pSRKGm-parB-mCherry or pSRKGm-mCherry-ftsZ , respectively . Treatment of S . meliloti , R . etli and A . tumefaciens cells with fluorescently-labeled D-amino acid HADA was performed as previously described [75] . Briefly , cells were grown for 24 h in liquid medium in glass tubes to an OD600 of 0 . 4–0 . 6 . 80 μl of the cultures were then mixed with 0 . 25 μl 100 mM HADA solution and incubated for 2 . 5 min ( A . tumefaciens ) , 3 min ( S . meliloti ) or 3 . 5 min ( R . etli ) at 30 oC with shaking at 800 rpm . After addition of 186 μl 100% ethanol and 5–20 min incubation at room temperature ( RT ) , cells were washed three times with 0 . 9% NaCl and subsequently placed onto 1% agarose pads . Cell viability was assessed by DNA staining with the fluorescent intercalating agent propidium iodide . 100 μl of S . meliloti liquid cultures ( OD600 0 . 4–0 . 8 ) were mixed with 1 μl of 2 mg/ml propidium iodide stock solution and incubated for 5 min at room temperature . Cells were washed three times with 0 . 9% NaCl and subsequently placed onto 1% agarose pads . Concentrated S . meliloti cell suspensions were high pressure frozen ( HPF Compact 02 , Wohlwend , CH ) and freeze-substituted ( AFS2 , Leica , Wetzlar , Germany ) in a medium based on acetone , containing 0 . 25% osmium tetroxide , 0 . 2% uranyl acetate and 5% ddH2O according to the following protocol: -90°C for 20 h , from -90°C to -60°C in 1 h , -60°C for 8 h , -60°C to -30°C in 1 h , -30°C for 8 h , -30°C to 0°C in 1 h , 0°C for 3 h . Still at 0°C , samples were washed three times with acetone before a 1:1 mixture of Epon 812 substitute resin ( Fluka , Buchs , CH ) and acetone was applied at room temperature for 2 h . The 1:1 mixture was substituted with pure resin to impregnate the samples overnight . After another substitution with fresh Epon , samples were polymerized at 60°C for 2 days . The sample containing polymerized Epon blocks were then trimmed with razor blades and cut to 50 nm ultrathin sections using an ultramicrotome ( UC7 , Leica , Wetzlar , Germany ) and a diamond knife ( Diatome , Biel , Switzerland ) . Sections were applied onto 100 mesh copper grids coated with pioloform . For additional contrast , mounted sections were post-stained with 2% uranyl acetate for 20–30 min and subsequently with lead citrate for another 1–2 min . The sections were finally analyzed and imaged using a JEM-2100 transmission electron microscope ( JEOL , Tokyo , Japan ) equipped with a 2k x 2k F214 fast-scan CCD camera ( TVIPS , Gauting , Germany ) . Strain Rm2011 rgsPdpl ectopically expressing rgsP variants from P*rgsP was grown in triplicates in liquid TY medium without IPTG and harvested in the exponential growth phase 24 h after inoculation . Quantification of intracellular c-di-GMP was performed as previously described [76] . Briefly , cells were collected by centrifugation and nucleotides were extracted three times with acetonitrile/methanol/water ( 2:2:1 ) , dried and subjected to liquid chromatography-tandem mass spectrometry . c-di-GMP was normalized to total protein , determined using Bradford reagent ( Bio-Rad ) . Heterologous protein expression and purification was performed as previously described [17] . E . coli BL21 ( DE3 ) harboring expression plasmids were grown in LB medium in flasks to OD600 of 0 . 5–0 . 6 and protein expression was induced with 0 . 4 mM IPTG overnight at RT . Cells were lysed using French press ( pressure 1 , 000 lb/in2 ) and the lysates were centrifuged for 60 min at 24 , 000 ×g and 4°C . Cleared lysates were applied to His SpinTrap columns ( GE Healthcare ) following the manufacturer’s instructions and eluted with 0 . 5 M imidazole . Purity of isolated proteins was assessed by SDS-PAGE and Coomassie brilliant blue staining . Protein concentration was determined using Bradford reagent ( Bio-Rad ) . [α-32P]-labeled c-di-GMP was synthesized using purified Caulobacter crescentus His6-DgcA at 10 μM from GTP and [α-32P]-GTP ( 0 . 1 μCi/μl ) at 1 mM in the reaction buffer ( 50 mM Tris-HCl , 300 mM NaCl , 10 mM MgCl2 , pH 8 . 0 ) , overnight at 30°C . The reaction was then treated with 5 units of calf intestine alkaline phosphatase ( Fermentas ) for 1 h at 22°C to hydrolyze unreacted GTP and stopped by incubation for 10 min at 95°C . The precipitated proteins were removed by centrifugation ( 10 min , 20 , 000 ×g , 22°C ) and the supernatant was used for the PDE activity and the c-di-GMP binding assays . c-di-GMP binding was determined using a differential radial capillary action of ligand assay ( DRaCALA ) with [α-32P]-labeled c-di-GMP , as previously described [77] . This assay is based on the ability of dry nitrocellulose to prevent diffusion of bound protein-ligand complexes and thereby separate them from free ligand . Reaction mixtures ( 50 μl ) containing [α-32P]-labeled c-di-GMP and 20 μM of indicated protein in the binding buffer ( 10 mM Tris , 100 mM NaCl , 5 mM MgCl2 , pH 8 . 0 ) were incubated for 10 min at RT . 10 μl of this reaction mixture was spotted onto nitrocellulose membrane and allowed to dry prior to exposing a phosphor-imaging screen ( Molecular Dynamics ) . Data were collected using a STORM 840 scanner . DGC and PDE activities were determined as previously described [78 , 79] . Reaction mixtures ( 40 μl ) containing purified proteins at 10 μM , in reaction buffer ( 50 mM Tris-HCl , 300 mM NaCl , 10 mM MgCl2 , pH 8 . 0 ) were first pre-incubated for 5 min at 30°C . DGC reactions were initiated by adding GTP/[α-32P]-GTP ( 0 . 1 μCi/μl ) to 1 mM , incubated at 30°C for the indicated periods of time and stopped by adding an equal volume of 0 . 5 M EDTA . PDE reactions were initiated by adding [α-32P]-labeled c-di-GMP and stopped by adding an equal volume of 0 . 5 M EDTA after indicated time periods . 2 μl of the PDE or DGC reaction mixtures were spotted on polyethyleneimine-cellulose TLC chromatography plates , developed in 2:3 ( v/v ) 4 M ( NH4 ) 2SO4/1 . 5 M KH2PO4 ( pH 3 . 65 ) . Plates were dried prior to exposing a phosphor-imaging screen ( Molecular Dynamics ) . Data were collected and analyzed using a STORM 840 scanner ( Amersham Biosciences ) . Co-IP and protein identification by mass spectrometry was performed as previously described including small modifications [80] . Cultures of Rm2011 rgsP-3×flag , Rm2011 rgsM-3×flag and control strain Rm2011 harboring the empty vector pWBT were grown in TY medium in flasks to an OD600 of 0 . 6 and cross-linked with 0 . 1% formaldehyde for 15 min at RT . Reaction was quenched by adding glycine at a final concentration of 0 . 35 M . Cells were washed , resuspended in lysis buffer ( 50 mM Tris-HCl , 150 mM NaCl , 1 mM EDTA , 1% Triton X-100 , 2 mM phenylmethylsulfonyl-fluorid [PMSF] , pH 7 . 4 ) and lysed using a French press ( pressure 1 , 000 lb/in2 ) . Cleared lysates obtained after ultracentrifugation ( 100 , 000 ×g , 1 h , 4°C ) were incubated with anti-FLAG M2 affinity gel ( FLAG Immunoprecipitation Kit , Sigma ) overnight at 4°C on a rolling shaker . Bound proteins were eluted with 3×FLAG peptide solution . For mass-spectrometry analysis , proteins were digested by Sequencing Grade Modified Trypsin ( Promega ) at 37°C overnight . The mass spectrometric analysis was performed using an Orbitrap Velos Pro mass spectrometer ( Thermo Fisher Scientific ) . An Ultimate nanoRSLC-HPLC system ( Dionex ) , equipped with a custom 20 cm x 75 μm C18 RP column filled with 1 . 7 μm beads was connected online to the mass spectrometer through a Proxeon nanospray source . 1-15 μl of the tryptic digest were injected onto a C18 pre-concentration column . Automated trapping and desalting of the sample was performed at a flow rate of 6 μl/min using water/0 . 05% formic acid as solvent . Separation of the tryptic peptides was achieved with the following gradient of water/0 . 05% formic acid ( solvent A ) and 80% acetonitrile/0 . 045% formic acid ( solvent B ) at a flow rate of 300 nl/min: holding 4% B for 5 min , followed by a linear gradient to 45% B within 30 min and linear increase to 95% solvent B in additional 5 min . The column was connected to a stainless steel nanoemitter ( Proxeon , Denmark ) and the eluent was sprayed directly towards the heated capillary of the mass spectrometer using a potential of 2 , 300 V . A survey scan with a resolution of 60 , 000 within the Orbitrap mass analyzer was combined with at least three data-dependent MS/MS scans with dynamic exclusion for 30 s either using CID with the linear ion-trap or using HCD combined with orbitrap detection at a resolution of 7 , 500 . Data analysis was performed using Proteome Discoverer ( Thermo Fisher Scientific ) with SEQUEST and MASCOT ( version 2 . 2; Matrix science ) search engines using either SwissProt or NCBI databases . The combined transmembrane topology and signal peptide prediction online tool PHOBIUS [27] was used to predict regions with high hydrophobicity within candidate protein interactions partners of RgsP and RgsM . Bacterial two-hybrid analysis was performed as previously described [31] . The adenylate cyclase-deficient strain E . coli BTH101 was co-transformed with plasmids carrying rgsP∆GGDEF∆EAL or rgsM translationally fused to T25 and T18 fragments of Bordetella pertussis adenylate cyclase . Transformant colonies were grown in 100 μl LB supplemented with antibiotics at 30°C for 6 hours with shaking 1 , 200 rpm . 10 μl of each culture was spotted onto LB agar plates containing kanamycin , ampicillin , X-Gal and IPTG . Plates were imaged after 30 h of incubation at 30°C . β-galactosidase activity was determined as previously described with small modifications [52] . Cells grown on LB agar containing IPTG were resuspended in 1 ml of Z-buffer ( 60 mM Na2HPO4 , 40 mM NaH2PO4 , 10 mM KCl , 1 mM MgSO4 , pH 7 . 0 ) , in triplicates and the OD600 was recorded . Cell permeabilization was facilitated by addition of 50 μl of chlorophorm and 50 μl of 0 . 05% SDS . The aqueous phase was mixed with an equal volume of Z-buffer containing 50 mM β-mercaptoethanol , and ortho-nitrophenyl-β-D-galactopyranoside ( ONPG ) was added to the final concentration of 0 . 5 mg/ml . A420 was recorded using the Infinite M200 PRO fluorescence reader ( Tecan ) . Miller Units ( MU ) were calculated as follows: MU = 1000*A420/ ( t*V*OD600 ) . t represents the time in min and V the volume in ml . Western blot analysis was performed as previously described [68] . Briefly , S . meliloti and E . coli strains expressing 3×flag-tagged rgsP or rgsM variants were grown in glass tubes supplemented with corresponding antibiotics . Unless otherwise specified , cells were grown in TY medium without IPTG and collected at an OD600 of 0 . 4–0 . 8 24 h after inoculation . Cells were adjusted to an OD600 of 1 , 10 μl of lysed cells were loaded to SDS-PAGE gel and separated proteins were transferred to PVDF membrane ( Thermo Fisher Scientific ) . RgsP protein variants were detected using anti-FLAG M2-Peroxidase ( HRP ) antibody ( Sigma-Aldrich ) . Membranes were incubated with Pierce ECL Western Blotting Substrate ( Thermo Fisher Scientific ) and imaged using the luminescence image analyzer LAS-4000 ( Fujifilm ) . Isolation of PG sacculi was achieved following a published protocol [81] . Briefly , S . meliloti strains were grown in 400 ml TY and LB media in flasks for 24 h at 30°C until OD600 reached 0 . 4–0 . 8 . E . coli strains were grown in 400 ml LB for 4 h at 37°C until OD600 reached 0 . 5–0 . 9 . Cells were harvested by centrifugation , resuspended in PBS buffer , added dropwise to boiling 8% SDS solution and stirred for 30 min . PG sacculi were pelleted by ultracentrifugation in a Beckman Coulter Optima at 440 , 000 ×g , at ambient temperature for 1 h and resuspended in water . Centrifugation and resuspension steps were repeated until the sacculi were free of SDS as verified by a published assay [82] . The sacculi were then incubated with 100 μg/ml of α-amylase in reaction buffer ( 10 mM Tris-HCl , 10 mM NaCl , pH 7 . 0 ) for 2 h at 37°C , followed by incubation with 200 μg/ml pronase E for 1 h at 60°C , to remove high molecular weight glycogen and proteins , respectively . The enzymes were removed by incubation in 4% SDS solution for 15 min at 80°C . Sacculi were washed free of SDS as described above and resuspended in 0 . 02% sodium azide . Purified sacculi were digested with the muramidase cellosyl ( Hoechst , Frankfurt , Germany ) at 37°C overnight . The reaction was terminated by boiling the sample at 100°C for 10 min . The samples were centrifuged ( 15 , 000 ×g for 10 min ) and the soluble muropeptides were recovered from the supernatant , reduced with NaBH4 and separated by HPLC as described for E . coli [81] . Muropeptides were then separated on a C18 reversed-phase column ( ProntoSIL ) using an Agilent 1220 infinity HPLC system . Separation was carried out over a 180 min linear gradient from buffer A ( 50 mM sodium phosphate , pH 4 . 31 ) to buffer B ( 75 mM sodium phosphate , pH 4 . 95 , 15% methanol ) . Muropetides were detected by their absorbance at 205 nm . Muropeptides of interest corresponding to major peaks were manually collected and analyzed by tandem mass spectrometry ( MS/MS ) as previously described [83] . Purified PG ( ~100 μg ) from S . meliloti strain Rm2011 was centrifuged at 15 , 000 ×g , 4°C for 14 min and resuspended in binding buffer ( 10 mM Tris-maleate , 10 mM MgCl2 , 50 mM NaCl , pH 6 . 8 ) . 10 μg of protein of interest was incubated with or without PG in a final volume of 100 μl , and incubated at 4°C for 30 min . Samples were centrifuged as described above and the supernatant was collected ( supernatant fraction ) whilst the pellet was resuspended in 200 μl of binding buffer . Another centrifugation step as described above was carried out ( wash fraction ) . Bound proteins were released from PG by incubation with 100 μl of 2% SDS , at 4°C for 1 h before being collected by a final centrifugation step as done at earlier steps . The proteins present in the different fractions were analyzed by SDS-PAGE . Proteins ( 2 μM final concentration ) were incubated with 10 μg of sacculi from S . meliloti , overnight at 37°C , in a volume of 100 μl ( containing 10 mM HEPES-NaOH , 1 mM ZnCl2 , 150 mM NaCl , 0 . 05% Triton X-100 , pH 7 . 5 ) . Proteins were inactivated by boiling for 10 min . Cellosyl was added ( 1 μM final concentration ) and the samples were incubated overnight again at 37°C . Samples were boiled for 10 min and the soluble muropeptides were collected by centrifugation at 15 , 000 ×g for 10 min , and taking the supernatant fraction . Muropeptides were reduced with NaBH4 and analyzed by HPLC as described above . | Bacteria face the challenge of maintaining their peptidoglycan cell wall integrity during growth and division . The enzymes involved in cell wall biogenesis are tightly regulated and targeting of these enzymes by β-lactams cause cell death . In contrast to the well-characterized mode of dispersed cell wall formation in many rod-shaped bacteria , the mechanisms controlling polar cell wall formation in α-proteobacteria are largely unknown . Seven-transmembrane receptors ( 7TMRs ) are widespread in eukaryotes and prokaryotes but only few of them have been functionally studied in bacteria . Here we suggest that a 7TMR-DISM protein cooperates with a putative peptidoglycan-hydrolyzing protein , to facilitate unipolar cell wall formation and cell division in the Rhizobiales . This 7TMR-DISM protein also contributes to degradation of the second messenger cyclic di-GMP . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"cell",
"walls",
"cell",
"physiology",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"microbiology",
"light",
"microscopy",
"cell",
"polarity",
"plant",
"science",
"microscopy",
"materials",
"science",
"cell",
"growth",
"cellular",
"structures",
"... | 2018 | Seven-transmembrane receptor protein RgsP and cell wall-binding protein RgsM promote unipolar growth in Rhizobiales |
The COP9 signalosome ( CSN ) is a highly conserved multifunctional complex that has two major biochemical roles: cleaving NEDD8 from cullin proteins and maintaining the stability of CRL components . We used mutation analysis to confirm that the JAMM domain of the CSN-5 subunit is responsible for NEDD8 cleavage from cullin proteins in Neurospora crassa . Point mutations of key residues in the metal-binding motif ( EXnHXHX10D ) of the CSN-5 JAMM domain disrupted CSN deneddylation activity without interfering with assembly of the CSN complex or interactions between CSN and cullin proteins . Surprisingly , CSN-5 with a mutated JAMM domain partially rescued the phenotypic defects observed in a csn-5 mutant . We found that , even without its deneddylation activity , the CSN can partially maintain the stability of the SCFFWD-1 complex and partially restore the degradation of the circadian clock protein FREQUENCY ( FRQ ) in vivo . Furthermore , we showed that CSN containing mutant CSN-5 efficiently prevents degradation of the substrate receptors of CRLs . Finally , we found that deletion of the CAND1 ortholog in N . crassa had little effect on the conidiation circadian rhythm . Our results suggest that CSN integrity plays major roles in hyphal growth , conidial development , and circadian function in N . crassa .
The COP9 signalosome ( CSN ) is an evolutionarily conserved multifunctional complex in eukaryotes; it is composed of eight subunits ( CSN1–CSN8 ) in plants and mammals [1] . The CSN was initially discovered to be an important regulator of photomorphogenesis in Arabidopsis thaliana [2] and was later found to participate in a wide range of processes [1] , [3] . The CSN potentially influences these cellular pathways by regulating the activity of cullin-RING ubiquitin ligases ( CRLs , e . g . , CRL1 , CRL3 , and CRL4 complexes in most eukaryotes ) [3] , [4] , [5] . CRLs are a big family of ubiquitin ligases that share a common cullin/RING-E2 module [6] , [7] , [8] . They are necessary for substrate ubiquitination in a cascade of enzymatic reactions involving E1 , E2 , and E3 [9] . Under the control of the CSN-regulated ubiquitin-proteasome pathway , cells coordinate the expression of an array of genes involved in the regulation of growth and development in order to respond to environmental signals , such as light , temperature , and changes in nutrient conditions [1] , [10] . Loss-of-function mutations in CSN subunits result in dysfunction of hundreds of CRLs [3] , which explains the pleiotropic phenotypes observed in CSN mutants [1] , [3] , [7] . In 2002 , Deshaies and his colleagues first described that the CSN-5 metalloprotease ( JAMM ) motif is required for removing the ubiquitin-like protein NEDD8 from Cul1 [11] . Later studies confirmed that the isopeptidase activity of the CSN complex is responsible for cullin deneddylation in eukaryotes [1] , [3] , [4] , [5] , [12] . In this process , the CSN binds to CRL E3 ligase and cleaves NEDD8 from cullins via the catalytic activity of its CSN-5 subunit , and then inhibits CRL activity [5] , [13] , [14] , [15] . Thus , the deneddylation activity of CSN requires the metalloprotease motif located in the CSN-5 subunit and the functional core subunits of the CSN [11] , [16] . However , CSN-5–dependent metalloprotease activity is not essential in Schizosaccharomyces pombe , as no obvious phenotype was detected in csn-5 deletion strains [17] , [18] . The physiological importance of CSN deneddylation activity in development and cell differentiation was examined in Drosophila melanogaster , in which the lethality of csn-5Δ/Δ animals was rescued by expression of a CSN-5 transgene but no adult flies were recovered upon equivalent expression of CSN-5 ( D148N ) ( loss of deneddylation activity ) [11] , [19] . In CSN-5-downregulated HeLa cells , however , the accelerated degradation of c-Jun was rescued equally by over-expression of either the JAMM domain mutant CSN-5D151N or wild-type CSN-5 [20] . These results suggest that the requirement for neddylation/deneddylation cycle of cullins is not absolutely necessary during normal growth and certain developmental stages . In plants , genetic studies suggest that although neddylation/deneddylation cycle is not absolutely necessary during early embryonic development and germination , it is required during seedling establishment and the later developmental stages [12] , [21] . In Aspergillus nidulans , deletion of csnE/csn-5 or mutation in JAMM domain results in a block in fruiting body formation at the primordial stage , with a few other observed phenotypic changes , such as light-dependent signaling [22] , [23] . Although deneddylation is a major activity of the CSN , it alone cannot explain all of the phenomena described above . These observations raise the possibility that the CSN may have other functional activities in addition to its deneddylation activity . Recent genetic evidence suggest that the CSN has one additional major function: it controls the stability of CRL ubiquitin ligases in vivo by mediating assembly/disassembly of CRL complexes and by protecting substrate receptors in CRLs from degradation [3] , [24] , [25] , [26] , [27] , [28] . A recent structural and biochemical study showed that the protective effect of the CSN on DDB2 and CSA autoubiquitination in CRL4 complexes does not require CSN-5–mediated deneddylation activity [29] . However , both of the CSN activities occur when CSN associates with cullins in CRL E3 complexes . Furthermore , there is also a tight correlation between CSN deneddylation activity and the ability of the complex to modulate the stability of CRLs [3] . Thus , it is difficult to determine which function is more important for growth and development through regulation of CRL activity , or how these two functions cooperate with each other in regulating CRL dynamicity in eukaryotes . In A . thaliana , the MPN ( Mpr-Pad1-N-terminal domain ) subunits CSN-5 and CSN-6 are essential for the structural integrity of the CSN holocomplex [12] . Several studies have shown that point mutations in the JAMM metal-binding site of CSN-5 do not interfere with the proper assembly of CSN complexes in S . pombe , A . thaliana , and A . nidulans [11] , [21] , [23] . In N . crassa , CSN-5 is not an essential gene; the deletion mutant can survive , and displays obvious growth and developmental defects , making it an excellent model system for investigating the distinctions between the deneddylation and CRL complex assembly/disassembly functions of the CSN [16] . The CSN takes part in a wide range of cellular and developmental processes in N . crassa , including hyphal growth , conidial formation , light and temperature responses , and circadian clock function [16] , [24] . To further investigate its biological function in vivo , we created a series of point mutations in the JAMM metal-binding motif of the CSN-5 subunit to disrupt the deneddylation activity of the CSN complex . In those mutant strains , the integrity of the CSN and its interactions with Cul1 and Cul4 were not affected . Surprisingly , mutated CSN-5 almost retained the ability to restore the phenotypic defects of a csn-5KO strain and partially maintained the stability of the SCFFWD-1 complex , which was able to carry out degradation of the clock protein FREQUENCY ( FRQ ) in vivo . Moreover , the stability of four other substrate receptors of CRLs can be efficiently restored by the CSN containing mutant CSN-5 . However , deletion of the CAND1ortholog in N . crassa had little effect on conidiation circadian rhythm and the degradation of FRQ . Our results suggest that the integrity of CSN plays major roles in hyphal growth , conidial development , and circadian function in N . crassa .
The N . crassa genome encodes seven COP9 signalosome subunits ( CSN-1–CSN-7 ) [16] , [24] . Several studies have shown that the JAMM metal-binding sites in the MPN domain of CSN-5 are required for metalloprotease activity in the CSN [11] , [21] , [23] . When the CSN-5 protein sequence was used in a BLAST search against protein databases , a highly conserved MPN domain in the N . crassa CSN-5 subunit was identified . As shown in Figure 1A , three conserved residues corresponding to His127 , His129 , and Asp140 lie within the putative metal-binding motif ( EXnHXHX10D ) of the N . crassa CSN-5 JAMM domain . To determine whether these conserved residues form the metalloprotease-like active site of JAMM , we used the JAMM domain of Archaeoglobus fulgidus as a template to generate the tertiary structure of N . crassa CSN-5 [30] . Because of the low similarity between these two JAMM domains , the generated structure was poor . Thus , we instructed SWISS-MODEL to automatically select a template protein for generating the structure of N . crassa CSN-5 [31] . SWISS-MODEL selected the pre-mRNA splicing factor Prp8 as template ( Protein Data Bank [PDB] accession number 2P8R ) for N . crassa CSN-5 . The functional sites were mapped into predicted structure according to the structural alignment with AfJAMM ( PDB accession number 1R5X ) . As shown in Figure 1B , His127 , His129 , and Asp140 within EXnHXHX10D of the N . crassa CSN-5 JAMM corresponded to the putative metal-binding motif as metalloprotease-like active site in AfJAMM [30] , [32] . To confirm the contribution of CSN-5 to CSN-mediated deneddylation of cullins , we mutated these three highly conserved amino acids ( H127A , H129A or D140N ) using site-directed mutagenesis . We then introduced quinic acid ( QA ) –inducible Myc-tagged wild-type CSN-5 or one of the three mutant CSN-5 constructs into a csn-5KO strain expressing Myc-Cul1 protein . As shown in Figure 1C , Myc-CSN-5 , Myc-CSN-5H127A , Myc-CSN-5H129A , and Myc-CSN-5D140N were expressed in the csn-5KO strains in the presence of QA . Expression of Myc-tagged wild-type CSN-5 in the csn-5KO strain resulted in a decrease in hyperneddylated Cul1 to the level of the wild-type strain ( Figure 1C ) , indicating that the Myc-tagged CSN-5 protein was functional for CSN deneddylation activity . In contrast , expression of mutant CSN-5 ( H127A , H129A , or D140N ) failed to decrease the hyperneddylated Cul1 in the csn-5KO strain ( Figure 1C ) . Similarly , hyperneddylation of Cul3 ( Figure 1D ) and Cul4 ( Figure 1E ) in the csn-5KO strain was rescued by expressing the Myc-tagged wild-type CSN-5 , but not by any of the mutated Myc-CSN-5s . This indicates that the metal-binding motif of JAMM is essential for CSN-mediated deneddylation of cullins . Because all of the Cul3 and Cul4 was neddylated while not all of the Cul1 was neddylated in the csn-5KO strain and csn-5KO strains complemented by JAMM-domain mutant CSN-5 , we rechecked Cul1 modification in the csn mutants . As shown in Figure S1 , c-Myc antibody detected three specific protein bands in first generation of csn-5KO or csn-6KO transformants and two specific bands in the csn-1KO transformants . In most positive transformants , there was slightly less unneddylated Cul1 than neddylated Cul1 , but the signal remained strong . This is different from the studies in yeast , plants , and fruit fly in which deletion of csn-5 results in hyperneddylation of Cul1 [11] , [21] , [33] . Possible explanations are that N . crassa genome codes for another deneddylase in addition to CSN complex or there is large amount of newly synthesized Cul1 proteins . We next examined the neddylation of Cul4 using a polyclonal antibody that recognizes the N terminus of N . crassa Cul4 . As shown in Figure 1F , only the neddylated Cul4 was detected in csn-5KO strain , while in the wild-type strain , most of the detected Cul4 was unneddylated . Next , we transferred endogenous csn-5 promoter-driven constructs of either wild-type CSN-5 or CSN-5 with JAMM triple point mutations ( H127A , H129A , and D140N ) ( hereafter referred to as CSN-5tri ) into a csn-5KO strain expressing Myc-Cul1 protein . Myc-CSN-5 and Myc-CSN-5tri were expressed in the csn-5KO strains ( Figure 1G ) . Similar to what we observed in csn-5KO expressing CSN-5 with a single point mutation ( Figure 1C ) , expression of the CSN-5tri failed to decrease the hyperneddylation of Cul1 in the csn-5KO strain ( Figure 1G ) as well . Interestingly , the amount of unneddylated Cul1 in csn-5KO strains expressing either single ( Figure 1C ) or triple ( Figure 1G ) point mutant CSN-5 was less than that in a csn-5KO strain . Furthermore , expression of CSN-5tri in the csn-5KO strain also failed to decrease hyperneddylated Cul4 to the levels observed in the wild-type or csn-5KO strain complemented with wild-type CSN-5 ( Figure 1H ) . Taken together , these data confirm that the JAMM domain metal-binding motif of N . crassa CSN-5 is essential for the deneddylation activity of the CSN . To examine whether the JAMM metal-binding site of CSN-5 functions in growth and development , we analyzed the phenotypes of the csn-5KO strain expressing either Myc-tagged wild-type or mutant CSN-5 . On minimal slants with QA , the csn-5KO strain produced fewer aerial hyphae and conidia than the wild-type strain ( Figure 2A ) . Expression of wild-type CSN-5 in the csn-5KO strain restored aerial hyphal growth and conidial formation to levels similar to those in the wild-type strain ( Figure 2A ) . Surprisingly , when csn-5KO , Myc-CSN-5H127A; csn-5KO , Myc-CSN-5H129A; and csn-5KO , Myc-CSN-5D140N strains ( hereafter referred to as csn-5H127A , csn-5H129A , and csn-5D140N , respectively ) were grown in minimal slants containing QA , the transformants exhibited hyphal formation and conidiation that were the same as the wild-type strain and the csn-5KO , Myc-CSN-5 strain ( Figure 2A ) . We next measured the growth rates of the wild-type strain , the csn-5KO strain , and the transformants by race tube assay in constant darkness . Interestingly , the growth of csn-5H127A , csn-5H129A , and csn-5D140N strains was slightly faster than that of the wild-type strain ( approximately 4 . 2 cm per day vs . 3 . 7 cm per day , respectively ) and the csn-5KO , Myc-CSN-5 strain ( Figure 2B ) . These results suggest that these CSN-5s with a point mutation within the JAMM metal-binding motif function similarly as the wild-type CSN-5 on N . crassa growth and conidiation . In QA-containing race tubes , the conidiation rhythms of the csn-5H127A , csn-5H129A , and csn-5D140N strains ( a period of about 22 . 5 h ) were pretty much ( only slightly longer ) to those of the wild-type and csn-5KO , Myc-CSN-5 strains ( a period about 22 . 2 h ) ( Figure 2C ) in constant darkness after light entrainment . To characterize the effect on light response of each CSN-5 point mutation , we further examined the light-entrained conidiation rhythm of each csn-5KO transformant during light–dark ( LD ) cycles ( 12 h light/12 h dark ) . As shown in Figure 2D , although the LD cycles entrained the conidiation rhythm of the csn-5KO strains expressing wild-type CSN-5 or mutant CSN-5 , however , the conidiation bands of the csn-5H127A , csn-5H129A , and csn-5D140N strains were broader than those of the wild-type and csn-5KO , Myc-CSN-5 strains . Similarly , 12 h 27°C/12 h 22°C temperature cycles entrained the conidiation rhythm of the csn-5H127A , csn-5H129A , and csn-5D140N strains , but not the patterns of conidiation bands ( Figure 2E ) . Taken together , these results suggest that point mutations within CSN-5 are functional in growth and conidiation , and partially functional in circadian rhythm , light response , and temperature-entrained clock process . The loss of deneddylation activity of the JAMM domain mutations may be due to the disruption of the CSN complex . To examine this , we tested the interactions between the CSN-6 subunit and wild-type or mutant CSN-5s . Myc-tagged CSN-6 was co-expressed with Flag-tagged CSN-5 or mutant CSN-5 proteins in csn-5KO strains . As shown in Figure 3A , the Flag-tagged CSN-5 strongly interacted with Myc-tagged CSN-6 in an immunoprecipitation reaction , suggesting that they were both in the intact CSN complexes . As expected , the Flag antibody pulled down the Myc-tagged CSN-6 protein in the csn-5KO strain co-expressing Myc-CSN-6 and each of the mutant Flag-CSN-5 proteins ( Figure 3A ) , similar to what was observed in the csn-5KO strain co-expressing CSN-6 and wild-type CSN-5 . This result indicates that the point mutations within the CSN-5 JAMM metal-binding motif did not affect the interactions between the CSN-5 and CSN-6 subunits and those two MPN proteins within PCI ( Proteasome , COP9 , eukaryotic Initiation factor 3 ) complexes may form dimers . To further examine whether Myc-His-tagged CSN-5 point mutants are incorporated into a larger molecular mass CSN complex , we performed gel filtration and followed by Western blot analysis . As shown in Figure 3B , like wild-type CSN-5 , CSN-5H127A , CSN-5H129A , and CSN-5D140N fusion proteins were eluted in larger molecular mass fractions , suggesting that each of the Myc-tagged CSN-5 point mutants can be incorporated into the intact CSN complex . Using protein affinity purification followed by Mass Spectrometry analysis , we further examined whether the CSN complex is properly assembled with CSN-5 point mutants . Myc-His-tagged CSN-5H127A , CSN-5H129A , CSN-5D140N , or wild-type CSN-5 was purified on a nickel column followed by immunoprecipitation with a c-Myc monoclonal antibody . As shown in Figure 3C , similar immunoprecipitated protein profiles were detected in the Myc-His-CSN-5H127A , Myc-His-CSN-5H129A , Myc-His-CSN-5D140N , and Myc-His-CSN-5 ( wild-type CSN-5 ) samples , but not in the wild-type strain ( negative control ) . Liquid chromatography–mass spectrometry/mass spectrometry ( LC-MS/MS ) analysis of excised gel bands led to the identification of seven subunits , from CSN-1 to CSN-7a , in the Myc-His-CSN-5 purified products and in the Myc-His-CSN-5H127A purified products ( Figure 3C ) . Taken together , these results confirm that the integrity of the CSN complex is not affected by mutations within the JAMM motif of CSN-5 in N . crassa . Next , we examined whether CSN complexes with mutant CSN-5 subunits can still interact with Cul1 protein . As shown in Figure 3D , both wild-type CSN-5 and each of the mutant CSN-5 proteins co-immunoprecipitated with Cul1 protein . We further examined whether CSN complexes with mutant CSN-5 subunits can also interact with Cul4 protein in vivo by IP/western blotting experiments . As shown in Figure 3E , the Myc-tagged wild-type CSN-5 co-immunoprecipitated with the neddylated and unneddylated Cul4 , indicating that the N . crassa CSN complex can interact with all species of Cul4 in vivo . Similarly , the Myc-tagged mutant CSN-5s also co-precipitated with Cul4 ( Figure 3E ) , further confirming that mutations within the JAMM metal-binding motif of CSN-5 do not interfere with interaction between CSN and cullins . These results strongly suggest that the point mutations within the JAMM metal-binding motif abolish NEDD8 isopeptidase activity but have no effect on the integrity of the CSN or on its interactions with cullins . In N . crassa , the clock protein FREQUENCY ( FRQ ) is a negative regulator in the negative feedback loop that controls the circadian clock under constant conditions [34] , [35] . Impaired FRQ degradation in csn-2 mutants results in the loss of circadian rhythm [24] . To investigate whether the mutant CSN-5s can rescue circadian rhythm defects in the csn-5KO strain , we examined the degradation of FRQ protein in the wild-type and csn-5KO strains expressing wild-type CSN-5 or mutant CSN-5s after addition of the protein synthesis inhibitor cycloheximide ( CHX ) . In the wild-type strain , the FRQ was gradually degraded after CHX treatment , with a half-life of about 2 . 5 h ( Figure 4A and 4B ) . However , in the csn-5KO strain , the degradation of FRQ was mostly blocked ( Figure 4A and 4B ) , similar to what was observed in the csn-2KO strain and the fwd1RIP mutant [24] , [36] . As shown in Figure 4A and 4B , the expression of Myc-tagged wild-type CSN-5 in the csn-5KO strain restored the degradation of FRQ to wild-type levels , so that the conidiation period on race tubes was similar to that of the wild-type strain ( Figure 2C ) . We next checked FRQ degradation in the csn-5KO strain expressing CSN-5 proteins with mutations in the JAMM metal-binding site . As shown in Figure 4A and 4B , the expression of Myc-tagged CSN-5H127A , CSN-5H129A , or CSN-5D140N in the csn-5KO strain partially rescued the degradation of FRQ in the csn-5KO strain . FRQ was degraded slightly slower in the mutants than the wild-type strain or the csn-5KO strain complemented by wild-type CSN-5 , with a half-life of ∼5 h , consistent with the prolonged period of the conidiation rhythms in the csn-5KO strains expressing the mutant CSN-5 , indicating that both deneddylation activity and integrity of CSN are needed in this process . Taken together , these results demonstrate that CSN-5 with point mutations in the JAMM metal-binding site partially restore the SCF-mediated FRQ degradation . Previous studies showed that FRQ ubiquitination and degradation is mediated by the SCFFWD-1 E3 ligase complex [24] , [36] , and that the stability of E3 ligase components is controlled by CSN in vivo [3] , [7] , [16] , [24] . Because the ectopic expression of mutated CSN-5 partially rescued both the circadian conidiation rhythm and FRQ degradation in the csn-5KO strain , we decided to check whether CSN with mutant CSN-5 can prevent the degradation of components of the SCFFWD-1complex . As shown in Figure 5A , Myc-Cul1 was stable after induced expression of Myc-CSN-5 in the csn-5KO strain , with a half-life of >9 h in the presence of CHX , similar to that of the wild-type strain . In the csn-5 mutant , however , both the neddylated and unneddylated Myc-Cul1 became very unstable , with a half-life about 1 . 5 h ( Figure 5A and 5D ) [16] . Interestingly , the expression of JAMM mutant CSN-5 had a differential effect on the neddylated and unneddylated Cul1 . In mutant CSN-5 transformants , the stability of neddylated Cul1 was only partially rescued , with a half-life of >3 h in the presence of CHX ( Figure 5A and 5D ) , whereas the stability of unneddylated Cul1 was almost rescued , with a half-life of >12 h ( Figure 5A and 5D ) . These data indicate that although CSN containing JAMM mutated CSN-5 fails to cleave NEDD8 from neddylated Cul1 , it still functions to protect hyperneddylated and unneddylated Cul1 from degradation to a certain extent . In N . crassa , deletion of csn-5 or csn-3 has no effect on the stability of SKP-1 protein in the SCFFWD-1 complex [16] . As expected , Myc-SKP-1 were very stable in the wild-type strain and the csn-5KO strain and in the complementation strains with mutant CSN-5 , with a half-life of >12 h ( Figure 5B and 5E ) . FWD-1 , the substrate-recruiting subunit of the SCFFWD-1 complex , was quite stable in the wild-type strain , whereas it became undetectable after only 3 h of CHX treatment in csn-5KO strain ( Figure 5C and 5F ) . In the csn-5H127A , csn-5H129A , and csn-5D140N strains , however , FWD-1 signals could still be detected after 6 h of CHX treatment ( Figure 5C and 5F ) , indicating that CSN with mutated CSN-5 partially functions to protect F-box proteins from degradation . This finding further confirms that regulation of SCF-mediated FRQ degradation by the CSN is a key step in the N . crassa circadian clock . Therefore , both the deneddylation activity and the integrity of the CSN are important for preventing the degradation of components of the SCFFWD-1 complex . We next asked whether CSN with mutated CSN-5 still functions to protect other CRL substrate receptors from degradation . N . crassa SCON-2 , an F-box protein involved in regulating sulfur metabolism , was previously shown to interact with SKP-1 and is very unstable in a csn-2KO strain [24] , [37] . We compared the stability of Myc-SCON-2 in wild-type , csn-5KO and csn-5KO expressing wild-type CSN-5 or mutant CSN-5H127A strains . The half-life of Myc-SCON-2 was approximately 12 h in the wild-type and csn-5KO expressing wild-type CSN-5 strains in the presence of CHX . Myc-SCON-2 was very unstable in the csn-5 mutant and became undetectable after 3 h of CHX treatment ( Figure 6A and 6B ) . In the csn-5H127A strain , the detection of Myc-SCON-2 signal extended to 6 h after CHX treatment ( Figure 6A and 6B ) . FBP94 ( NCU04785 ) , another F-box–containing protein in N . crassa , can also interact with SKP-1 ( data not shown ) . As shown in Figure 6C and 6D , FBP94 was quite stable in the wild-type strain and csn-5KO strain complemented with Myc-CSN-5 , whereas in the csn-5KO strain it became undetectable after only 6 h of CHX treatment . In the csn-5H127A strain , detection of FBP94 signal extended to 12 h after CHX treatment ( Figure 6C and 6D ) . Therefore , CSN complex with mutated JAMM domain can partially function in maintaining the stability of other F-box–containing adaptor proteins . In a previous study , we determined that the N . crassa Cul3 protein interacts with BTB1 protein , and both proteins become unstable in the csn-5KO strain [16] . The half-life of Myc-BTB1 was >12 h in the wild-type and csn-5KO expressing wild-type CSN-5 strains in the presence of CHX , whereas in the csn-5KO strain it became undetectable after 6 h of CHX treatment ( Figure 6E and 6F ) . As expected , in the csn-5H127A strain , BTB1 signals were detectable at 12 h after CHX treatment ( Figure 6E and 6F ) , indicating that CSN with the JAMM mutated CSN-5 still partially functions to protect the substrate adaptor proteins of CRL3 from degradation . We also investigated whether CSN with the JAMM mutated CSN-5 regulates the substrate receptor protein of CRL4 in a similar manner . N . crassa Cul4 was previously shown to interact with DCAF11 , a putative substrate receptor of CRL4DCAF11 [16] , [38] . As shown in Figure 6G and 6H , the half-life of Myc-DCAF11 was >12 h in the wild-type and csn-5KO expressing wild-type CSN-5 strains in the presence of CHX , whereas in the csn-5KO strain it became undetectable after 6 h of CHX treatment [16] . As expected , in the csn-5H127A strain , the detection of DCAF11 signal was extended to 9 h after CHX treatment ( Figure 6G and 6H ) . Taken together , these in vivo results indicate that the CSN complex containing mutant CSN-5 efficiently prevents degradation of substrate receptors of CRLs . Current models suggest that the activity and assembly of CRLs are controlled by cycles of CRL deneddylation and CAND1 binding of deneddylated cullins [39] , [40] , [41] , [42] , [43] , [44] , [45] , [46] . In plants and worms , CAND1 mutants exhibit defects consistent with a positive role in regulating the function of a subset of CRLs [40] , [47] , [48] , [49] . However , in yeast and human cells , loss of CAND1 has little effect on the abundance of neddylated cullins , suggesting that the neddylation/deneddylation cycle may function independently of CAND1 [50] , [51] . To test whether CAND1 is involved in maintaining the function of CRLs in N . crassa , we examined the role of CAND1 in the regulation of circadian conidiation rhythm and proper functioning of the SCFFWD-1 complex . We first measured the growth rates of the wild-type and cand1KO strains by race tube assay under constant darkness . The growth of the cand1KO strain ( about 3 . 0 cm per day ) was slightly slower than that of the wild-type strain ( about 3 . 7 cm per day ) , suggesting that CAND1 is involved in regulating hyphal growth . After entrainment by light , like the wild-type strain , the cand1KO strain exhibited a robust circadian conidiation rhythm with a period of about 22 h at 25°C in constant darkness ( Figure 7A ) , suggesting that CAND1 is not required for circadian rhythms in N . crassa . To test whether CAND1 functions in a manner similar to the CSN , we examined the conidiation rhythms of the cand1 mutant in LD cycles ( 12 h light/12 h dark ) . As shown in Figure 7B , the conidiation rhythms of the cand1KO strain were entrained by LD cycles , indicating that unlike CSN , CAND1 is not required in light regulation of the circadian clock . We also examined the responses of the cand1 mutant to temperature entrainment using race tube assays . As expected , in 12 h 27°C/12 h 22°C temperature cycles , as shown in Figure 7C , like the wild-type strain , the conidiation rhythm of the cand1KO strain was synchronized by the temperature cycles , indicating that CAND1 is not required for the temperature-entrained conidiation process . These results suggest that CAND1 does not play a significant role in the regulation of circadian rhythm in N . crassa . Deletion of cand1 also had no effect on degradation of the clock protein FRQ , which is the substrate of the SCFFWD-1 ubiquitin ligase complex in N . crassa ( Figure 8A and 8B ) . We also examined the stability of FWD-1 of the SCFFWD-1 complex in the cand1KO strain . As shown in Figure 8C and 8D , FWD-1 was very stable in the cand1 mutant , as in the wild-type strain , with a half-life of >12 h . Together , these results suggest that CAND1 is not required for regulation of the circadian rhythm and for maintaining the proper function of the SCFFWD-1 complex in N . crassa .
As the key regulator of CRLs , both deneddylation and maintenance of CRL stability by the CSN occurs when the CSN binds to CRLs . Thus , it is difficult to distinguish which function is more important for maintaining the proper function of CRLs in eukaryotes . To precisely determine the function of the CSN in maintaining the stability of CRLs , we sought to separate the two functional aspects of CSN from one other in N . crassa . In those csn-5KO strains expressing CSN-5 proteins with different point mutations in the JAMM metal-binding motif , the deneddylation activity was disrupted , while the assembly of the CSN complex and interactions between CSN and cullin proteins were not affected . Therefore , this system has great potential as a model for distinguishing between the two activities of the CSN . A recent study suggests that neddylated Cul1 and Cul3 are unstable in D . melanogaster csn mutant cells due to a defect in CSN deneddylation activity , whereas unneddylated cullins are stable in csn-5 mutant cells [33] . The results presented here show that unstable forms of Cul1 in the csn-5 mutant were partially restored by expression of mutant CSN-5 protein without deneddylation activity . Like the stability of unneddylated Cul1 and Cul3 in D . melanogaster CSN-5–defective cells [33] , unneddylated Cul1 remained stable in the csn-5H127A , csn-5H129A , and csn-5D140N strains , similar to that in the wild-type strain ( Figure 5A ) , indicating that CSN integrity with catalytically dead CSN-5 effectively maintains the stability of cullins . Studies in D . melanogaster and A . nidulans CSN-5 mutants indicated that the CSN deneddylation activity is essential for cell differentiation and developmental initiation [11] , [19] , [22] , [33] . However , in the A . thaliana fus6/C231 mutant ( a CSN1 N-terminal deletion mutant ) , although the Cul1 neddylation still works in a wild-type pattern , it was lethal and exhibited severe gene expression defects [52] . This genetic evidence raises questions concerning whether the CSN has other important functions aside from its deneddylation activity . The accelerated degradation of c-Jun in HeLa cells in which CSN-5 is downregulated is rescued equally by over-expression of the deneddylation mutant CSN-5D151N or wild-type CSN-5 [20] . These data suggest that two activities of CSN may function parallel for regulating the activity of CRLs . Bennett et al . found that Cul1K720R ( a constitutively unneddylated Cul1 mutant ) assembles with CSN , SKP-1 , and most F-box proteins to the same extent as wild-type Cul1 [51] . Our IP experiments also show that wild-type CSN interacts with both neddylated and unneddylated Cul4 . These findings suggest that the CSN can interact with CRLs independent of the prior neddylation of cullins . In plants , genetic results also suggest that during early embryo development and germination , neddylation/deneddylation cycling is not absolutely required , although it becomes more important during seedling establishment and later in development [21] , suggesting that the CSN has distinct biochemical functions that orchestrate development in the appropriate spatial and temporal setting . Protection of substrate receptors by the CSN has been described for the CRLs in vivo [24] , [25] , [26] , [27] , [28] . We found that CSN with mutated CSN-5 had a contribution to the stabilities of five receptor proteins of CRLs in vivo . These results provide evidence for the idea that the abundance of adaptor modules ( rather than cycles of neddylation/deneddylation and CAND1 binding ) drives CRL network organization [51] . This possibility is supported by our genetic observations that the csn-5H127A , csn-5H129A , and csn-5D140N strains exhibited normal growth and conidiation phenotypes . The integrity of CSN was maintained in these JAMM mutation complementation strains , thus it can serve as a platform to recruit other proteins for regulating the activities of CRLs , such as the recruitment of UBP12 in yeast , as well as USP15 in human [14] , [53] . In addition , a non-catalytic CSN itself may stabilize the substrate receptors of CRLs . A very recent study has shown that the protective effect of the CSN on DDB2 and CSA autoubiquitination is independent of CSN-5 mediated deneddylation in vitro [29] . These results suggest that the partial rescue of stability of substrate receptors by the catalytically dead CSN is mainly dependent on its protective effect . Therefore , the stability of cullins and some substrate receptors of CRLs are dependent on both deneddylation activity and integrity of the CSN in N . crassa . The csn-5KO strain exhibits abnormal conidiation rhythms in DD , which cannot be entrained by either LD or temperature cycles , indicating that light and temperature regulation of the conidiation process is impaired in this mutant [16] . We found that degradation of the clock protein FRQ is impaired in the csn-5KO strain , especially when protein synthesis is completely blocked . To further characterize the molecular mechanism of how the CSN regulates the conidiation rhythm , we focused on the SCFFWD-1 ubiquitin ligase , which controls the N . crassa circadian rhythm by ubiquitinating FRQ [36] . Our results demonstrated that defective FRQ degradation in the csn-5KO strain is due to the drastically reduced stability and levels of FWD-1 and Cul1 proteins in the SCFFWD-1 complex . Ectopic expression of mutant CSN-5 without deneddylation activity restored the defects of growth and conidiation in the csn-5KO strain , and almost restored the defects of the circadian conidiation rhythm in DD and FRQ degradation in the csn-5KO strain . Our data further showed that the low levels of FWD-1 in the csn-5KO strain were dramatically increased after expression of each of the CSN-5 proteins with point mutations in the JAMM metal-binding site , however , the increased stability and levels of the components in the SCFFWD-1 ubiquitin ligase are not enough to fully restore the degradation of FRQ to wild-type level , indicating that regulation of FRQ degradation plays a key role in maintaining the precise period length of conidiation rhythm in N . crassa . This is further supported by the finding that accelerated degradation of c-Jun in HeLa cells in which CSN-5 is downregulated can be rescued equally by over-expression of the deneddylation mutant CSN-5D151N or wild-type CSN-5; however , accelerated c-Jun degradation is not rescued in CSN-1– or CSN-3–downregulated cells by over-expression of wild-type CSN-5 [20] . Furthermore , the degradation of EB1 ( microtubule-end-binding protein 1 ) is accelerated by over-expression of wild-type CSN-5 or CSN-5D151N in HeLa cells [20] . These results suggest that the integrity of CSN might contribute more to regulating the stability of some substrates of CRLs . Current models suggest that the CRL complex is controlled by cycles of CRL deneddylation and CAND1 binding [7] . Our experiments further suggested that CAND1 , a putative regulator of CRLs , is not required for maintenance of SCFFWD-1 ubiquitin ligase activity and circadian rhythm in N . crassa . These data provide additional evidence that the CSN is an important regulator of the circadian clock in N . crassa through maintenance of SCFFWD-1 ubiquitin ligase stability . In conclusion , the results of our experiments indicate that even without deneddylation activity , the N . crassa CSN can still regulate hyphal growth , conidial development , and circadian function by regulating the activities of E3 ubiquitin ligases . Because the function of the CSN in the regulation of CRL activities is conserved in higher eukaryotes , we propose that the CSN may have a similar role in plants and animals .
The N . crassa strain 87-3 ( bd , a ) was used as the wild-type strain in this study . The bd ku70RIP strain , which was generated previously [54] , was used as the host strain for creating the cand1 knockout mutants . We also used csn-5KO , csn-2KO and csn-5KO , his-3 strains that were generated previously [16] . The 301-6 ( bd , his-3 , A ) strain and the csn-5KO , his-3 strain were used as the host strains for the his-3 targeting construct transformation [24] . Liquid culture conditions were the same as described previously [34] . For QA-induced protein expression , 0 . 01 M QA ( pH 5 . 8 ) was added to liquid medium containing 1× Vogel's medium , 0 . 1% glucose , and 0 . 17% arginine . The medium for the race tube assay contained 1× Vogel's , 0 . 1% glucose , 0 . 17% arginine , 50 ng/mL biotin , and 1 . 5% agar [55] . For race tubes containing QA ( 10−3 M ) , glucose was excluded from the medium . All three JAMM point mutations of CSN-5 were generated using the Quikchange Site-Directed Mutagenesis Kit ( Stratagene ) . pUC19-CSN-5 was used as the template for mutagenesis . Afterwards , the mutated CSN-5 DNA fragments were subcloned into either the pqa-5Myc-6His or pqa-3Flag vectors . The triple point mutant CSN-5 ( H127A , H129A and D140N ) generated from pUC19-CSN-5 was subcloned into the endogenous csn-5 promoter-driven vector pcsn-5-Myc-His-CSN-5 , resulting in pcsn-5-Myc-His-CSN-5tri . The previously constructed plasmids pqa-Myc-Cul1 , pqa-Myc-His-Cul3 , pqa-Myc-His-Cul4 , pqa-Myc-His-CSN-6 , pqa-Myc-His-SCON-2 , pqa-Myc-His-FBP94 , pqa-Myc-His-BTB1 , and pqa-Myc-His-DCAF11 were also used for his-3 targeting transformation in the csn-5KO , his-3 and 301-6 ( bd , his-3 , A ) strains [16] and cotransformation in the csn-5H127A , csn-5H129A , and csn-5D140N strains . GST-Cul4 ( containing Cul4 amino acids 1–113 ) fusion protein was expressed in RIL cells and the recombinant protein was purified and used as the antigen to generate rabbit polyclonal antiserum , as described previously [56] . The csn-5KO , Myc-His-CSN-5H127A , csn-5KO , Myc-His-CSN-5H129A , or csn-5KO , Myc-His-CSN-5D140N strain , wild-type strain ( negative control ) , and csn-5KO , Myc-His-CSN-5 strain ( positive control ) were cultured for approximately 24 h in constant light ( LL ) in liquid medium containing QA ( 0 . 01 M QA , 1× Vogel's medium , 0 . 1% glucose , and 0 . 17% arginine ) . Approximately 10 g of tissue from each strain grown in LL was harvested . The purification procedure was the same as described previously [16] . Fractions containing purified Myc-tagged CSN proteins were immunoprecipitated by adding 25 µL of c-Myc monoclonal antibody-coupled agarose beads ( 9E10AC , Santa Cruz Biotechnology ) . The precipitates of each sample were analyzed by SDS-PAGE ( 4%–20% acrylamide ) , which was subsequently silver stained following the manufacturer's instructions ( ProteoSilver Plus , Sigma ) . Specific bands in the Myc-His-CSN-5 purified products or in the Myc-His-CSN-5H127A purified products were excised and subjected to tryptic digestion and LC-MS/MS . The protocol of gel filtration chromatography was the same as described previously [16] , [21] . Briefly , purified proteins ( 400 µg ) were loaded onto a Superdex™ 200 ( GE ) gel filtration column that was equilibrated with 25 mL ( 150 mM NaCl , 20 mM Tris Cl pH 7 . 4 ) . The proteins were eluted in the same buffer at a flow rate of 0 . 3 mL/min . Fractions of 0 . 4 mL were collected starting from the onset of the column void volume ( 8 . 0 mL ) and finishing at 18 mL ( 25 fractions ) . 20 µL of each fraction were prepared in 20 µL of 2× SDS loading buffer , separated by 7 . 5% SDS-PAGE , and then examined by Western blot analysis using c-Myc antibody ( 9E10 , Santa Cruz Biotechnology ) . Protein extraction , quantification , western blot analysis , protein degradation assays , and immunoprecipitation assays were performed as described previously [24] , [56] . Western blot analyses using a monoclonal c-Myc antibody ( 9E10 , Santa Cruz Biotechnology ) or Flag antibody ( F3165-5MG , Sigma ) were performed to identify the positive transformants . Immunoprecipitates or equal amounts of total protein ( 40 µg ) were loaded into each protein lane for SDS-PAGE . After electrophoresis , proteins were transferred onto a PVDF membrane , and western blot analysis was performed using c-Myc antibody , Flag antibody , FWD-1 antiserum , FRQ antiserum , or Cul4 antiserum . | Cullin-RING E3 ubiquitin ligases ( CRLs ) play important roles in regulating a wide range of processes , such as signal transduction , transcription , cell cycle progression , circadian rhythm , and development , via the ubiquitin-proteasome pathway . The activity and stability of CRLs is precisely controlled by the COP9 signalosome ( CSN ) , an evolutionarily conserved multisubunit protein complex . Under the control of the CSN , CRL activity can be either downregulated via cleavage of NEDD8 ( an ubiquitin-like protein ) from cullin proteins ( deneddylation ) or preserved by maintaining the stability of CRL components . We generated point mutations of key residues in the JAMM domain of the CSN-5 subunit to disrupt CSN deneddylation activity , thereby creating a series of mutants containing the intact CSN complex but lacking deneddylation activity . Surprisingly , hyphal growth , conidial development , circadian rhythm , and stability of the SCFFWD-1 complex in these CSN-5 point mutants were comparable to that observed in wild-type N . crassa . Furthermore , we showed that CSN containing mutant CSN-5 efficiently prevents degradation of the substrate receptors of CRLs . Finally , deletion of the N . crassa ortholog of CAND1 ( cullin-associated NEDD8-dissociated protein 1 ) had little effect on conidial development and the circadian clock . Our results suggest that the integrity of the CSN is important for growth and development in N . crassa . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"developmental",
"biology",
"model",
"organisms",
"neurospora",
"crassa",
"genetics",
"yeast",
"and",
"fungal",
"models",
"biology",
"microbial",
"growth",
"and",
"development",
"microbiology",
"genetics",
"and",
"genomics",
"gene",
"function"
] | 2012 | Neurospora COP9 Signalosome Integrity Plays Major Roles for Hyphal Growth, Conidial Development, and Circadian Function |
To facilitate viral infection and spread , HIV-1 Nef disrupts the surface expression of the viral receptor ( CD4 ) and molecules capable of presenting HIV antigens to the immune system ( MHC-I ) . To accomplish this , Nef binds to the cytoplasmic tails of both molecules and then , by mechanisms that are not well understood , disrupts the trafficking of each molecule in different ways . Specifically , Nef promotes CD4 internalization after it has been transported to the cell surface , whereas Nef uses the clathrin adaptor , AP-1 , to disrupt normal transport of MHC-I from the TGN to the cell surface . Despite these differences in initial intracellular trafficking , we demonstrate that MHC-I and CD4 are ultimately found in the same Rab7+ vesicles and are both targeted for degradation via the activity of the Nef-interacting protein , β-COP . Moreover , we demonstrate that Nef contains two separable β-COP binding sites . One site , an arginine ( RXR ) motif in the N-terminal α helical domain of Nef , is necessary for maximal MHC-I degradation . The second site , composed of a di-acidic motif located in the C-terminal loop domain of Nef , is needed for efficient CD4 degradation . The requirement for redundant motifs with distinct roles supports a model in which Nef exists in multiple conformational states that allow access to different motifs , depending upon which cellular target is bound by Nef .
The HIV-1 accessory protein , Nef , affects the biology of the infected cell in several ways to achieve conditions optimal for viral replication and spread . Nef alters the intracellular trafficking of important immune molecules , such as class I and II major histocompatibility complex proteins ( MHC-I and MHC-II ) , CD4 , CD28 , and DC-SIGN [1]–[5] . Nef-dependent reduction of surface MHC-I protects HIV-infected primary T cells from recognition and killing by HIV-specific cytotoxic T lymphocytes ( CTLs ) in vitro [6] . Moreover , disruption of MHC-I expression by HIV-1 and SIV Nef provides a selective advantage under immune pressure in vivo [7]–[10] . CD4 downregulation by Nef is also essential for efficient viral spread . The rapid removal of CD4 prevents viral superinfection [11] , and enables optimal viral particle production by eliminating detrimental CD4/HIV envelope interactions in the infected cell [12] , [13] . Mutagenesis of protein-protein interaction domains has revealed that Nef uses genetically separable mechanisms to affect MHC-I and CD4 transport . Specifically , disruption of MHC-I surface expression requires an N-terminal α helix , a polyproline repeat , and an acidic domain in Nef [14] , [15] , while CD4 downregulation requires an intact dileucine motif , two diacidic motifs , and a hydrophobic pocket in Nef [15]–[18] . Amino acids necessary for the myristoylation [19] , [20] and oligomerization [21] of Nef are required for the disruption of both MHC-I and CD4 surface expression . Nef has the capacity to affect MHC-I transport at multiple subcellular locations; Nef blocks the export of newly-synthesized MHC-I from the secretory pathway and Nef expression results in a small increase in the rate of MHC-I internalization [22] . To accomplish this , Nef directly binds to the cytoplasmic tail of MHC-I early in the secretory pathway [23]–[26] . The Nef-MHC-I complex then actively recruits the clathrin adaptor protein complex AP-1 , which targets MHC-I from the TGN to the endo-lysosomal network where it is ultimately degraded [25] . Recruitment of AP-1 primarily requires a methionine at position 20 in the N-terminal α helical domain of Nef and a tyrosine residue in the cytoplasmic tail of MHC-I . Additionally , the acidic and polyproline domains of Nef have recently been shown to stabilize this interaction [27] , [28] . The normal function of AP-1 is to target proteins into the endosomal pathway and then recycle them back to the TGN . Thus , the AP-1 interaction with the Nef/MHC-I complex explains the targeting of MHC-I containing vesicles to the endosomal pathway and to the TGN . However , it does not explain accelerated degradation of MHC-I , hence other cellular factors may be involved [25] . The mechanism of Nef-induced CD4 internalization and degradation has been derived , in part , from correlating Nef function with the requirement for domains in the C-terminal flexible loop region of Nef that bind to cellular factors . The Nef dileucine motif ( ExxxLL165 ) is needed for CD4 internalization and it binds to adaptor protein complexes AP-1 , AP-2 , and AP-3 [16] , [29]–[36] . In addition , a diacidic motif , which is also required , enhances the interaction of Nef with AP-2 [37] . There is separate evidence that this diacidic motif may recruit the H subunit of the vacuolar ATPase ( V1H ) [38] to promote AP-2 recruitment [39] . Because the normal role of AP-2 is to link cargo to clathrin and promote internalization , it makes sense that this molecule would be necessary and indeed , the involvement of AP-2 has now been confirmed using RNAi knockdown in a number of cell systems [40]–[42] . After CD4 is internalized , it is targeted to lysosomes for degradation . There is evidence that this step requires β-COP [18] , a component of COP-1 coats implicated in endosomal trafficking as well as transport through the early secretory pathway [43]–[45] . Specifically , there are defects in the Nef-dependent transport of CD4 into acidified vesicles at the non-permissive temperature in cells harboring a temperature sensitive ε-COP mutant [18] . Nef directly interacts with β-COP [46] , and a second diacidic motif in the C-terminal loop domain of Nef has been demonstrated to mediate this interaction [18] , [47] , although , this result has not been reproducible by another group [48] . To more clearly understand the mechanism of altered MHC-I and CD4 trafficking observed in Nef-expressing cells , we directly compared these two processes in T cells that expressed Nef . We confirmed that Nef primarily affected MHC-I and CD4 at different subcellular locations and we demonstrated that the cytoplasmic tails of the respective molecules dictated which pathway was utilized . Despite the differences in initial trafficking , we found that HLA-A2 and CD4 co-localized in a discrete subset of vesicular structures . Upon further inspection , we determined that these structures also contained markers of late endosomes ( Rab7 ) and to a lesser extent , the lysosomal marker , LAMP-1 . Electron microscopy ( EM ) revealed that CD4 and HLA-A2 were found within MVBs of Nef-expressing T cells . HLA-A2 ( but not CD4 ) was also found in tubulovesicular structures adjacent to the Golgi . In Nef expressing cells , reduction of β-COP expression reduced the targeting of HLA-A2 from the TGN to LAMP-1+ compartments and stabilized CD4 expression within endosomal compartments . Finally , we identified two separate domains within Nef that were necessary for these activities and for β-COP binding . These data support a model in which both MHC-I and CD4 are ultimately targeted to the lysosomes in Nef expressing cells by a final common pathway .
It is known that Nef binds to the cytoplasmic tails of both CD4 and MHC–I , but that it affects them differently . To better understand the similarities and differences governing these two pathways , we examined the trafficking of CD4 , HLA-A2 and a chimeric molecule in which the wild type HLA-A2 cytoplasmic tail was substituted with the CD4 cytoplasmic tail ( HA-A2/CD4 ) . A flow cytometric analysis of steady state surface expression revealed that Nef dramatically reduced steady state surface expression of all three molecules ( Figure 1A ) . Consistent with prior studies , we found that CD4 was rapidly internalized from the cell surface in Nef expressing T cells , whereas wild type HLA-A2 was not ( Figure 1B ) . Substitution of the CD4 tail for the HLA-A2 cytoplasmic tail was sufficient to confer this phenotype ( Figure 1C ) . Conversely , prior studies have shown that Nef disrupts cell surface expression of MHC-I by blocking the transport of newly synthesized MHC-I from the TGN to the cell surface [22] , [23] . As shown in Figure 1D , Nef inhibited HLA-A2 forward transport by approximately 75% , whereas CD4 was unaffected at Nef levels that had a clear effect on HLA-A2 transport . Slight effects on CD4 could be observed at higher Nef levels ( Figure 1D , lane 8 ) . The substitution of the HLA-A2 cytoplasmic tail with the CD4 tail reduced the ability of Nef to disrupt forward trafficking ( Figure 1E ) . Thus , sequences in the cytoplasmic tails of CD4 and HLA-A2 determine how Nef disrupts their trafficking . To better understand the similarities and differences between MHC-I and CD4 trafficking in Nef-expressing cells , we compared the steady-state distribution of these molecules in T cells using confocal microscopy ( Figure 2A ) . We found that Nef expression caused the bulk of MHC-I to cluster in the perinuclear region where , in agreement with many other studies [14] , [30] , [49] , it co-localized with markers of the TGN ( data not shown ) . Interestingly , we also identified a subset of HLA-A2 that co-localized with CD4 in vesicular structures ( Figure 2A; arrows show example vesicles ) . To further identify these structures , we simultaneously stained for HLA-A2 , CD4 , and organelle markers using 3-color confocal microscopy ( summarized in Table S1 ) . Our results indicated that CD4 was mainly found in discrete vesicular structures , which also contained HLA-A2 ( 91 . 9% of the CD4+ vesicles co-localized with HLA-A2 , Table S1 ) and markers of late endosomes and lysosomes . Overall , the best marker for structures containing both HLA-A2 and CD4 was Rab7 ( 94% , of CD4+ vesicles co-localized with Rab 7 , Table S1 and Figure 2A , arrowheads mark example vesicles ) . CD4 and HLA-A2 were also found to co-localize with markers of lysosomes , such as LAMP-1 . However , the vesicles with the most intense LAMP-1 staining did not contain either HLA-A2 or CD4 , possibly because of degradation . Consistent with this , the co-localization of HLA-A2 and CD4 was dramatically increased when the cells were treated with bafilomycin , which inhibits degradation in acidic compartments ( Figure S1 ) . Thus , the normal steady-state co-localization of HLA-A2 and CD4 in Nef expressing cells was limited because degradation prevented accumulation in this compartment . To further discern these structures , we also examined them using electron microscopy ( EM ) . In agreement with the confocal data , our EM analysis revealed that compared with control cells in which both HLA-A2 and CD4 were found on the cell surface ( Figure 2B , panel 1 ) , in Nef-expressing T cells , the majority of CD4 was found in MVBs , co-localizing with HLA-A2 ( Figure 2B , panel 2 ) . In addition , we also noted substantial HLA-A2 , but not CD4 , accumulating in tubulovesicular structures adjacent to Golgi stacks ( Figure 2B , panel 3 ) . In separate experiments these structures were also found to contain AP-1 ( Figure 2C ) . Based on these studies , it appears that the majority of HLA-A2 resides in tubulovesicular structures in the region of the TGN with AP-1 , whereas at any given time , a small subset can be found in the endosomal compartment with CD4 . To further elucidate the similarities and differences between these pathways , we examined the role of known Nef-interacting proteins implicated in intracellular trafficking . AP-1 is a heterotetrameric adaptor protein involved in protein sorting from the TGN and it has been previously demonstrated to interact with MHC-I molecules in Nef expressing HIV-infected primary T cells and to direct MHC-I into the endolysosomal pathway [25] . Nef is also known to interact with β-COP [46] , a component of COP-1 vesicles also involved in endosomal trafficking [43]–[45] . Indeed , expression of wild type COP 1 components is needed for targeting CD4 into acidic vesicles in Nef-expressing cells [18] . To compare and contrast the requirement for these factors in Nef-dependent CD4 and HLA-A2 trafficking , we knocked down their expression using lentiviral vectors expressing short hairpin RNAs ( shRNAs ) [50] . All of these studies were performed in T cells and new cell lines were generated for each experiment to eliminate the possibility that long term growth in culture would select for cells that had compensated for the defect . Using this system , we obtained good knock down of the μ1 subunit of AP-1 and β-COP ( Figure 3A–C ) . ( A small apparent effect of shβ-COP on μ1 levels observable in Figure 3A was not significant when adjusted for protein loading in the experiment shown here or in replicate experiments [Figure 3B] . We also did not observe any effect of another siRNA directed against a different target site in β-COP on μ1 expression [Figure S2] . ) Because β-COP is known to be important for intra-Golgi and ER-to-Golgi trafficking , we asked whether the Golgi structure or MHC-I trafficking were drastically affected by reduced β-COP expression . We found that there was only a small reduction in the normal transport of MHC-I to the cell surface ( 35% reduction , Figure 3D ) . In addition , cells lacking β–COP generally maintained overall Golgi structure as assessed by the intracellular localization of giantin , a transmembrane protein normally residing in the cis and medial Golgi [51] ( Figure 3E ) . In contrast , brefeldin A , an inhibitor of an ARF1 GEF necessary for β-COP activity obliterated the normal Golgi staining ( Figure 3E , panel 9 ) . The relatively mild phenotype of this knock-down compared to the drastic effects of brefeldin A , suggests that brefeldin A has effects other than just disrupting COP 1 coats by blocking ARF1 activity . Having established that knocking down β-COP allowed relatively normal forward trafficking of HLA-A2 , we proceeded to assess the effect of knocking down β-COP or AP-1 in Nef-expressing cells . Consistent with previous publications [25] , we found that knocking down the ubiquitously expressed form of AP-1 ( AP-1A [52] ) largely reversed the effect of Nef on HLA-A2 ( p<10−4 ) , but had a smaller and less significant effect ( p<0 . 02 ) on CD4 surface expression ( Figure 4A and 4B ) . Surprisingly , we also observed that knocking down β-COP expression inhibited MHC-I downmodulation by Nef and had a small but statistically significant effect on CD4 downmodulation ( p<10−3; Figure 4A and 4B ) . The small effect of β-COP on CD4 surface expression indicated that β-COP was not necessary for CD4 internalization and downmodulation from the cell surface . However , further studies were needed to determine whether β-COP was required to degrade the CD4 after it was internalized . Prior studies had determined that expression of β-COP was necessary for acidification of CD4-containing vesicles and thus it was hypothesized that β-COP was needed to target vesicles containing internalized CD4 for lysosomal degradation . Therefore , we asked whether the role of β-COP in MHC-I trafficking was also to promote MHC-I degradation . To examine this , we utilized an assay we had developed , which measures the loss of mature , endo H–resistant HA-tagged HLA-A2 in Nef expressing cells by western blot analysis . This assay system is based on previous data demonstrating Nef-dependent degradation of the mature form of MHC-I in a manner that is reversible by inhibitors of lysosomal degradation [25] . As shown in Figure 4C , under normal , steady state conditions , most of the HLA-A2 is resistant to endo H digestion , indicating that it has matured through the Golgi apparatus ( Figure 4C , lane 2 ) . However , when Nef was expressed , we observed a dramatic reduction in total MHC-I and a decrease in the ratio of endo H resistant to sensitive protein ( Figure 4C compare lanes 2 and 18 , see also Figure S3 ) . Consistent with a role for AP-1 , we observed that AP-1A shRNA largely reversed this effect of Nef ( Figure 4C , compare lanes 18 and 20 . See also Figure 4D for quantification ) . To detect degradation of molecules containing a CD4 tail , we used HA-A2/CD4 ( Figure 1 ) and found that Nef expression accelerated the degradation of endo H resistant forms of this molecule ( Figure 4C , compare lanes 6 and 22 ) . However , we found that there was no effect of reduced AP-1A expression on Nef-dependent degradation of molecules containing the CD4 tail ( Figure 4C , compare lanes 22 and 24 . See also Figure 4D for quantification ) . When β-COP expression was reduced , we observed a small increase in the amount of immature , endo H–sensitive protein ( Figure 4C , compare lanes 10 and 12 ) , consistent with the 35% reduction in export of MHC-I to the cell surface shown in Figure 3D . However , we also noted that reduction in β-COP expression reduced the Nef-dependent degradation of the mature , endo H resistant form of these molecules ( Figure 4C , compare lanes 26 and 28 . See also Figure 4D for quantification ) implicating β-COP in this pathway . We were also able to confirm the model that β-COP is involved in Nef-dependent CD4 degradation as treating cells with β-COP shRNA reduced the degradation of the A2/CD4 chimeric molecule ( Figure 4C , compare lanes 30 and 32 . See also Figure 4D for quantification ) . We next directly examined the effect of reducing β-COP expression on Nef-dependent trafficking by confocal microscopy . For these experiments , cells were infected with HIV or were transduced with Nef-expressing adenoviral vectors and then the fate of internalized CD4 was assessed by confocal microscopy . Using this assay system , we observed fairly rapid internalization of CD4 in Nef-expressing cells , followed by loss of CD4 staining by 30 minutes ( Figure 5A , compare control cells in row 1 to Nef-expressing cells in row 3 ) . However , in T cells expressing β-COP shRNA , there was a three-to-four fold increase in the number of CD4-containing vesicles , consistent with a role for β-COP in promoting maturation of these vesicles into degradative compartments ( Figure 5A , compare control treated Nef-expressing cells in row 3 to shβ-COP–expressing cells in row 4 ) . Reduction of β-COP expression yielded similar results whether Nef was introduced using HIV infection or via adenoviral vectors ( Figure 5B and 5C ) . Confocal analysis of MHC-I intracellular localization revealed that expression of β-COP shRNA in control cells increased the intracellular accumulation of MHC-I , consistent with the slowing of export we observed in cells deficient in β-COP ( Figure 5D , compare rows 1 and 2 ) . Infection with Nef-expressing HIV resulted in the loss of cell surface MHC-I and an increase in intracellular MHC-I , some of which co-localized with LAMP-1 ( Figure 5D , compare rows 1 and 3 ) . Under these conditions , reduction of β-COP expression reduced the degree of colocalization with LAMP-1 ( Figure 5D , compare rows 3 and 4 ) . To enhance our ability to observe trafficking of MHC-I into LAMP-1+ compartments , we treated the cells with bafilomycin , which inhibits the vacuolar ATPase and thus acidification and degradation within lysosomal compartments . As previously reported [25] , bafilomycin treatment enhanced our ability to detect MHC-I in LAMP-1+ compartments in Nef-expressing T cells ( Figure 5D , compare rows 3 and 7 ) . The expression of β-COP shRNA decreased LAMP-1 colocalization with MHC-I , consistent with a role for β-COP in targeting MHC-I for degradation in lysosomal compartments in Nef expressing T cells ( Figure 5D , compare rows 7 and 8 ) . Similar results were observed whether Nef was introduced using HIV or adenoviral vectors ( Figure 5E and 5F ) . We also examined co-localization of HLA-A2 and CD4 in cells that expressed β-COP shRNA . We observed that reduction of β-COP expression resulted in increased staining of both proteins , and did not disrupt their co-localization ( Figure S4 ) . Thus , β-COP was not necessary for targeting these proteins into a common endosomal pathway , but rather was needed for their subsequent targeting into a degradative pathway . To further explore the molecular mechanism for the similarities and differences in MHC-I and CD4 trafficking in Nef-expressing T cells , we asked whether these molecules differed as to how well they bound Nef or cellular factors . As expected , we found that HIV Nef bound to both the HLA-A2 and the CD4 tail ( Figure 6A , right panel ) . However , AP-1 only co-precipitated with molecules containing the HLA-A2 cytoplasmic tail ( Figure 6A , right panel ) . The chimeric molecule with the CD4 cytoplasmic tail did not bind AP-1 in Nef-expressing T cells ( Figure 6A , right panel ) . In these experiments , we noted that the expression level of A2/CD4 was lower than for wild type HLA-A2 , which could explain this difference . Therefore , we confirmed these data using a fusion protein containing either HLA-A2 or A2/CD4 directly fused to full length HIV-Nef protein . In previously published experiments it was shown that the HLA-A2/Nef fusion protein co-precipitated AP-1 in a manner that depended on sequences both in Nef and in the HLA-A2 cytoplasmic tail [25] . Here we show again that the HLA-A2 cytoplasmic tail was necessary for this interaction and , moreover , that the CD4 tail could not substitute for it ( Figure 6B , right panel ) . The Nef-β-COP interaction is well-described in the literature [46] and there is evidence that β-COP interacts with a diacidic motif ( E154/155 ) within the Nef C-terminal loop [18] . However , this region of Nef has never been implicated in MHC-I trafficking . To provide further evidence that β-COP is needed to promote MHC-I degradation , we sought to identify a region of Nef that is needed both for MHC-I degradation as well as β-COP binding . We therefore examined a panel of mutations ( M20A , V10EΔ17–26 and E62–65Q ) that are specifically defective at disrupting MHC-I trafficking [14] , [15] , [26] , [53] . We also examined a Nef mutant , D123G , that is defective at both CD4 and MHC-I downmodulation [21] . The relative activity of these Nef mutants in MHC-I and CD4 downmodulation is shown in Figure 7A and quantified in Figure 7B . We then examined the relative ability of each of these mutant molecules to co-precipitate with β–COP . As shown in Figure 7C , we found that the V10E Δ17–26-Nef , which is defective at MHC-I downmodulation , was also defective at binding to β-COP ( compare lanes 3 and 5 ) . Interestingly , this deletion mutant is also defective at interacting with AP-1 [25] . However , the β-COP binding site was separable from the AP-1 interaction site because M20 , which is located within the deleted region , is needed for AP-1 interaction [25] , [27] ) , but was not necessary for β-COP binding to Nef ( Figure 7C , compare lanes 3 and 4 ) . Mutation of the Nef dimerization motif [D123G , [21]] , which disrupts a number of Nef functions , including MHC-I and CD4 downmodulation , also reduced binding to β-COP ( Figure 7C , compare lanes 3 and 7 ) . Finally , mutation of the Nef acidic domain ( E62–65Q ) , which disrupts binding to MHC-I [26] , AP-1 [27] , [28] and PACS-1 [54] , did not affect binding to β-COP ( Figure 7 , compare lanes 3 and 6 ) . As expected , we found that V10EΔ17–26 Nef , which was defective at β-COP binding , was also defective at inducing the degradation of the endo H resistant form of HLA-A2 ( Figure 7D , upper panel , compare lanes 3 and 4 with lanes 5 and 6 ) . In contrast , V10EΔ17–26 Nef was not defective at A2/CD4 degradation based on western blot analysis ( Figure 7D , lower panel , compare lanes 3 and 4 with lanes 5 and 6 ) . These data suggested that there may be another interaction domain that recruits β-COP to the Nef-CD4 complex to promote CD4 degradation . This would be consistent with the faint band observable in the V10EΔ17–26-Nef mutant immunoprecipitation ( Figure 7C , lane 5 , longer exposure ) and prior publications demonstrating that mutation of E154/155 also affected β-COP binding [47] . Thus , there may be two independent binding sites for β-COP within Nef , each of which governs the degradation of a different cellular factor . To further define the β-COP binding site , and to determine whether there were indeed two β-COP binding sites , we constructed additional Nef mutants . We focused on the arginine residues ( R17ER19MR21R22 ) within the Nef deletion Δ17–26 ) because previous studies had indicated that arginine rich regions could form β-COP-binding sites [55] . Flow cytometric analysis of MHC-I levels on cells expressing these mutants revealed that the R17/19 pair was necessary for maximal MHC-I downmodulation ( Figure 8A and 8B ) . In contrast , mutation of R21/22 did not significantly affect MHC-I downmodulation ( unpublished data ) . An assessment of Nef-induced degradation by pulse chase analysis of HA-HLA-A2 , revealed that mutating this motif also inhibited Nef-dependent degradation ( Figure 8C , compare lanes 5 and 7 , quantified in Figure 8D ) . Additionally , mutation of R17/19 reduced , but did not eliminate binding of β-COP to Nef in a manner similar to the effect of the Δ17–26 Nef mutation ( Figure 8E , compare lanes 3 and 4 ) . We next examined the diacidic motif ( E154/155 ) previously implicated in β-COP binding . As shown in Figure 8A and 8B , mutation of this motif did not disrupt MHC-I downmodulation , in fact downmodulation was somewhat enhanced . Additionally , we found that mutation of this motif did not reduce MHC-I degradation ( Figure 8C , compare lanes 5 and 11 , see also quantification in 8D ) . However , in agreement with prior results , we observed a partial defect in β-COP binding with this mutant ( Figure 8E , compare lanes 3 and 6 , [18] , [47] . However , this defect was less reproducible ( observed in two out of four experiments ) than that observed with disruption of R17/19 ( consistently observed in five out of five experiments ) , suggesting that binding to R17/19 can mask the defect observed with mutation of E154/155 under certain conditions . To provide additional data supporting the possibility that both sites contributed to β-COP binding , we constructed a double mutant , R17/19 A and E154/155A ( R/E ) . As shown in Figure 8E , lane 5 , binding of R/E to β-COP was further reduced relative to binding of Nef proteins containing single mutations in each motif , strongly implicating both motifs in β-COP binding . The phenotype of the double mutant was highly reproducible in 5 out of 5 experiments . Interestingly , the R/E double mutant was not more defective than R17/19A at downmodulating MHC-I ( Figures 8A and 8B ) or at promoting MHC-I degradation ( Figure 8C , compare lanes 7 and 9 , quantified in 8D ) , indicating that Nef did not utilize the E154/155 binding site to recruit β-COP for MHC-I degradation . Conversely , we confirmed prior reports that the E154/155A mutant was defective at CD4 degradation ( Figure 9A , compare lanes 3 and 6 ) and determined moreover that there was no significant effect of mutating R17/19 on CD4 degradation , either alone or in combination with E154/155A ( Figure 9A , compare lanes 3 and 4 ) . It is also worth noting that , in contrast to what was observed with HLA-A2 , we did not observe a clear correlation between the relative CD4 surface expression and the relative level of total cellular CD4 ( compare Figure 8B and 9B ) , indicating that there was a complex relationship between total cellular CD4 and the fraction expressed on the cell surface . Because the R17/19 motif is directly adjacent to M20 , which is necessary for AP-1 recruitment [25] , [27] , we also examined whether these mutations , which affect β-COP binding , also disrupted AP-1 co-precipitation . To accomplish this , we used our standard AP-1 recruitment assay in which proteins co-precipitating with MHC-I HLA-A2 were detected by western blot analysis . As shown in Figure S5 , mutation of R17 , 19 ( and E154/155 ) decreased AP-1 binding only slightly . Thus , the defects in MHC-I downmodulation and degradation noted with mutation of R17 , 19 resulted primarily from defects in β-COP binding .
Expression of HIV Nef in infected cells protects them from lysis by CTLs and this activity of Nef is due to downmodulation of MHC-I surface expression . The Nef protein also prevents superinfection and promotes viral spread by removing the viral receptor , CD4 from the cell surface ( for review see [56] ) . We provide evidence that sequences in the cytoplasmic tail of these molecules are important for determining whether Nef disrupts their trafficking from the cell surface or at the TGN . These data , that swapping cytoplasmic domains switches the initial pathways taken by HLA-A2 and CD4 in the presence of Nef , may seem somewhat obvious . Nef is always the same and thus one might conclude that this information has to be contained in the modulated protein . However , it was also possible that the ectodomain affected Nef responsiveness by binding to other transmembrane proteins or by altering intracellular trafficking . This was certainly a possibility for MHC-I for which it is clear that the efficiency of peptide loading can affect trafficking and we have found that trafficking rates affect responsiveness to Nef and AP-1 binding [23] . Prior studies have demonstrated that Nef initially binds to hypo-phosphorylated forms of the MHC-I cytoplasmic tail early in the secretory compartment [23] , but binding does not affect normal transit through the Golgi apparatus and into the TGN [25] . The Nef-MHC-I complex then recruits the AP-1 heterotetrameric clathrin adaptor protein using a binding site that is created when Nef binds the MHC-I cytoplasmic tail . This binding site requires a methionine from the N-terminal α helix of Nef and a tyrosine residue in the MHC-I cytoplasmic tail [25] . Additionally , there is evidence that this complex is stabilized by the acidic and polyproline domains of Nef [27] , [28] . Formation of this complex results in the re-direction of MHC-I trafficking in such a way that it is targeted to lysosomes for degradation [25] . However , cellular proteins that normally bind AP-1 are not degraded , but rather recycled to the TGN ( Figure 9C ) . Here we present new evidence that Nef utilizes β-COP to promote trafficking to degradative compartments ( Figure 9C ) . Knocking down expression of β-COP inhibited the degradation of MHC-I and it did so by blocking the transport of MHC-I from intracellular vesicles to LAMP-1+ compartments . We also provide results here that confirm β-COP is necessary for degradation of CD4 in lysosomal compartments . Thus , we propose that AP-1 and AP-2 deliver MHC-I and CD4 respectively to endosomal compartments where β-COP displaces AP-1 and AP-2 to target MHC-I and CD4 for lysosomal degradation ( Figure 9C ) . As described above , we found that knocking down β-COP with shRNA resulted in stabilization of internalized CD4 , however the effect on CD4 surface expression was small , but still significant . In contrast , there was a greater effect of β-COP knockdown on HLA-A2 surface expression . This might suggest that the role of β-COP in the modulation of these targets was different , rather than the same . However , this apparent paradox can be explained by our model shown in Figure 9C . As indicated , differences in response to β-COP knockdown can be explained by differences in the intracellular pathways of these proteins before they interact with β-COP . MHC-I is engaged in an AP-1-dependent endosome-to-TGN loop , and MHC-I could “leak” out to the cell surface from the TGN in the absence of β-COP , whereas CD4 may be unable to return to the cell surface from its endosomal compartment . Consistent with this , we also noted a lack of correlation between degradation and surface expression of CD4 ( but not MHC-I ) when Nef mutants that were defective in β-COP binding were examined . These data indicate that there is a complex relationship between total cellular CD4 and the fraction that is present on the cell surface and thus intracellular pools need to be directly examined to assess degradation rather than relying on surface expression as an indicator of the efficiency of this process . It is also noteworthy that shRNA knockdown of β-COP did not fully reverse Nef-dependent MHC-I and CD4 degradation . This may have resulted from incomplete knockdown of β-COP . However , we also observed a similar phenotype with Nef mutants defective at β-COP binding . Failure to fully reverse degradation may be secondary to a default degradative pathway that exists for all proteins delivered to endosomal pathways . Alternatively , there may be other ways Nef targets these proteins to lysosomes , which have yet to be identified . Our studies indicate that there are at least three domains needed for Nef to interact efficiently with β-COP . One of these domains ( D123 ) , is required for dimerization of Nef and is needed to affect a variety of Nef functions [21] . Another region lies within the N-terminal α helical domain of Nef that is specifically required for disruption of MHC-I trafficking and for interactions with AP- 1 [25] . This binding site for β-COP is distinct from that used by AP-1 , because recruitment of β-COP does not require Nef's acidic domain or Nef M20 , whereas AP-1 does [25] , [27] . The fact that these Nef mutants bind β-COP , but are still defective at MHC-downmodulation [53] makes sense , because these mutants are also unable to bind the MHC-I cytoplasmic tail [26] . Additional mutants , which focused on the highly conserved stretch of arginines in the N-terminal alpha helical domain of Nef ( R17XRMRR22 ) , revealed that the regions involved in AP-1 and β-COP binding were very closely apposed . However , we determined that mutation of R17/19 affected primarily β-COP binding , with only a minimal effect on AP-1 interaction . Thus , these two Nef-interacting proteins have distinct and separable amino acid requirements for binding . The identification of a β-COP binding domain within a region of Nef that is also required for Nef to accelerate MHC-I degradation confirms the requirement for β-COP in this pathway . In addition , the residual binding of β-COP to these Nef mutants provided suggestive data that another binding site for β-COP existed . Indeed , we were able to confirm prior evidence that a diacidic motif within the C-terminal loop of Nef also promoted an interaction with β-COP and that mutation of this motif reduced CD4 degradation [47] . Finally , we demonstrated that mutation of both the RXR and the diacidic motifs resulted in the greatest defect in β-COP binding . The double mutant did not however result in a greater defect in either MHC-I or CD4 degradation , indicating the role of each motif is distinct and not additive . The discovery of two distinct β-COP binding motifs helps explain why some groups could not confirm the role of the diacidic motif in β-COP binding [48] as both motifs need to be mutated to reliably eliminate an interaction between β-COP and Nef . There is precedent for such redundancy . For example , there are two AP-1 binding sites within Nef; a dileucine motif within the C-terminal flexible loop [16] , [31] , [32] , [33] as well as a second site that forms upon binding of Nef to the MHC-I cytoplasmic tail . Despite the presence of two AP-1 signals , only one is active in the context of the natural Nef-MHC-I complex [25] , [27] . The dileucine motif in the C-terminal flexible loop can become activated to affect MHC-I transport , but only when Nef is artificially fused to the MHC-I cytoplasmic tail [27] . This result indicates there is no inherent inability of this signal to affect MHC-I traffic but rather that something else , such as the structure of the natural complex , causes the dileucine motif to be inactive [27] . The dileucine motif at position 164 is located close to the diacidic motif at position 154 that binds β-COP to promote CD4 degradation . The fact that both of these motifs are inactive when Nef is bound to MHC-I , suggests that much of the C-terminal flexible loop region of Nef is inaccessible under these conditions . Thus , Nef behaves as though it assumes different structural forms in different contexts to differentially expose distinct trafficking signals . We also present evidence that knock down of β-COP yielded a distinct phenotype from BFA treatment . As described above , BFA is a chemical inhibitor of ARF1 , that is known to trigger the reversible collapse of the cis-medial Golgi compartments to the ER [57]–[59] by inhibiting an ARF-specific guanine nucleotide-exchange protein ( ARF-GEF ) [60] , [61] . Because ARF1 activity is necessary for recruitment of β-COP to membranes [62] , it was possible that the dramatic effects of BFA resulted from the inability for β-COP to function normally . However , our results demonstrating that knockdown of β-COP had no effect on overall Golgi structure indicate that the dramatic effects of BFA are not due solely to disruption of β-COP function in the Golgi . Given the important role of β-COP in the Golgi , it is surprising that β-COP bound to Nef does not also affect transport of MHC-I through the ER/Golgi . It is possible that our inability to detect an effect of Nef on early transport of MHC-I [25] may be a result of the cell type chosen for these studies . T cells , which are an important natural target of HIV , normally traffic MHC-I through the early secretory pathway slowly [23] and thus it might be difficult to further reduce the trafficking speed through an interaction with β-COP . Interestingly , another group has reported a reduced ER-Golgi exit rate for MHC-I in Nef-expressing HeLa cells [63] , which normally transport MHC-I more rapidly than T cells [23] . We have made similar observations in astrocytoma cells expressing higher levels of Nef than typically needed to observe MHC-I downmodulation ( Roeth and Collins , unpublished observations ) . Further studies will be needed to determine whether this effect of Nef plays a role in more physiologically relevant cell systems and whether this effect of Nef might be dependent on β-COP expression . A recent report indicates that the effect of Nef on internalization of MHC-I , which is only minimally apparent in our system , occurs via a PI3-kinase dependent pathway [64] . This publication reported that CEM cells , which were used in our study , have less PTEN ( a phosphatase that inhibits PI3-kinase ) than another T cell line used in their study ( H9 ) . This deficiency might make it relatively more difficult for us to detect an effect of chemical PI3-kinase inhibitors , but would not affect our ability to detect a PI3-kinase-dependent trafficking pathway . In fact , one would expect the opposite , that the PI3-kinase-dependent pathway would be more active in our system . However , we have found that Nef has a relatively small effect on internalization of MHC-I , and mainly affects MHC-I protein export and degradation . These data have been corroborated in HIV-infected primary T cells [22] , [26] , which were also found to much lower levels of PTEN than H9 cells did [64] . From a teleological perspective , it makes sense that Nef would have evolved to target early forms of MHC-I , which harbor antigens derived from the newly synthesized viral proteins . Older forms of MHC-I already on the cell surface would be bound to normal cellular antigens and would in fact be protective as they would inhibit killing by natural killer cells that are stimulated to lyse cells with abnormally low MHC-I expression . On the other hand , it makes sense that Nef , an early viral protein , would have evolved to target surface CD4 to rapidly and efficiently remove CD4 in order to prepare the cell for rapid release of viral particles and to render the cell resistant to re-infection . Meanwhile , a late protein , Vpu , is expressed in infected cells and specifically targets the newly synthesized CD4 for degradation , preventing any additional CD4 from reaching the cell surface [65] . In sum , we have found that the HIV Nef protein commandeers the cellular trafficking machinery efficiently by utilizing their natural activities for abnormal purposes . The fact that these pathways may end in a final common step raises the important possibility that inhibitors might be developed that could block multiple Nef functions .
CEM T cells stably expressing HA-tagged HLA-A2 ( CEM HA-HLA-A2 ) have already been described [25] . Cell lines stably expressing YFP-tagged Rab7 or HA-HLA-A2/CD4 were made by transducing cells with murine retroviral constructs ( MSCV YFP-Rab7 or MSCV HA-A2/CD4 ) as previously described [22] , followed by culture in selective media . MSCV YFP-Rab7 was constructed by cloning a filled-in a Kpn I-Xho I fragment from pEYFP-Rab7 [66] into MSCV puro [67] . MSCV HA-A2/CD4 was constructed using PCR mutagenesis . The first round PCR produced two products: the first utilized 5′ primer ( primer 1 ) 5′-CGGGATCCACCATGCGGGTCACGGCG-3′ and 3′ primer ( primer 2 ) 5′-CTCTGCTTGGCGCCTTCGGTGCCACATCACAGCAGCGACCAC-3′ with MSCV HA-HLA-A2 as the template [25] . The second utilized 5′ primer ( primer 3 ) 5′-GTGGTCGCTGCTGTGATGTGGCACCGAAGGCGCCAAGCAGAG-3′ and 3′ primer ( primer 4 ) 5′-CCTCGAGTCAAATGGGGCTACATGTCTTCTGAAATCGGTGAGGGCACTGG-3′ using CD4 as the template . The second round utilized primers 1 and 4 from the previous PCR reactions plus 1 µl of each purified first round PCR reactions as template . The resulting product was digested with BamHI and XhoI and ligated into MSCV 2 . 2 [67] digested with BglII and XhoI . MSCV A2/Nef has been described [26] . MSCV HA-A2/CD4/Nef was constructed using a PCR mutagenesis approach . The first round PCR produced two products: the first utilized 5′ primer ( primer 1 ) 5′-CGGGATCCACCATGCGGGTCACGGCG-3′ and 3′ primer ( primer 2 ) 5′-CCACTTGCCACCCATACTAGTAATGGGGCTACATGT-3′ with MSCV HA-A2/CD4 as the template . The second utilized 5′ primer ( primer 3 ) 5′-ACATGTAGCCCCATTACTATGATGGGTGGCAAGTGG-3′ and 3′ primer ( primer 4 ) 5′- GCGAATTCTCAGCAGTTCTTGAAGTACTC-3′ with NL4-3 Nef open reading frame as template . The second round utilized primers 1 and 4 from the previous PCR reactions plus 1 µl of each purified first round PCR reactions as template . The resulting product was digested with BamHI and EcoRI and ligated into MSCV IRES GFP [68] digested with BglII and EcoRI . Nef mutants were made by using the PCR mutagenesis approach described previously ( Wonderlich et al . 2008 ) . The mutagenesis primers were as follows: R17/19A 5′-TGGCCTACTGTAGCGGAAGCAATGAGACGAGCT-3′ and EE154–155AA 5′-GTTGAGCCAGATAAGGTAGCAGCGGCCAATAAAGGAGAGA-3′ . Each primer , plus its reverse complement were utilized together with additional 5′ and 3′ primers to generate the mutated product . Wild type NL4-3 Nef [MSCV A2/Nef IRES GFP ( Roeth et al 2005 ) ] was used as a template for the PCR reaction , except for the double mutant , R17/19A/EE154–155AA , in which the MSCV R17/19A Nef IRES GFP was used as the template . Each mutated PCR product was digested and cloned into MSCV IRES GFP [68] as described previously ( Wonderlich et al . 2008 ) . The FG12 shRNA lentiviral vectors were constructed as previously described [50] . Briefly , complementary primers were annealed together and ligated into vector pRNAi [69] digested with BglII and HindIII . The sequences of the primers were as follows ( the target sequence is underlined ) : shNC ( an siRNA directed at GFP , with several base changes [25] ) - sense 5′-GATCCCCGCTCACACTGAAGTTAATCTTCAAGAGAGATTAACTTCAGTGTGAGCTTTTTGGAAA-3′ , antisense 5′-AGCTTTTCCAAAAAGCTCACACTGAAGTTAATCTCTCTTGAAGATTAACTTCAGTGTGAGCGGG-3′ , shβ-COP- sense 5′-GATCCCCTGAGAAGGATGCAAGTTGCTTCAAGAGAGCAACTTGCATCCTTCTCATTTTTGGAAA-3′ , antisense 5′-AGCTTTTCCAAAAATGAGAAGGATGCAAGTTGCTCTCTTGAAGCAACTTGCATCCTTCTCAGGG-3′; shμ1A- ( a mixture of two lentiviruses was used ) ( 1 ) sense 5′GATCCCCTGAGGTGTTCTTGGACGTCTTCAAGAGAGACGTCCAAGAACACCTCATTTTTGGAAA-3′ , antisense 5′-AGCTTTTCCAAAAATGAGGTGTTCTTGGACGTCTCTCTTGAAGACGTCCAAGAACACCTCAGGG-3′ , ( 2 ) sense 5′- GATCCCCCGACAAGGTCCTCTTTGACTTCAAGAGAGTCAAAGAGGACCTTGTCGTTTTTGGAAA-3′ , and antisense 5′- AGCTTTTCCAAAAACGACAAGGTCCTCTTTGACTCTCTTGAAGTCAAAGAGGACCTTGTCGGGG-3′ . The pRNAi constructs were digested with XbaI and XhoI to remove the promoter and shRNA sequence . The resulting fragment was ligated into FG12 [50] , digested with XbaI and XhoI . Adenovirus was prepared by the University of Michigan Gene Vector Core facility . Adenoviral and HIV ( HXB-EP [6] ) transductions of T cells [25] or 373 mg astrocytoma cells [49] have been described previously . Murine retroviral vector ( MSCV ) expressing Nef was prepared as described previously ( Roeth et al . 2005 ) , except that in some cases the retroviral vector supernatants were concentrated by spinning at 14000 RPM for four hours at 4°C . The viral pellet was then resuspended in media to yield a twenty-fold concentrated stock . Lentiviruses expressing shRNA were generated using an approach similar to that already described [50] . Briefly , 293 cells were transfected with the FG12 constructs described above plus pRRE [70] , pRSV-Rev [70] and pHCMV-G [71] using Lipofectamine 2000 ( Invitrogen ) . Supernatants from the transfected cells were collected and used to transduce CEM T cells using a spin-transduction protocol . Intact cells were stained for flow cytometry analysis as previously described [24] . Briefly , HLA-A2 was detected with BB7 . 2 [72] that had been purified as previously described [22] . Endogenous CD4 was detected using RPA-T4 from Serotec . The secondary antibody was goat anti-mouse-phycoerythrin ( BioSource , 1∶250 ) . For experiments using the GFP-expressing FG12 lentivirus for shRNA expression , the GFP-positive cells were gated to identify the subset of transduced cells ( generally >90% of cells ) . Endocytosis assays were performed as previously described with minor modification [22] . Briefly , cells were washed once with Endocytosis Buffer [D-PBS , 10 mM HEPES , 10 µg/ml BSA ( NEB ) ] , then stained with primary antibody ( described above ) for 20 minutes on ice . After washing , the cells were resuspended in RPMI supplemented with 10% fetal bovine serum , 10 mM HEPES buffer , 2 mM L-glutamine , penicillin and streptomycin ( R10 ) ( pre-warmed to 37°C ) and replicate aliquots were removed and placed on ice for each time point . Cells were then washed and stained with goat anti-mouse-phycoerythrin ( BioSource , 1∶250 ) and the samples were analyzed using a FACScan flow cytometer ( Becton Dickinson ) . Flow cytometry data was processed using FlowJo v4 . 4 . 3 software ( Treestar Corp . ) . The mean fluorescence at time zero was set to 100% , and this value was used to calculate the relative surface staining at each subsequent time point . CEM cells transduced with adenoviral vectors as previously described [22] were first incubated in pre-label media [RPMI –Cys –Met ( Specialty Media , Inc . ) +10% dialyzed FBS ( Invitrogen ) ] for 15 minutes at 37°C . Pulse labeling was performed in pre-label media with 150–200 µCi/ml Pro-mix-L [35S] ( >1000 Ci/mmol; Amersham Pharmacia ) for 30 minutes at 37°C . The cells were then chased in R10 media for 15 minutes at 37°C , followed by two washes with D-PBS . To label the protein that reached the cell surface , the cells were resuspended in D-PBS containing 0 . 5 mg/ml EZ-Link sulfo-NHS-LC-Biotin ( Pierce ) , and incubated at 37°C for 1 hour . Surface biotinylation was quenched by washing the cells in D-PBS+25 mM Lysine ( Fisher ) . For Figure 1D , immunoprecipitation of proteins from cell lysates was performed as previously described [25] , except that one-third of the total lysate was used for the HLA-A2 immunoprecipitation while two-thirds of the material was used to recover CD4 . For immunoprecipitations of 35S labeled proteins , 5 µg of BB7 . 2 and 2 . 5 µg RPTA4 ( BD Pharmingen ) were used for HLA-A2 and CD4 respectively . In Figure 1E and 3D , the total cell lysate was immunoprecipitated with anti-HA ascites ( HA . 11 , Covance ) . For Figures 1D , 1E and 3D , recovered proteins were released from the beads by boiling in 100 µl of 10% SDS . One third was analyzed directly by SDS-PAGE ( Total ) . The remaining two thirds was brought to a total volume of 1 ml in RIPA Buffer [25] , and 40 µl of avidin-agarose ( Calbiochem ) was added to recover biotinylated proteins . After 2 hours at 4°C , the beads were washed three times with 1 ml RIPA buffer and proteins were separated by SDS-PAGE ( Surface ) . Adeno-transduced CEM cells were adhered to glass slides , fixed , permeabilized , and stained for indirect immunofluorescence as previously described [25] . Bafilomycin treatment was performed as described previously [25] . The following antibodies were utilized to localize proteins via microscopy: Figure 2 , and Figures S1 and S4; anti-CD4 ( S3 . 5 , Caltag Laboratories ) and anti-HLA-A2 ( BB7 . 2 ) ; Figure 3; anti-giantin ( Covance ) ; Figure 5; anti-CD4 antibody ( S3 . 5 , Caltag Laboratories ) , anti-LAMP-1 ( H4A3 , BD Pharmingen ) and anti-HLA-A2 ( BB7 . 2 ) . Secondary antibodies were obtained from Molecular Probes and were used at a dilution of 1∶250: Giantin , Alexa Fluor 546 goat anti-rabbit; CD4 , Alexa Fluor 546 goat anti-mouse IgG2a; LAMP-1 , Alexa Fluor 546 goat anti-mouse IgG1; BB7 . 2 ( Figures 2 , 5D and S4 ) , Alexa Fluor 647 goat anti-mouse IgG2b; BB7 . 2 ( Figure S1 ) , Alexa Fluor 488 goat anti-mouse IgG2b . See Table S2 for a summary of antibodies used to gather data for Table S1 . For the microscopy based internalization assay in Figure 5A , CEM T cells were allowed to adhere to glass slides , and placed on ice . The cells were washed once with wash buffer ( D-PBS , 10 µg/ml BSA ( NEB ) and 2% goat serum ) , incubated with anti-CD4 antibody ( S3 . 5 , Caltag Laboratories , IF , 1∶25 ) for 20 minutes , washed once with wash buffer , incubated with Alexa fluor 546 goat anti-mouse IgG2a ( Molecular Probes , 1∶250 ) for 20 minutes and washed once with wash buffer . The zero time point was fixed with 2% paraformaldehyde , while the remaining time points incubated at 37˚C for the indicated time . The cells were then fixed with 2% paraformaldehyde . Images were collected using a Zeiss LSM 510 confocal microscope and processed using Adobe Photoshop software . Three-dimensional projections of cells were generated from Z-stacks using Zeiss LSM Image Examiner software . Otherwise , single Z sections through the center of the cell were displayed . Electron microscopy with CEM cells transduced with adenovirus was performed by the Harvard Medical School ( HMS ) Electron Microscopy Facility . Frozen samples were sectioned at −120°C , the sections were transferred to formvar-carbon coated copper grids and floated on PBS until the immunogold labeling was carried out . The gold labeling was carried out at room temperature on a piece of parafilm . All antibodies and protein A gold were diluted in 1% BSA . The diluted antibody solution was centrifuged 1 minute at 14 , 000 rpm prior to labeling to avoid possible aggregates . Grids were floated on drops of 1% BSA for 10 minutes to block for unspecific labeling , transferred to 5 µl drops of primary antibody and incubated for 30 minutes . The grids were then washed in 4 µl drops of PBS for a total of 15 minutes , transferred to 5 µl drops of Protein-A gold for 20 minutes , washed in 4 µl drops of PBS for 15 minutes and 6 µl drops of double distilled water . Contrasting/embedding of the labeled grids was carried out on ice in 0 . 3% uranyl acetete in 2% methyl cellulose for 10 minutes . Grids were picked up with metal loops ( diameter slightly larger than the grid ) and the excess liquid was removed by streaking on a filter paper ( Whatman #1 ) , leaving a thin coat of methyl cellulose ( bluish interference color when dry ) . The grids were examined in a Tecnai G2 Spirit BioTWIN transmission electron microscope and images were recorded with an AMT 2k CCD camera . For the western blot analysis in Figures 3A , 4C , 7D , 9A , S2 , and S3 , cells were lysed in PBS 0 . 3% CHAPS , 0 . 1% SDS pH 8 , 1 mM PMSF , normalized for total protein and separated by SDS-PAGE . Endo H ( NEB ) digestion was performed according to the manufacturer's protocol . Staining of the western blot was performed using anti-Nef ( AG11 , [73] ) and anti-β-COP ( M3A5 [74] ) , which were purified as previously described [22] . Additional antibodies used were HA ( Covance ) and μ1 ( RY/1 [75] ) . The secondary antibody for anti-Nef , β-COP , and HA was HRP-rat anti-mouse IgG1 ( Zymed ) and for anti-μ1 was HRP-goat anti-rabbit ( Zymed ) . For Figure 6B , the IP-western experiment was performed as previously published [26] . Briefly , parental CEM T cells were spin-transduced with murine retroviral supernatant expressing either empty vector , A2/Nef or A2/CD4/Nef . At 72 hours post transduction , the cells were incubated in 20 mM NH4Cl for 4 hours . The cells were then treated with DTBP ( Pierce ) for 40 minutes , quenched per the manufacturer's protocol , and lysed in PBS with 0 . 3% Chaps and 0 . 1% SDS . The lysate was pre-cleared and immunoprecipitated with HLA-A2 with BB7 . 2 chemically crosslinked protein A/G beads ( Calbiochem ) [25] . The immunoprecipitates were washed in TBS with 0 . 3% CHAPS and 0 . 1% SDS . A more stringent IP protocol was used in Figures 6A , 7C , 8E , and S5 . For these experiments , CEM cells were transduced with control , wild type Nef , or mutant Nef expressing adenovirus ( Figure 6A and 7C ) or concentrated MSCV ( Figures 8E and S5 ) . At 48 hours post-transduction , the cells were incubated in 20 mM NH4Cl for 16 hours . The cells were not crosslinked and were lysed in digitonin lysis buffer ( 1% digitonin ( Wako ) , 100 mM NaCl , 50 mM Tris pH 7 . 0 , 1 mM CaCl2 , and 1 mM MgCl2 ) . After pre-clear , the lysates were immunoprecipitated with either BB7 . 2 ( Figures 6A and S5 ) or M3A5 ( Figures 7C and 8E ) crosslinked to beads . The immunoprecipitates were eluted and analyzed by western blot as described previously [26] . A total of 30 million CEM T cells transduced with wild type or mutant Nef using concentrated MSCV as described above were pulse labeled for 30 minutes with [35S]-methionine and cysteine . Half of the cells were collected as the zero time point and stored at −20 degrees . The remaining cells were then chased for 12 hours in RPMI , collected and stored at −20 degrees . Lysates were generated in lysis buffer ( PBS 0 . 3% CHAPS , 0 . 1% SDS pH 8 , 1 mM PMSF ) and precleared over night . They were immunoprecipitated for two hours with an anti-HLA-A2 antibody ( BB7 . 2 ) and washed once in radioimmunoprecipitation assay ( RIPA ) buffer ( 50 mM Tris pH 8 , 150 mM NaCl , 1% NP-40 , 0 . 5% deoxycholate , 0 . 1% SDS ) . The immunoprecipitates were then eluted by boiling in 10% SDS , reprecipitated with an antibody against HA ( HA . 11 , Covance ) , and washed two times in RIPA buffer . The final immunoprecipitates were then separated by SDS-PAGE , the gel was dried down and analyzed using a phosphorimager . | HIV is unique among viral pathogens in its capacity to cause chronic and progressive disease in almost all infected people . To accomplish this , HIV must evade the host immune response , especially cytotoxic T lymphocytes ( CTLs ) , which normally function to lyse virally infected cells . HIV encodes a factor , Nef , which protects HIV infected cells from lysis by anti-HIV CTLs . To prevent CTL lysis , Nef interferes with the expression of host MHC-I , which is needed for CTL recognition of infected targets . A clear understanding of how Nef works has been hampered by its many complex functions . In addition to MHC-I , Nef protein disrupts the expression of multiple other cellular targets using different mechanisms and it is unclear how one protein can accomplish all these tasks . Here , we provide evidence that Nef acts as a highly flexible adaptor protein that is capable of utilizing different protein binding domains depending on which cellular target it is bound to . For example , we present evidence that Nef binding to MHC-I creates novel motifs that result in the recruitment of AP-1 and subsequently β-COP . This series of events results in the mis-localization of MHC-I from the cell surface to cellular degradative compartments , where MHC-I is destroyed . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"biology",
"immunology/immunity",
"to",
"infections",
"virology/immune",
"evasion"
] | 2008 | HIV-1 Nef Targets MHC-I and CD4 for Degradation Via a Final Common β-COP–Dependent Pathway in T Cells |
Human African Trypanosomiasis ( HAT ) in West Africa is a lethal , neglected disease caused by Trypanosoma brucei gambiense transmitted by the tsetse Glossina palpalis gambiensis . Although the littoral part of Guinea with its typical mangrove habitat is the most prevalent area in West Africa , very few data are available on the epidemiology of the disease in such biotopes . As part of a HAT elimination project in Guinea , we carried a cross-sectional study of the distribution and abundance of people , livestock , tsetse and trypanosomes in the focus of Boffa . An exhaustive census of the human population was done , together with spatial mapping of the area . Entomological data were collected , a human medical survey was organized together with a survey in domestic animals . In total , 45 HAT cases were detected out of 14445 people who attended the survey , these latter representing 50 . 9% of the total population . Potential additional carriers of T . b . gambiense were also identified by the trypanolysis test ( 14 human subjects and two domestic animals ) . No trypanosome pathogenic to animals were found , neither in the 874 tsetse dissected nor in the 300 domestic animals sampled . High densities of tsetse were found in places frequented by humans , such as pirogue jetties , narrow mangrove channels and watering points . The prevalence of T . b . gambiense in humans , combined to low attendance of the population at risk to medical surveys , and to an additional proportion of human and animal carriers of T . b . gambiense who are not treated , highlights the limits of strategies targeting HAT patients only . In order to stop T . b . gambiense transmission , vector control should be added to the current strategy of case detection and treatment . Such an integrated strategy will combine medical surveillance to find and treat cases , and vector control activities to protect people from the infective bites of tsetse .
Human African Trypanosomiasis ( HAT , or sleeping sickness ) is a lethal , neglected disease caused by a trypanosome ( Trypanosoma brucei gambiense in West Africa ) transmitted by an insect vector , the tsetse fly ( Diptera: Glossinidae ) . In West Africa , Guinea is the country with the highest prevalence for HAT , especially in the littoral part [1] , where the vector is Glossina palpalis gambiensis . From North to South , are found the active foci of Boffa , Dubreka and Forecariah , where infection rates in humans are generally around 0 . 5–1% , but can go up to 5% in some villages [2]–[4] . This littoral part is characterised by the mangrove habitat , with mangrove trees Avicennia spp . , Rhizophora spp . This mangrove habitat generally consists in both a Guinean savannah on the mainland , and of numerous islands harbouring tsetse flies , many of which are inhabited . Very few data are available on the epidemiology of sleeping sickness in mangrove habitats , although this biotope is currently the one harbouring the highest prevalence in West Africa . There is also strong suspicion that HAT could also be highly prevalent in mangroves of Central Africa [5] , [6] . There is growing information that the mangrove systems have proved very difficult to active detection and treatment programs . Hence , elimination , which represents actual WHO's aim , is challenged in such difficult habitats . However , it is not known so far if these difficulties are due to particular behaviour of the human communities living in mangrove which puts them at high contact with tsetse , or to particular tsetse populations or tsetse/trypanosomes interactions . Accordingly , as part of a major Government of Guinea/IRD programme to tackle HAT in the mangrove systems of West Africa with the support of WHO and the Bill and Melinda Gates Foundation , we carried a cross-sectional study of the distribution and abundance of people , livestock , tsetse and trypanosomes in the one of the important foci of HAT: the focus of Boffa .
All the activities conducted in this study have been preceded by meetings of the National Control Program against HAT with authorities at the national , regional , and local level to explain the objective of the work and to obtain the agreement of the populations . Both the Ministry of health ( for human survey ) and the Ministry of Agriculture and Livestock , through the National Direction of Livestock have i . ) approved our protocols , ii . ) given their agreement and iii . ) given the administrative authorizations for these activities . Then , every district and village was informed about the projects aims and activities by people from both ministries through various means including films , discussion forums , and slideshows . All human samples were collected within the framework of medical surveys conducted by the national HAT control program ( NCP ) according to the national Guinean HAT diagnostic procedures of the Ministry of Health . No samples other than those collected for routine screening and diagnostic procedures were collected for the purposes of the present study . All human samples were anonymized . All participants were informed of the objective of the study in their own language . Hence the attendance of the population to the medical survey reflects the people who were there and who accepted to participate to the survey . Only livestock owners who agreed to participate to the study had their animals sampled . In Boffa area , the geomorphology of the Rio Pongo estuary , with its extensive areas of sedimentation , has led to the development of halophile vegetation: the mangrove trees Avicennia spp . , Rhizophora spp . This mangrove habitat consists of numerous islands , many of which are inhabited , and which harbour G . p . gambiensis [7] . On these islands , people ( mainly Soussou and Baga ethnic groups ) cultivate irrigated rice , do salt extraction , cut wood , pick oysters , and fish . The mainland harbours a typical anthropised Guinean savanna landscape mainly colonized by Elaeis Guineensis and characterized by rivers bordered by gallery forests . Here , in addition to the above mentioned activities , people carry out market gardening in lower-lying areas and grow cassava , fonio , groundnut on the hillsides . Livestock does not constitute an important part of the economy , and looks more like a secondary activity . Mainly small ruminants ( goats and sheep ) are present , whereas cattle and pig breeding are restricted to some few families . Annual rainfall is around 3000 mm . An exhaustive census of the human population was carried out . All the inhabitants of villages were counted and their houses geo-referenced using Global Positioning Systems ( Garmin GPSmap76Cx , USA ) . In addition , all tracks , water sources ( pumps , wells , holes of water , springs ) , bridges ( car , pedestrian ) and pirogue jetties ( small , big , frequented by one family or several ) were also mapped . Data on the number and types of domestic animals observed in each household was also recorded . The distribution and abundance of tsetse was assessed using biconical traps . In addition to the classical trap deployment made by walking on and between islands , the extensive network of channels meant that traps were also deployed by boat . So , in addition to the classical biconical traps , some floating traps [8] ( see Fig . 1 ) were also deployed . A total of 344 traps were deployed for 48 h of consecutive trapping . When possible , cages were harvested daily . Tsetse were identified according to species and sex . To correct the great variability of the catches , data were normalized using a log+1 transformation . The median was then calculated with the 1st and 3rd quartiles , and were then back transformed . Females were individually aged by ovarian dissection , whereas the wing-fray method was used to estimate the male population's mean age . Males and females that were still alive were dissected in order to look for possible trypanosome infection using microscopy in the midgut , the proboscis , and the salivary glands . As the Boffa area is constituted of a western and an eastern parts , separated by the Rio Pongo river which can reach more than one kilometer wide , the question was to know if the tsetse from the two parts were genetically differentiated , or alternatively if they belonged to one single panmictic population . Such information indeed has important consequences for designing tsetse control strategies ( reviewed in [9] ) . HAT was diagnosed by mass screening of the population with the routinely used Card Agglutination Test for Trypanosomiasis ( CATT , [24] ) performed on capillary-collected blood ( CATT-B ) . For each CATT-B positive person , additional blood was collected in heparinized tubes and a two-fold plasma dilution series in CATT buffer ( CATT-P ) was tested to assess the end titer , i . e . the highest dilution still positive . All individuals showing a CATT-P dilution end titer of 1/4 or greater ( positive CATT-P ) , were submitted to microscopic examination of lymph node aspirate whenever swollen lymph nodes were present , and 350 µl of buffy coat was examined using the mini-anion exchange centrifugation test on buffy-coat ( mAECT-BC , [25] ) . After centrifugation of the heparinized tube , one 1 . 5 ml eppendorf tube of plasma was sampled for every CATT-P positive subject for subsequent analyses ( trypanolysis , see below ) . Staging of the disease was done by counting white cells in the cerebrospinal fluid: if the number of cells was <6 , the patient was considered to be in the 1st stage; early stage 2 was between 6 and 20 cells , and late stage 2 corresponded to cases with more than 20 cells . At the end of the survey , all parasitologically positive subjects were treated according to the national treatment procedure taking into account staging results . During the initial geographical demographic survey , information had been gathered on the number and species of domestic animals present in the area . Power calculations suggested that a total sample size of 300 animals would provide a reliable estimate of the prevalence of trypanosome infection in the livestock population . For sampling , priority was given to good geographical coverage of the area , of course adapted to the low number of sites where domestic animals occurred ( see fig . 2 ) . Priority was then also given to sample animals known to harbour pathogenic trypanosomes , i . e . cattle and pigs , these latter having also been described as potential reservoir of human trypanosomes . For all animals sampled , blood was taken at the jugular vein in a 10 ml heparinized vacutainer tube . From this tube , a 75 µl capillary tube was taken , centrifuged . Packed cell volume ( PCV ) was measured , and parasitological diagnosis for trypanosomes was made according to the buffy-coat technique of Murray et al . [26] , which has been recognized as one of the most sensitive . All the animals sampled have been given a free trypanocide treatment ( Veriben at 3 . 5 mg/kg ) , except the pigs , for which this drug is not recommended . A deworming treatment ( Bolumisol , Laprovet ) was provided free to all animals sampled . All the blood and plasma samples , together with tsetse samples collected in the focus of Boffa were first kept in a transportable cool box with ice , and then systematically transferred twice a day in a −20°C vehicle freezer ( Engel ) . The samples were then transported back to CIRDES Bobo-Dioulasso where all the molecular analyses were done .
In total , 25 287 inhabitants ( 5 761 on the West side and 19 526 on the East side of the Rio Pongo ) were censused . According to this census , in the Rio Pongo mouth ( 734 km2 ) , the human density is around 34 inhabitants per square kilometer . Sixty seven percent of the population is located on the mainland , and 33% on the islands . Figure 2 shows high spatial heterogeneity in terms of human settlement . Most inhabitants from continental villages have a lot of activities in the mangrove ( rice cultivation , salt extraction , fishing , etc . ) . Hence there is a very high human mobility between the mainland and the islands ( for agricultural activities ) and from the islands to the mainland ( market , water supply etc… ) . This mobility , partly due to the fact that more than 50% of the population is under 25 years old , is illustrated by the important number of pirogue jetties , water point supplies , and bridges georeferenced ( see Fig . 3 ) , all these points being crucial for tsetse-human contact . Livestock production is not an important part of the local economy , and looks more like a secondary activity . Mainly small ruminants ( goats and sheep ) are present , whereas cattle and pig breeding are restricted to some few families . In total , 14 , 445 persons were screened with CATT-B . Our population census estimated that 25287 people inhabit the area . In addition , 3088 people that were not included in the census attended the medical survey . Accordingly , we estimated that 50 . 9% ( 14445/28375 ) of the population was screened . A total of 324 subjects displayed a CATT-B positive result , among which 118 were CATT-P positive ( mean seroprevalence = 0 . 82±0 . 08% ) . Among these 118 CATT-P positive subjects , 45 were confirmed to be HAT cases according to the parasitological tests ( mean prevalence = 0 . 31±0 . 05% ) . The remaining 73 subjects ( 0 . 50%±0 . 06 ) were seropositive but parasitologically negative ( hence labelled SERO ) , and did not receive treatment . Out of these 73 SERO , 14 ( 19 . 2% ) were positive using the immune trypanolysis test ( TL ) , hence labelled SEROTL+ , indicating present or past contact with T . b . gambiense . All the 45 HAT cases were TL positive ( TL+ ) , confirming a T . b . gambiense infection . Of these , 33 ( 73 . 3% ) were positive for lymph node aspirate microscopic examination , and 45 ( 100% ) were positive for mAECT-BC . Staging was performed on 43 HAT ( two patients declined to come to Boffa hospital ) : 20 were diagnosed with late second stage of the disease , 9 were in early 2nd stage , and 17 were in 1st stage , with CSF white cell counts ranging from 0 to 1460/µl . Two late stage 2 twins who were 10 months old were treated with melarsoprol instead of NECT because of perfusion constraints . All other treated patients were given NECT . Neither important side effects nor death were registered during treatment of the patients . Figure 5 shows the spatial distribution of the HAT cases and of the SEROTL+ . HAT patients and SEROTL+ subjects were not evenly distributed at the geographic scale of the focus , since most of them were located in the mangrove/mainland ( savannah ) interface . No significant differences were noted between the number of females ( n = 20 ) and males ( n = 25 ) with respect to HAT . Mean age of HAT cases was 28 . 7 ( ranging from 10 months to 60 years old ) with 31 HAT cases being 16 to 45 years old , and thus representing the active population . This means that 69% of the HAT cases were found in this active part of the population , although in this particular part of the population the attendance to the medical survey was low ( 40 . 29% ) . A total of 303 animals were sampled , consisting of 158 cattle ( all being trypanotolerant Ndama breed ) , 49 pigs , 103 sheep and 3 goats , and were parasitologically examined for trypanosomes . Only one cattle was positive , showing a T . theileri-like infection . No pathogenic trypanosome was found . Mean PCV value of the 303 animals , all types confounded , was 33 . 3±6 . 75 . Out of the 158 cattle sampled , only 4 had a PCV under 25 , this value being the usual one under which a parasitological infection is suspected in areas where trypanosomiasis is found . Out of 273 animal samples on which the trypanolysis test was implemented , two ( one pig and one goat ) were positive to LiTat 1 . 3 , indicating a contact with T . b . gambiense ( see Fig . 5 ) .
The settlement and the landscape of Boffa focus are very similar to those which can be found on other parts of the Guinean littoral . From the Mellacoree river mouth in the south , at the border with Sierra Leone , to the Rio Compony river mouth in the north , near the border with Guinea Bissau , the Guinean littoral is characterized by a mangrove ecosystem , except for the headlands of Conakry and Cape Verga [31] and the important HAT foci in Guinea ( Boffa , Dubréka and Forécariah ) are all located in or around this mangrove ecosystem [2] , [32] . Although the population of the littoral is multiethnic ( Nalou , Baga , Temne ethnic groups ) , the Soussou group is clearly the most important . These populations conduct several activities in the mangrove that expose them to the bite of tsetse flies and so to the risk of getting HAT . In a recent study of Forecariah focus , in the southernmost part of the Guinean mangrove , the identified risk activities were rice cultivation , water supply at backwater and use of pirogue jetties [32] . In Boffa , high human mobility between the mainland and the islands is certainly an important factor that increases the risk of transmission from infected tsetse to humans , and that favors the spread of the disease . First , because this mobility is probably responsible for high human/tsetse contact at pirogue jetties and in narrow mangrove channels ( highly human- frequented places with high tsetse densities ) , and second , because this mobility also contributes to a low attendance of population to medical surveys , in particular the active part of the population , which is also the most at risk . In many instances , when the medical survey team arrived in a village , where important sensitization always came along census campaign that preceded medical survey , many of the people were not present , simply because their activities in the mangrove were more important for them . It has to be noted , as part of the explanation , that these mangrove activities , have more constraints than classical agricultural ones on the mainland . For instance , people have to take into account tide hours as a first priority for their movements . These results can certainly be generalized to other HAT foci of the African coast with mangrove habitat , e . g . Equatorial Guinea , Gabon , etc . Considering tsetse distribution , two different types of habitats can be distinguished in the area: The northern and the southern ones . The northern one , on the mainland , looks like the classical distribution of the riverine species G . palpalis gambiensis in humid savannahs such as northern Ivory Coast , Burkina Faso , Mali , with tsetse being strongly associated to the forest galleries bordering water courses , and being almost absent from the savannah itself ( e . g . [33] , [34] ) . Here the highest densities are indeed found along the rivers , and near sources which offer conserved vegetation and suitable host abundance . On the opposite , in the southern part , when approaching the coast and mangrove channels , tsetse become progressively almost evenly distributed , with highest densities in narrow mangrove channels with Rhizophora vegetation , which represent hunting areas for tsetse , and also near natural watering points and pirogue jetties frequented by humans . Given the absence of genetic differentiation observed between the tsetse sampled using 9 independent microsatellite loci , it is likely that Boffa provides a habitat for a single big panmictic tsetse population , which was a priori questionable given the wideness of Rio Pongo River ( sometimes more than 1 km ) which could have acted as a barrier between the two banks of the river . This confirms earlier observation on another sleeping sickness focus of the Guinean littoral , Dubreka , where mainland mangrove was also found to host a single large and panmictic tsetse fly population [10] . With such results , tsetse eradication at the scale of the focus of Boffa is clearly not the way to go because it is impossible to target the whole tsetse population , hence emphasis will be put to reduce ( i . e . not eradicate ) tsetse densities in order to reduce tsetse/human contact and stop T . b . gambiense transmission . The medical survey results confirmed the endemic situation in the Boffa focus with an overall prevalence of 0 . 3% in humans . Transmission of T . b . gambiense probably occurs mainly at the mangrove/savannah interface which is believed to be the main sites of human contact with tsetse flies . The active population ( 20–45 years old ) is by far the most affected . This part of the population comprises 70% of the cases detected , although only 40% attended the medical survey . Transmission is probably linked to high human tsetse contacts occurring during human activities conducted in the mangrove , mainly around sites such as pirogue jetties , water supply points , and mangrove channels where many tsetse were captured . These sites will constitute priority targets for any vector control operation aiming at reducing tsetse/human contact . The proportion of patients being in the 1st stage of the disease ( 39% ) is greater than in other foci in littoral Guinea , since this proportion was only 2% in Dubreka [2] , and 16% in Forecariah [4] . This suggests that in Boffa , HAT transmission is still active , and this underlines the pressing need for active intervention in this focus . Despite important efforts to sensitize the communities , only half of the whole population attended the medical survey and was therefore medically screened . This attendance even decreases to 40 . 29% when considering only the active population , which represents the majority of the cases ( 69% ) . Given that the overall prevalence was 0 . 3% , and assuming the prevalence is the same in the non screened population ( which is very conservative ) , we therefore expect around 60 more HAT cases to be still living in the area . Moreover , application of TL proved that an additional 14 SERO-TL+ individuals ( at least ) indeed had a contact with T . b . gambiense and should be considered as potential additional carriers [35] . They confirm once more ( see also [4] , [36] ) the potential impact SERO subjects can have in the maintenance of transmission in HAT foci , especially with HAT control strategies targeting HAT patients only . In the absence of any prophylactic treatment against sleeping sickness , this number of T . b . gambiense carriers is still growing since transmission is active . This emphasizes the need for intervention ( s ) that would overcome this constraint . The presence of two 10 months old patients , the mother of whom was HAT diagnosed the year before ( M . C . , unpublished data ) , represents an additional indirect evidence for the anciently suspected existence of vertical T . b . gambiense transmission ( reviewed in [37] ) . The absence of pathogenic trypanosomes in domestic animals was not expected , but is very likely given both their absence using parasitology ( BCT method ) on more than 300 animals sampled , and also given the high mean PCV values found . These results suggest that animal trypanosomiasis is not a major veterinary problem in this area . The same result had been found in Loos islands , near Conakry , where no domestic animal had been found infected with trypanosomes out of 104 sampled [38] . The fact that none of the infected tsetse had trypanosomes identified using PCR suggests that the trypanosomes circulating may come from reptiles for instance , such as T . grayi for which we did not do PCR . Reptiles ( crocodiles , monitor lizards ) are numerous in this mangrove area . But it may also suggest that unrecognised , possibly pathogenic , trypanosome species exist that were not identified . This can be particularly evoked for the two midgut+proboscis infections which would have been interpreted as a Nannomonas infection if only parasitological methods had been used . Hence the only pathogenic trypanosome identified in Boffa focus is T . b . gambiense , its presence having been confirmed in humans , and in two domestic animals thanks to the trypanolysis test which is specific for T . b . gambiense [3] . T . b . gambiense has not been found in tsetse , confirming the usual ( but poorly understood ) very low ( <0 . 1% ) mature infection rates of T . b . gambiense in tsetse , even in active sleeping sickness foci ( [39] . The same has been reported for T . b . rhodesiense ( see [40] ) . To summarize , our results suggest the absence of pathogenic trypanosomes in domestic animals in the focus of Boffa . We show the presence of T . b . gambiense in humans , and a contact between T . b . gambiense and some domestic animals ( one pig and one goat in our study ) . The vectorial capacity of G . p . gambiensis in this focus seems very low , confirming what was found on another area of the Guinean littoral , the Loos islands , where Kagbadouno et al . [38] also reported an absence of trypanosome in the G . p . gambiensis dissected . This is in contrast with other areas where G . p . gambiensis occurs , such as the savannah areas of Burkina Faso or Mali for instance , where G . p . gambiensis can usually be found infected with animal trypanosomes , with infection rates generally ranging from 2 to 10% [41] , [42] . The G . p . gambiensis from littoral Guinea may also be genetically different from the one from the West African savannah , this being currently investigated ( P . S . , unpublished data ) . Nonetheless , even a very small number of tsetse having a T . b . gambiense mature infection are able to spread these trypanosomes , since the infection by T . b . gambiense alters tsetse saliva and modifies the behaviour of the tsetse , favoring the risk of human infection [43] . The “good news” here is that , given the very low proportion of tsetse infected with T . b . gambiense , a human needs to be bitten a great number of times before being bitten by an infective tsetse . Hence , any intervention that will reduce tsetse numbers and tsetse/human contact will also reduce the number of tsetse infected , and will protect people from an infective bite . Interventions directed against animal trypanosomiasis have been advocated as a useful entry point for controlling HAT [44] . However , in the littoral HAT foci of West Africa , and in many Central African foci , livestock are scant and/or trypanosomiasis is not an important constraint for livestock production . In these mangrove foci , interventions should be applied at a relatively small scale and have to be aimed specifically on eliminating HAT . In conclusion , the prevalence of pathogenic trypanosomes in the human population , combined with low attendance at medical surveys and to an additional population of human carriers of T . b . gambiense living in the community , highlights the importance of implementing new strategies . We suggest that in order to stop T . b . gambiense transmission in Boffa and similar foci in West and Central Africa , vector control should be added to the current strategy of case detection and treatment . Such an integrated strategy of control will combine medical surveillance and vector control activities , the former finding and treating cases while the latter will protect people from the infective bites of tsetse . The recent development of insecticide-treated targets that are effective for G . p . gambiensis [45] offers the prospect of a method that can be applied in the mangrove systems . In addition there is an absolute need to follow up seropositive people , and to target more efficiently the population at risk if HAT is to be eliminated from the focus . | Human African Trypanosomiasis ( HAT ) in West Africa is a lethal , neglected disease caused by Trypanosoma brucei gambiense transmitted by the tsetse fly Glossina palpalis gambiensis . Although the littoral part of Guinea with its typical mangrove habitat is the most prevalent area in West Africa , very few data are available on the epidemiology of the disease in such biotopes . We carried out a cross-sectional study of the distribution and abundance of people , livestock , tsetse and trypanosomes in the active focus of Boffa . We only found T . b . gambiense in the area , no other trypanosome . T . b . gambiense was found parasitologically in humans ( 45 cases ) , and suspected serologically in other humans and in two animals . Tsetse flies were present in high densities in places very frequented by humans , such as pirogue jetties , and watering points . Our results confirm the importance of medical surveys to find cases and treat them , but also point out the limit of strategies targeted at HAT patients only . If sleeping sickness is to be eliminated , a vector control component must be added to the strategy of case detection and treatment , and this latter must be directed to the population the most at risk . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"infectious",
"disease",
"epidemiology",
"african",
"trypanosomiasis",
"population",
"genetics",
"parasitic",
"diseases",
"effective",
"population",
"size",
"neglected",
"tropical",
"diseases",
"tsetse",
"fly",
"zoolo... | 2012 | Epidemiology of Sleeping Sickness in Boffa (Guinea): Where Are the Trypanosomes? |
We investigated the relationship between bacterial load in Buruli ulcer ( BU ) lesions and the development of paradoxical reaction following initiation of antibiotic treatment . This was a longitudinal study involving BU patients from June 2013 to June 2017 . Fine needle aspirates ( FNA ) and swab samples were obtained to establish the diagnosis of BU by PCR . Additional samples were obtained at baseline , during and after treatment ( if the lesion had not healed ) for microscopy , culture and combined 16S rRNA reverse transcriptase/ IS2404 qPCR assay . Patients were followed up at regular intervals until complete healing . Forty-seven of 354 patients ( 13% ) with PCR confirmed BU had a PR , occurring between 2 and 42 ( median 6 ) weeks after treatment initiation . The bacterial load , the proportion of patients with positive M . ulcerans culture ( 15/34 ( 44% ) vs 29/119 ( 24% ) , p = 0 . 025 ) and the proportion with positive microscopy results ( 19/31 ( 61% ) vs 28/90 ( 31% ) , p = 0 . 003 ) before initiation of treatment were significantly higher in the PR compared to the no PR group . Plaques ( OR 5 . 12; 95% CI 2 . 26–11 . 61; p<0 . 001 ) , oedematous ( OR 4 . 23; 95% CI 1 . 43–12 . 5; p = 0 . 009 ) and category II lesions ( OR 2 . 26; 95% CI 1 . 14–4 . 48; p = 0 . 02 ) were strongly associated with the occurrence of PR . The median time to complete healing ( 28 vs 13 weeks , p <0 . 001 ) was significantly longer in the PR group . Buruli ulcer patients who develop PR are characterized by high bacterial load in lesion samples taken at baseline and a higher rate of positive M . ulcerans culture . Occurrence of a PR was associated with delayed healing . ClinicalTrials . gov NCT02153034 .
Buruli ulcer ( BU ) is a neglected tropical disease caused by infection with Mycobacterium ulcerans ( M . ulcerans ) which is common in rural parts of West African countries including Ghana . It causes large , disfiguring skin ulcers mainly in children aged 5 to 15 years although persons of any age can be affected [1] . Access to treatment in rural areas is limited and many patients present with late stage disease because of fear , suspicion about conventional medicine and the economic consequences for poor families [2] . The incidence of the disease is highly focal , and in Ghana for example , most cases occur in particular parts of the Ashanti Region [3] . The mode of transmission remains unknown but there have been major advances in understanding the mechanism of disease since the establishment of the WHO Buruli ulcer Initiative in 1998 together with improved diagnosis and management . The initial BU lesion is a subcutaneous painless nodule tethered to the skin or an intradermal plaque . These enlarge over a period of days to weeks and ulcerate in the centre . Ulcers are usually painless and have a necrotic base and irregular undermined edges [4 , 5] . The mainstay of treatment is the combination of rifampicin and streptomycin or clarithromycin but additional treatment such as debridement and skin grafting , and early basic management with appropriate dressings and physiotherapy when an ulcer is close to a joint can minimize complications [6 , 7] . The most common complication is paradoxical reaction occurring during or after treatment in 8–12%[5 , 8] of patients in Africa . In an Australian population the phenomenon has been reported to occur more frequently ( 21% ) in elderly patients [9] . Another study in Africa , also reported similar frequency of paradoxical reaction occurrence ( 22% ) associated with trunk localization , larger lesions and genetic factors [10] . The time to development of paradoxical reaction varies widely between patients from antibiotic initiation , from few weeks in some patients to several months in others [5 , 8 , 9] . Paradoxical reactions cause anxiety to both patient and carer with the possibility that it represents uncontrolled or recurrent infection and indeed it is likely that earlier perceptions that antibiotics were ineffective for management of Buruli ulcer may have been influenced by such reactions . Culture of samples from the lesions are usually negative if the reaction occurs after completion of antibiotics but this does not exclude persistent infection since the sensitivity of culture for Mu is only 35–60%[11 , 12] . Paradoxical reaction is thought to be due to an immunological response to residual M . ulcerans antigens which are known to persist for many months after successful treatment[9] . The immunological mechanism underpinning paradoxical reactions requires clearer elucidation in order to design appropriate evidence-based interventions for this important clinical phenomenon . Even though , several studies have associated paradoxical reactions with larger lesions , its relation to bacterial load has not been demonstrated . The aim of the present study was to investigate the clinical forms of paradoxical reactions in relation to their time of occurrence , the lesion type and bacterial load as potential risk factors for their occurrence .
From June 2013 to June 2017 , patients with clinically suspected Buruli ulcer were screened at Agogo Presbyterian hospital , Tepa , Dunkwa and Nkawie-Toase Government hospitals . The diagnosis of Buruli ulcer was confirmed by M . ulcerans IS2404 PCR . Patients who had already started antibiotic therapy or refused to participate were excluded . All categories of BU lesions were included . Demographic data of all participants , details of the timing and nature of paradoxical reactions were collected prospectively using WHO BU01 and study designed laboratory forms . The dimensions of lesions were documented using Silhouette ( ARANZ Medical , Christchurch , New Zealand ) , a 3-dimensional imaging and documentation system together with digital photographs[13] . Patients were reviewed by an experienced clinician every 2 weeks up to 8 weeks and thereafter every month up to one year after completion of treatment . The time of complete healing was documented for all patients . All the patients recruited into the study were given combination antibiotic therapy of either rifampicin and clarithromycin or rifampicin and streptomycin for 56 days as recommended [7] . Two fine needle aspirates ( FNA ) were taken from non-ulcerated lesions; for patients with ulcerated lesions , two swabs from the undermined edges of ulcers were taken to confirm the diagnosis of BU by microscopy and PCR . The presence of viable bacteria was determined by taking samples for culture and 16S rRNA reverse transcriptase/IS2404 qPCR assay . Samples were collected at baseline and at weeks 4 , 8 , 12 and 16 ( only if lesions remained unhealed ) . When a paradoxical reaction occurred , samples were taken from those lesions for culture and 16S rRNA reverse transcriptase/ IS2404 qPCR assay . Paradoxical reactions were defined by the presence of one or both of the following features as previously described: ( i ) an initial improvement in the clinical appearance of an M . ulcerans lesion during or after antibiotic treatment , followed by an episode of new inflammation , with or without pus formation , with significant enlargement of a healing lesion or its surrounding tissues or ( ii ) the appearance of a new lesion ( s ) [14] . Clinical samples were transported to the laboratory in appropriate transport media and processed immediately upon arrival at the laboratory . All routine laboratory tests and molecular assays were conducted at Kumasi Centre for Collaborative Research in Tropical Medicine ( KCCR ) . For laboratory confirmation of Buruli ulcer disease , smear microscopy for acid-fast bacilli , culture on Lowenstein-Jensen medium and IS2404 qPCR were performed by well-established methods as previously described[15–17] . A final diagnosis of Buruli ulcer was based on the IS2404 qPCR result which was the most sensitive test . FNA and swab samples were transported from the study site to the KCCR laboratory stabilized in 500 μl RNA protect ( Qiagen , UK ) . Whole transcriptome RNA and whole genome DNA were extracted separately from the same clinical sample . The RNA and DNA isolation was carried out within 5 hours of sample collection using the AllPrep DNA/RNA Micro kit ( Qiagen , UK ) . RNA extracts were reverse transcribed into cDNA using Quantitect kit as described elsewhere[13] . The cDNA prepared was subjected to qPCR for detection of human glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ) mRNA [17] . The detection of the GAPDH was for quality assurance purposes to confirm correct sample collection and to exclude false negative 16S rRNA RT qPCR results . All whole transcriptome RNA extracts from Buruli ulcer lesions tested positive when subjected to GAPDH mRNA RT qPCR at baseline . The cDNA was then subjected to 16S rRNA qPCR and DNA to IS2404 qPCR to increase the specificity for M . ulcerans and for quantification of the bacterial load as previously described[16] . Ten-fold serial dilutions of known amounts of a plasmid standard of IS2404 ( 99 bp ) and 16S rRNA ( 147 bp ) ( Eurofins MWG Operon , Ebersberg , Germany ) were included with PCR amplification for preparation of a standard curve . M . ulcerans bacillary loads in original clinical samples were calculated based on threshold cycle values per template of IS2404 qPCR ( standard curve method ) adjusted to the whole amount of DNA extract and the known copy number of 207 IS2404 copies per M . ulcerans genome on average . The raw data generated from the study were entered in Microsoft Excel ( Microsoft Corporation , Redmond , WA ) and analyzed using GraphPad Prism version 5 . 0 ( GraphPad Software , Inc . , La Jolla , CA ) STATA statistical package ( StataCorp ) . Continuous variables such as age , IS2404 copies and M . ulcerans 16SrRNA copies were compared using the Mann-Whitney U test . Chi-square test was used to compare the frequencies of all categorical variables , except the location of lesions which was compared using the Fisher’s exact test . The frequency and percentage of missing values for each variable were collected , analysed and reported . When there were missing values for the variables of interest/outcome , exclusion of observations with missing values for the variables of interest/outcome was considered . Highly incomplete covariates ( >33% of observations missing ) were excluded from analyses . The Kaplan-Meier survival analysis was used to determine the effect of developing paradoxical reaction on time to healing . Simple proportions of positive AFB and culture among the paradoxical reaction and the non-paradoxical reaction participants were also calculated . Logistic regression was performed to assess incidence rates and association of variables with PR . Univariate analysis was done to determine crude rate ratios and a multivariate analysis performed adjusting for age , gender and location of lesion to test associations with characteristics assessed at pretreatment . P value < 0 . 05 was considered statistically significant in all the analyses . Verbal and written informed consent was obtained from all eligible participants and from parents or legal representatives of participants aged 18 years or younger . Ethical approval was obtained from the Committee of Human Research Publication and Ethics , School of Medical Sciences , Kwame Nkrumah University of Science and Technology , Kumasi , Ghana ( CHRPE/AP/229/12 ) and registered with ClinicalTrials . gov identifier NCT02153034 .
A flowchart indicating the recruitment of patients is shown as Fig 1 . Forty-seven ( 13% ) out of 354 patients in the study developed paradoxical reactions . Of the 32 lesions that enlarged , 24 ( 51% ) were just warm and enlarged , and 8 ( 17% ) were also warm and pus filled , with or without pain . When enlargement happened , it was always during antibiotic therapy . Fifteen patients ( 32% ) developed new lesion ( s ) : 14 of them had a new lesion developing close to the original site and one had multiple new lesions around the existing one ( Fig 2 ) . Thirty-two ( 68% ) paradoxical reactions occurred during antibiotic treatment . The median ( IQR ) time from the start of antibiotic administration to development of paradoxical reaction was 6 ( 4–11 ) weeks but the time to occurrence of an enlarged lesion ( 6; 4–8 weeks ) was shorter compared to that for a new lesion ( 10; 5–28 weeks: p<0 . 01 ) . A higher proportion ( 26/29 = 90% ) of patients with nodules and plaques with paradoxical reaction had an enlarged lesion in comparison to patients with oedematous lesions and ulcers ( 6/18 = 33%: p<0 . 001 ) . The majority of paradoxical reactions manifested as new lesions occurred in patients with oedematous lesions and ulcers ( 12/18 = 66% ) . No other variables significantly influenced the type of reaction that occurred . Patients who developed a paradoxical reaction had a significantly higher bacterial load both in terms of IS2404 copy numbers , median cps/ml ( IQR ) [500 ( 500–8000 ) vs 500 ( 500–500 ) , p = 0 . 020] and higher viable organisms measured by Mu 16SrRNA , median cps/ml ( IQR ) [500 ( 500–4875 ) vs 500 ( 0–1000 ) , p = 0 . 014] at baseline than patients with no paradoxical reaction ( Fig 3 ) . This was supported by the finding that a larger proportion of patients who developed a paradoxical reaction had positive AFB microscopy ( 61% ) compared to those who did not ( 31%; p = 0 . 003 ) . Similarly , the proportion of patients with positive culture results was significantly higher in those who developed paradoxical reaction ( 44% vs 24%; p = 0 . 025 ) ( Table 1 ) . Paradoxical reactions were related to the clinical form of the initial lesion . They were more common in patients with a plaque ( 27% ) or oedematous lesion ( 23% ) than in those with nodule ( 15% ) or ulcer ( 7% ) ( p <0 . 001 ) . Their incidence was significantly related to lesion category at presentation; 9% in category I and 10% in category III had a paradoxical reaction compared to 19% in category II ( p = 0 . 04 ) ( Table 2 ) . Paradoxical reaction was equally common in patients who received streptomycin or clarithromycin combined with rifampicin . Using a logistic regression model , multivariate analysis showed that plaque ( OR 5 . 42; 95% CI 2 . 25–13 . 04 ) ; p<0 . 001 ) , oedematous lesion ( OR 4 . 13; 95% CI 1 . 37–12 . 42; p = 0 . 012 ) , nodular lesion ( OR 2 . 63; 95% CI 1 . 12–6 . 17; p = 0 . 026 ) and category II lesions ( OR 2 . 37; 95% CI 1 . 19–4 . 71; p = 0 . 014 ) were strongly associated with development of paradoxical reaction adjusting for age , gender and location of lesion . Positive cultures for M . ulcerans and/or positive 16S rRNA results were found in two out of fifteen patients ( A and B in Fig 4 ) who had a paradoxical reaction after completion of antibiotic treatment . In patient A , a new lesion appeared on the right knee at week 10 at the same time as the original lesion on the right thigh re-ulcerated . Culture was positive from the new lesion . No additional antibiotics were administered and both lesions healed by week 24 . Patient B , a 16-year-old girl with an ulcer on the left upper arm , developed a new lesion close to the initial lesion at week 11 which tested positive to combined 16S rRNA/IS2404 qPCR assay . Both lesions healed completely at week 20 without further antibiotic therapy . All other paradoxical reactions before or after treatment had negative culture and 16S rRNA/IS2404 qPCR results . In 23 patients who had positive M . ulcerans 16S rRNA or culture results at week 4 , the median time to developing a PR was 6 weeks ( IQR 4–8 ) compared with 13 weeks ( IQR 6–26 , p = 0 . 015 ) in 10 patients with negative week 4 M . ulcerans 16s rRNA/culture results . The median time to complete healing for patients with paradoxical reaction was 28 weeks compared to 16 weeks for those with no PR ( p <0 . 001 ) ( Fig 5 ) . By the end of antibiotic treatment at 8 weeks only 2 of 47 ( 4% ) PR group patients had completely healed compared to 42 of 307 ( 12% ) in the no PR group . Patients with PR had a 1 . 58-fold increase ( 95% CI 1 . 23–2 . 10 ) in the time to complete healing of Buruli ulcer lesions compared to those who did not develop PR .
Paradoxical reactions have been reported previously in 8–12% of Buruli ulcer patients in Africa during or after treatment with antibiotics [5 , 8] . In the present study the overall incidence was similar at 13% but lower than that which was reported in an Australian cohort [9] , possibly because of the younger age distribution in this study compared to the elderly population in the Australian study; older age is a known risk of developing PR . We also recognised some distinctions in the clinical presentation . During antibiotic treatment for 8 weeks the common form of paradoxical reaction was re-enlargement of the lesion after healing had begun . This occurred more than 4 weeks after initiation of treatment by which time the necrotic tissue around the lesion had cleared by auto-debridement and it was usually associated with new inflammation , sometimes severe with pus formation . After completion of antibiotic treatment , paradoxical reactions consisted mainly of new inflammatory lesions adjacent to the original one , with or without new ulceration in the original lesion . This was more common when the initial lesion was an ulcer or oedematous lesion , possibly because bacteria were more widely disseminated around such lesion . It takes a longer time to kill the bacteria and clear mycolactone in the skin before inflammatory process due to dead bacteria in the skin starts which results in delayed PR even at distant sites occurring as new lesions . It is difficult to make a distinction between paradoxical reaction and treatment failure when viable M . ulcerans can still be detected by culture or by the 16S rRNA assay in lesions during or after antibiotic treatment [13] . This was the case in 2 of 11 patients who developed new inflammatory lesions after completion of antibiotics but they were included as paradoxical reactions because they had clinical features of inflammation such as pain and/or pus formation and in each case the inflammation settled without further antibiotic treatment or any other additional therapy . The diagnosis of paradoxical reactions is usually clinical but investigations such as AFBs , mycobacterial culture , PCR for IS2404 repeat sequence and histopathology may be done . Viable organisms may also be detected using the 16S rRNA assay . In the present study , 2 patients were culture and 16S rRNA positive at the time of development of PR . Other studies have reported negative mycobacteria cultures at the diagnosis of PR [5 , 9 , 18] . In our study , no additional treatment ( antibiotics , surgery or steroids ) was given when PR was detected . However , other treatments including aspiration of pus without additional antibiotics[18] , surgical excision [5] and administration of steroids [9] have been reported . It is impossible to estimate accurately the total M . ulcerans bacterial load in a Buruli ulcer lesion but using a simple sampling method and estimating bacterial load by the number of copies of IS2404 or 16S rRNA , we found an association between paradoxical reaction and high baseline bacterial load . This was supported by the finding that AFB detection and M . ulcerans culture were also more likely to be positive in these patients . The pathogenesis of paradoxical reaction in M . ulcerans disease is unknown but one hypothesis is that it is caused by an inflammatory reaction that , prior to antibiotic treatment , is suppressed by mycolactone , the M . ulcerans toxin . As the organisms are killed by antibiotics , mycolactone production ceases and its suppressive effect is lost causing a rebound of inflammation . This would be analogous to paradoxical reactions in M . tuberculosis and HIV co-infected patients when anti-retroviral treatment restores the immune response . In immune reconstitution inflammatory syndrome ( IRIS ) associated with HIV and M . tuberculosis or cryptococcal co-infection it has been postulated that antigens of the co-infecting pathogen accumulate before the immune response recovers leading to an excessive acute inflammatory response during anti-retroviral treatment[19] . There is a considerable disparity in the time to development of the paradoxical reaction . In this study it occurred from 4 to 28 weeks following initiation of therapy but 2 to 58 weeks is reported in other studies[9 , 20] . We predicted that a paradoxical response would happen earlier in patients in whom M . ulcerans was cleared rapidly from their lesion . In fact , the opposite was the case; when the M . ulcerans 16S rRNA assay and culture were negative at 4 weeks , the paradoxical reaction occurred later than in those whose tests were still positive at 4 weeks . Unfortunately , we were unable to measure mycolactone in this study so the observation remains difficult to explain . Further studies of the pattern of cytokine secretion during treatment may shed some light on the problem . | Buruli ulcer is a neglected tropical skin disease caused by the third most common pathogenic mycobacterium: Mycobacterium ulcerans . Paradoxical reaction , a phenomenon observed in some patients is characterised by worsening of existing lesion ( s ) with attendant pain and occurrence of new lesions during or after antibiotic therapy following an initial period of clinical improvement . This significantly affects treatment outcomes . In this clinical study , tissue samples obtained from patients were subjected to 16S rRNA/ IS2404 qPCR to measure bacterial load . This was to identify a link between bacterial load in BU lesions and the development of paradoxical reactions following initiation of antibiotic treatment . We found that 13% of participants developed PR . Patients who developed PR had higher baseline bacterial load; a higher rate of positive M . ulcerans culture and persistently positive culture during antibiotic treatment . Occurrence of a paradoxical reaction was associated with delayed healing . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"antimicrobials",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"drugs",
"immunology",
"tropical",
"diseases",
"microbiology",
"bacterial",
"diseases",
"physiological",
"processes",
"signs",
"and",
"symptoms",
"antibiotics",
"ulcer... | 2019 | Paradoxical reactions in Buruli ulcer after initiation of antibiotic therapy: Relationship to bacterial load |
Dispersal is a critical life history behavior for mosquitoes and is important for the spread of mosquito-borne disease . We implemented the first stable isotope mark-capture study to measure mosquito dispersal , focusing on Culex pipiens in southwest suburban Chicago , Illinois , a hotspot of West Nile virus ( WNV ) transmission . We enriched nine catch basins in 2010 and 2011 with 15N-potassium nitrate and detected dispersal of enriched adult females emerging from these catch basins using CDC light and gravid traps to distances as far as 3 km . We detected 12 isotopically enriched pools of mosquitoes out of 2 , 442 tested during the two years and calculated a mean dispersal distance of 1 . 15 km and maximum flight range of 2 . 48 km . According to a logistic distribution function , 90% of the female Culex mosquitoes stayed within 3 km of their larval habitat , which corresponds with the distance-limited genetic variation of WNV observed in this study region . This study provides new insights on the dispersal of the most important vector of WNV in the eastern United States and demonstrates the utility of stable isotope enrichment for studying the biology of mosquitoes in other disease systems .
The distance and direction of mosquito movement on the landscape are critical factors in the development of effective strategies for control of both nuisance and vector mosquito species . At small spatial scales , effective mosquito abatement using adult insecticides or larvicides needs to incorporate information on flight range of the intended mosquito target [1] . For example , when controlling Aedes aegypti , the vectors of dengue virus , insecticides are sprayed at homes of infected patients and in a specified proximity to the homes based on studies quantifying adult female dispersal [2]–[4] . Short range dispersal of Ae . aegypti has been quantified in dengue-endemic areas using genetic markers in relation to habitat structure , in particular presence of road networks , which act as barriers to mosquito dispersal and further influence the local distribution and risk of dengue cases in humans [5] , [6] . At large spatial scales , mosquito movement has been implicated in shaping the geographical spread of West Nile virus ( WNV ) across North America [7] , underscoring the importance of vector dispersal for shaping spatial patterns of disease transmission . Currently , many alternative strategies to insecticides for vector-borne disease control are being implemented , including sterile insect technique , biological control using Wolbachia , and genetically modified mosquitoes ( reviewed by [8] ) . For these disease control strategies to succeed and reduce the global burden of vector-borne diseases , a critical parameter necessary for field implementation of these strategies is the distance mosquitoes travel across the landscape . For example , control programs that release sterile , Wolbachia-infected , or genetically modified mosquitoes need detailed understanding of flight ranges to determine the appropriate spatial resolution of the release points . Simulation models of these intervention programs often incorporate parameters to represent adult mosquito dispersal [9] , [10] , although limited data on actual dispersal presents challenges to these models [11] . Given the importance of adult mosquito behavior , mosquito biologists have utilized mark-release-recapture studies for several decades to estimate mosquito dispersal distance and patterns . Diverse methods have been used to mark mosquitoes to study dispersal including dyes , paints , dusts , trace elements , and radioactive isotopes ( reviewed by [1] ) . The ideal insect marker should persist without inhibiting normal biology , be environmentally safe , cost-effective , and easy to use [12] . However , existing techniques to mark mosquitoes tend to be labor intensive , as they require rearing mosquitoes , marking them in large quantities , and then inspecting large numbers of individuals to detect re-captures [13] . Additionally , the process of rearing adults , marking them , and releasing them may change behavior compared to natural populations [1] , [14] . Further , the artificial release of mosquitoes inflates local populations that may contribute to pathogen transmission; this has led to studies where the proboscis has been glued or amputated to prevent feeding [15] , a process with potential consequences for mosquito behavior . In 2008 , a meeting of international experts in vector biology discussed critical needs in vector-borne disease control [16] . Among the research priorities highlighted , the panel listed improved technologies for marking insects for studying basic biology . To meet this challenge and to overcome limitations of previous techniques , Hamer et al . [17] developed a stable isotope method to mark naturally-occurring Culex pipiens . The laboratory and field experiments from this study suggested life-long marker retention in adults with no apparent impact on survival or body size . Stable isotopes occur naturally in the environment , are non-toxic and non-radioactive , and incorporate into living tissue , which make them safe and useful tracers [18] . Several studies have utilized stable isotopes to study dispersal of adult insects; 15N was added to streams , immature aquatic insects incorporated the rare isotope into structural body tissues , and then the emergent adult insects were captured at different distances from the enriched stream [19]–[21] . Additionally , mosquitoes have been enriched with stable isotopes in the context of Sterile Insect Technique programs [22] , [23] . Here we report the first application of stable isotopes to study mosquito dispersal in natural field conditions . We enriched naturally existing larval Culex mosquitoes with 15N in catch basins in Alsip , Illinois and used a large network of traps to capture marked females . This study demonstrates that female Culex mosquitoes were capable of flying up to 2 . 4 km with a logistic distribution function suggesting that 90% of the female Culex mosquitoes stayed within 3 km of their larval habitat . We discuss the advantages and disadvantages of the stable isotope enrichment of natural larval habitats and demonstrate how this approach could be a valuable new tool to study dispersal of medically important mosquitoes around the world .
From July to October 2010 and 2011 , we treated nine catch basins with 15N-potassium nitrate in Alsip , Illinois ( 41°41′14 . 56″N; 87°44′32 . 84″W ) . The catch basins are stormwater drains designed with a sump to prevent organic debris from entering pipes that lead to an outlet at a creek . The catch basins we treated were the terminal catch basins before water drained through an outlet into a small stream ( Stony Creek; Figure 1 ) . The maximum distance between treated catch basins was 153 m . We added stable isotopes with the quantity based on the amount of water present in the catch basin sumps [17] . Briefly , the volume of water in each catch basin sump was estimated under the assumption that sediment comprised 50% of the sump volume . Initial treatment began with a targeted enrichment of 2 . 0 mg of stable isotope per liter of water in the catch basin . In subsequent treatments that followed rain events , we reduced this amount to a half dose or quarter dose , depending on the amount of flushing that had occurred . In 2010 , we delivered 15N-potassium nitrate into the 9 catch basins on seven occasions for a total of 38 . 58 grams being delivered . In 2011 , we treated these same catch basins on 14 occasions for a total of 51 . 36 grams . Fourth instar larvae , pupae , and adult male and female Culex mosquitoes were subsampled from these catch basins and submitted for stable isotope analysis to monitor the level of enrichment and adjust the frequency and quantity of isotopic amendment accordingly . We estimated the production of 15N-enriched mosquitoes from the amended catch basins using previously described emergence traps [24] . These traps were necessary because S-methoprene is used by mosquito control agencies in the study region and the presence of larvae or pupae does not necessarily reflect adult emergence . These standardized emergence traps covered a known surface area inside the catch basin and allowed an estimation of the total production of marked adults leaving the catch basin . In 2010 and 2011 , we placed and continuously monitored three emergence traps in three of the treated catch basins from July to October . In addition , we estimated the number of Culex mosquitoes emerging from all nine catch basins per day from July to October [24] . To ensure that our enrichment activities were not affecting larval mosquitoes outside of our desired areas , we monitored down-stream enrichment by sampling immature mosquitoes and benthic invertebrates ( Chironomidae , Amphipoda , Ephemeroptera , Coleoptera ) in the Stony Creek; upstream , downstream , and at the outlet opening . We sampled these invertebrates on July 23 and August 31 in 2010 and on August 30 in 2011 . Invertebrates from the creek were submitted for stable isotope testing and all δ15N values represented natural abundance levels ( mean δ15N = 11 . 75 ) indicating no down-stream enrichment . We obtained permission to add stable isotopes to the environment by the municipalities , Cook County Department of Public Health , Illinois Department of Public Health , and the Illinois Environmental Protection Agency . Mosquitoes were trapped from May to October , 2010 and 2011 in Alsip , Blue Island , Chicago , Chicago Ridge , and Oak Lawn , Illinois . We deployed CDC light traps at 100 different locations and gravid traps at 40 locations in 2010 and 83 light trap locations and 33 gravid trap locations in 2011 ( Figure 1 ) . The closest mosquito trap was 17 . 6 m from the centroid of the nine enriched catch basins and the farthest trap was 3 . 3 km . The mean trap distance from the centroid of the catch basins was 1 . 47 km and 0 . 89 km in 2010 and 2011 , respectively . These trap locations were distributed in all directions from the catch basins and specific locations were dependent on obtaining permission from landowners . All locations were trapped once per week . Mosquitoes were identified by species and sex , and then pools of up to 50 female Culex spp . mosquitoes were tested for WNV using a quantitative RT-PCR [25] . Adult female Cx . pipiens and Cx . restuans collected in traps were pooled together as Culex spp . given the difficulty in distinguishing the two Culex species morphologically [26] . RNA was extracted using a MagMAX Viral Total RNA Isolation Kit ( Applied Biosystems , Foster City , California ) . A subset of female Culex spp . mosquitoes were placed in pools of up to five individuals and prepared for stable isotope testing as previously described [17] . In 2010 and 2011 , we deployed a Hobo weather station ( Onset Computer Corporation , Pocasset , MA ) placed 1 . 9 km from the amended catch basins . This station recorded hourly temperature , wind speed , wind direction , and precipitation . We calculated the average wind direction as a combined vector of the mean wind angle and speed [27] . The direction and speed were converted into N-S and E-W components and averaged over the July to September period . The average wind direction in 2010 was 216° ( southwest ) with a net speed of 1 . 9 kph and in 2011 was 170° ( south ) with a net speed of 0 . 66 kph . All 4th instar larvae , pupae , adult mosquitoes , and other aquatic invertebrates were stored at −20°C and processed for stable isotope analysis as previously described [17] . Briefly , samples were dried at 50°C for 24 h , encapsulated into tin capsules to create a sphere , arranged into a 96-well plate , and submitted for stable isotope analysis at the University of California-Davis Stable Isotope Facility using a PDZ Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20Ð20 isotope ratio mass spectrometer ( IRMS; Sercon Ltd . , Cheshire , United Kingdom ) . Additional analyses , that required short turn-around of results in order to guide field enrichment activities , were performed by Isotech Laboratories Inc . by using a Carlo Erba CHNS-O EA1108 ( CE Instruments , Milan , Italy ) coupled to a ThermoFisher Delta V Plus IRMS ( Thermo Fisher ScientiÞc , Bremen , Germany ) via a ThermoFinnigan ConFlo III interface ( Thermo Electron Corp . , Waltham , MA ) . Mosquito pools collected in the field and tested for stable isotopes were considered enriched if δ15N values were at least three standard deviations above the natural isotopic abundance of Culex mosquitoes in our study region [28] . We calculated the mean distance traveled ( MDT ) and incorporated a correction factor for each annulus given unequal trap density [29]–[32] . We calculated MDT for each year based on the equations in [33] . The estimated recaptures ( ER ) for each annulus were calculated as:The mean distance traveled ( MDT ) was calculated as: To estimate the probability of detecting a marked mosquito at different distances from the release point , we used a logistic regression in Program R [34] . Similar dispersion models have been used for relating the distance insects travel from a central point [35] . The data from two years were merged and each trap location ( n = 167 ) was given a 1 for detecting a marked individual or a zero for no marked individuals . We used the centroid of the 9 catch basins receiving stable isotopes to calculate the distance from each light or gravid trap . The logistic distribution function of plogis ( x ) = ( 1+tanh ( x/2 ) ) /2 was used for the predictions of detecting a marked mosquito at different distances from the release point .
During the enrichment of nine catch basins receiving 15N-potassium nitrate in 2010 and 2011 , a subsample of 4th instar larvae and pupae were collected and all were identified as Culex pipiens or Culex restuans . These immature specimens collected directly from the treated catch basins had a mean δ15N of 484 . 1±73 . 1 ( n = 73 ) while immature Culex pipiens collected from nearby untreated catch basins had a mean δ15N of 4 . 7±0 . 74 ( n = 15 ) . Using emergence traps , we estimated that the nine enriched catch basins produced 1 , 138 female and 769 male Culex spp . mosquitoes in 2010 and 2 , 624 female and 3 , 362 male Culex spp . mosquitoes in 2011 from July to October ( Figure 2 ) . The emergence of adults from the catch basins declined following large rain events ( e . g . greater than 1 cm per day [24] ) and the capture of marked pools from traps tended to occur following increased periods of emergence . A total of 343 larvae collected in catch basins or adults collected in emergence traps were identified to species during the two years and 333 ( 97% ) were Cx . pipiens and 10 ( 3% ) were Cx . restuans . In 2010 , we collected 271 , 594 female mosquitoes of which 30 , 261 were Culex spp . mosquitoes ( 11 . 1% ) . Of the 2 , 255 Culex spp . mosquito pools ( 23 , 068 individuals ) tested for WNV , 166 pools were positive with a peak infection rate of 42 . 6 per 1000 individuals ( 95% CI of 27 . 9–63 . 3 ) at the end of August . In 2011 , we collected 227 , 036 individual mosquitoes of which 15 , 263 were Culex spp . mosquitoes ( 6 . 7% ) . Of the 1 , 954 Culex spp . mosquito pools ( 11 , 639 individuals ) tested for WNV , 6 pools were positive with a peak infection rate of 2 . 31 per 1000 individuals ( 95% CI of 0 . 4–7 . 6 ) occurring in mid-August . In 2010 , 1 , 529 female Culex spp . mosquito pools ( 7 , 193 individuals ) were collected and tested for stable isotopes and 6 pools were enriched ( Figure 1 ) . In 2011 , 913 female Culex spp . mosquito pools ( 3 , 624 individuals ) were collected and tested for stable isotopes and 6 pools were enriched . The 12 marked pools had a mean δ15N of 285 . 4±198 . 4 . The mean δ15N of all un-enriched female Culex spp . mosquito pools was 6 . 6±0 . 04 . Based on the estimated number of enriched female Culex mosquitoes emerging from the treated catch basins , we obtained a re-capture rate of 0 . 52% in 2010 and 0 . 23% in 2011 , under the simplifying assumption that marked pools contained only one marked individual . The MDT for 2010 was 1 . 44 km and 2011 was 0 . 86 km . The closest trap containing a marked mosquito was 123 . 9 m from the release point and the farthest was 2 . 48 km ( mean = 0 . 9 km , S . E . = 0 . 19 ) . The marked female mosquito captured at 2 . 48 km occurred on September 22 , 2010 in a trap with a 68 degree bearing from the release point . During the previous night before this female was captured ( 8pm to 8am ) there was a mean wind speed of 3 . 3 km per hour ( gusts up to 16 . 7 km per hour ) with a mean bearing of 254 degrees ( WSW at 74 degrees ) , which is in the direction of the trap that captured the marked female mosquito . For the two years combined , the probability of detecting a marked mosquito at different distances from the release point was estimated using a logistic distribution function of y = ( 1+tanh ( ( −0 . 76*x – 1 . 74 ) /2 ) ) /2 ( Figure 3 ) . Based on this model , 80% of the marked mosquitoes stayed within 2 . 1 km of the release point and 90% stayed within 3 km .
This study used stable isotope enrichment to measure the dispersal of Culex spp . mosquitoes in an urban hot spot of WNV transmission . The majority of the dispersal studies of Culex mosquitoes to date have focused on Cx . quinquefasciatus or Cx . tarsalis ( reviewed by [1] ) and there are very few dispersal studies for Cx . pipiens in the U . S . Given the importance of Cx . pipiens as a primary enzootic vector of WNV in the eastern half of the U . S . north of 36° latitude [36] , [37] , the paucity of data is unfortunate . One exception is a mark-release-recapture study by Jones et al . [38] conducted on Cx . pipiens , using fluorescent dust near Washington , DC . However , this study was primarily designed to study survival . A second study by Ciota et al . [39] analyzed dispersal of Cx . pipiens in New York using a novel labeling approach that allowed mosquitoes emerging from natural container habitats to self-mark with fluorescent dust . Ciota et al . [39] report a MDT of 1 . 37 km for Cx . pipiens ( maximum flight range of 1 . 98 km ) and the current study reports a MDT of 1 . 15 km ( maximum flight range of 2 . 48 km ) , suggesting similar estimates between the two studies . Given the low detection probability of capturing a marked mosquito at large distances from the point of origin , these studies emphasize the potential for female Cx . pipiens to travel several kilometers from larval habitats of origin . However , the current dispersal estimates should be cautiously interpreted given the limited capture of marked female Cx . pipiens mosquitoes in the Ciota et al . [39] study ( n = 10 ) and in the current study ( n = 12 ) . Future studies with designs that capture more marked mosquitoes will reduce the uncertainty of such estimates . The mosquito dispersal documented in the current study is of direct relevance to the enzootic transmission of WNV and “spillover” to humans . From 2002 to 2012 , the study region defined by the seven annuli contained the geocoded addresses of 57 human cases of WNV ( Illinois Department of Public Health ) . Culex emerging from catch basins represent the same population that are part of the enzootic cycle feeding on birds [25] and are very likely also responsible for the bridge transmission to humans [40] . The study region has a radius of 3 . 5 km and includes 8 municipalities ( Alsip , Oak Lawn , Chicago Ridge , Worth , Chicago , Palos Heights , Blue Island , and Midlothian ) , in which mosquito control efforts vary considerably . Our estimates that nearly 20% of Culex mosquitoes were able to travel over 2 km from their larval environment demonstrates that the mosquito control efficacy in one small municipality can affect the level of WNV transmission and of risk of human exposure in adjacent regions . Moreover , Bertolotti et al . [41] studied the fine-scale genetic variation of WNV in this suburban Chicago study region and found significant negative spatial autocorrelation at distances beyond 4 km . This evidence of distance-limited viral transmission corresponds well with the distance female Culex mosquitoes moved in the current study . The stable isotope marking technique offers advantages and disadvantages over traditional mosquito dispersal studies . The ability to mark wild mosquitoes in natural containers with a non-invasive marker is an ideal approach to avoid artifacts of the marker or unnatural larval diet that may influence dispersal behavior [1] , [12] . Additionally , our previous laboratory experiment revealed low decay rates of 15N in mosquitoes held for 55 days post-emergence [17] . Based on the enrichment achieved in the field during the current study , the stable isotope marker should offer sufficient retention for the life of the mosquito , which is fortunate given the significance of old females for disease transmission . The advantage of the larval site label also comes with the disadvantage of not being able to control how many marked mosquitoes emerge . In the current study , the relatively wet summers of 2010 and 2011 compromised this study given that the rain events washed the water and larvae out of the catch basins ( Fig . 2; [24] , [42] . Additionally , local mosquito control efforts used S-methoprene based products to treat mosquitoes in the same catch basins we were enriching with stable isotopes , although the immediate effect on the study was not quantified . Another challenge of the larval site labeling approach is that the release of marked mosquitoes is over a prolonged time period and is thus not a defined release event . This confounds the effort to determine the age of the captured marked mosquitoes . Although this study estimated the number of marked mosquitoes emerging from the treated catch basins , the uncertainty associated with this estimate limits the ability to estimate the size of the adult mosquito population [1] . Another factor to consider with isotopic enrichment of mosquito larval environments in the field is the potential for downstream enrichment . In our case , we received many rain events during the study period , so dissolved potassium nitrate or microbes enriched with 15N would have washed into the Stony Creek at the catch basin outlets . However , we monitored aquatic invertebrates in these downstream areas and none were enriched , likely indicating that the large volume of water in the stream diluted the 15N to a concentration that was unable to bioaccumulate into the food chain and into invertebrates measurably . The stable isotope mark-capture study offers a unique perspective on mosquito dispersal and should broadly be useful for other mosquito-borne disease systems . Aedes aegypti , responsible for an estimated 50–100 million annual human cases of dengue virus [43] , is a container-breeding mosquito ideally suited for applying this stable isotope mark-capture study . Although Ae . aegypti is generally characterized as having limited dispersal [1] , [44] , the unique ability of life-long marker retention might yield a unique perspective on Ae . aegypti movement . This is especially important given current techniques of gene-driving and Wolbachia-induced population suppression or reduced vector competence aimed at the global elimination of dengue virus [8] . Understanding the distance between Anopheles spp . larval habitat and human exposure to malaria could improve control programs and mitigate disease transmission . Importantly , the enrichment of larval mosquito environments with stable isotopes is relatively inexpensive and easy to implement . Adult mosquitoes captured in traps need to be kept frozen or dried prior to the stable isotope analysis , either of which would be possible in remote field locations with limited facilities . Besides the labor and consumables to run mosquito traps , the most expensive aspect of this kind of project is the stable isotope analysis . Our project collected 2 , 442 pools between the two years that were tested at $8 per sample totaling $19 , 536 . Different stable isotope facilities charge variable amounts , and the cost per sample is dropping [18] . Given the ability to measure mosquito dispersion , including epidemiologically important old females , this tool should be useful for studying the dispersal behavior of other medically important arthropods . | The distance and direction of adult mosquitoes movement on the landscape are important processes in the spread of mosquito-borne diseases , and are critical to understand to the development of effective intervention programs . Here we present a novel approach to study adult mosquito dispersal by using stable isotope enrichment of natural larval habitats . We apply this technique in a focal hotspot of West Nile virus ( WNV ) transmission in suburban , Chicago , USA to measure dispersal of Culex spp . mosquitoes . We enriched larval mosquitoes in residential catch basins using 15N-potassium nitrate and captured adult mosquitoes in traps surrounding these catch basins . Of 10 , 817 adult female Culex mosquitoes trapped and tested for stable isotopes , 12 individuals were enriched with 15N , indicating they originated from the catch basins receiving stable isotope amendments . The mean dispersal distance was 1 . 15 km and maximum flight range was 2 . 48 km . Ninety percent of the female Culex mosquitoes stayed within 3 km of their larval habitat , which corresponds with the distance-limited genetic variation of WNV observed in this study region . This study provides new insights on the dispersal of the most important vector of WNV in the eastern United States and demonstrates the utility of stable isotope enrichment for studying the biology of mosquitoes in other disease systems . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"veterinary",
"diseases",
"urban",
"ecology",
"ecology",
"and",
"environmental",
"sciences",
"ecology",
"entomology",
"veterinary",
"virology",
"biology",
"and",
"life",
"sciences",
"zoology",
"veterinary",
"science"
] | 2014 | Dispersal of Adult Culex Mosquitoes in an Urban West Nile Virus Hotspot: A Mark-Capture Study Incorporating Stable Isotope Enrichment of Natural Larval Habitats |
Increased efforts in the assembly and analysis of connectome data are providing new insights into the principles underlying the connectivity of neural circuits . However , despite these considerable advances in connectomics , neuroanatomical data must be integrated with neurophysiological and behavioral data in order to obtain a complete picture of neural function . Due to its nearly complete wiring diagram and large behavioral repertoire , the nematode worm Caenorhaditis elegans is an ideal organism in which to explore in detail this link between neural connectivity and behavior . In this paper , we develop a neuroanatomically-grounded model of salt klinotaxis , a form of chemotaxis in which changes in orientation are directed towards the source through gradual continual adjustments . We identify a minimal klinotaxis circuit by systematically searching the C . elegans connectome for pathways linking chemosensory neurons to neck motor neurons , and prune the resulting network based on both experimental considerations and several simplifying assumptions . We then use an evolutionary algorithm to find possible values for the unknown electrophsyiological parameters in the network such that the behavioral performance of the entire model is optimized to match that of the animal . Multiple runs of the evolutionary algorithm produce an ensemble of such models . We analyze in some detail the mechanisms by which one of the best evolved circuits operates and characterize the similarities and differences between this mechanism and other solutions in the ensemble . Finally , we propose a series of experiments to determine which of these alternatives the worm may be using .
In recent years , connectomics – the assembly and analysis of comprehensive maps of neural connectivity – has been growing by leaps and bounds . Partial connectomes now exist for several organisms , including the nematode C . elegans [1] , [2] , the primate cerebral cortex of the macaque monkey [3] , the cortico-thalamic system of the cat [4] , and the mouse retina and primary visual cortex [5] . Recent efforts have increasingly been aimed at collecting data about the structural connectivity of the human brain at different levels of detail [6]–[9] . Furthermore , there have been several developments in high-throughput serial electron microscopy that continue to accelerate the rate and resolution of data collection [10] , [11] . In addition to the experimental assembly of connectome data , there has also been a growing interest in studying the large-scale network properties of these connectomes using graph theory [12]–[15] . The focus of this analysis has been on the global properties of the full network , such as small-world , scale-free properties , common motifs , degree distributions , vertex degrees , generalized eccentricities , number of complete subgraphs , clustering structures , etc . [16]–[23] . The dynamical consequences of network structure , such as signal flow and propagation of neuronal activity in response to artificial sensory stimulation [24] , has also begun to be examined . Thus , connectomics can provide important insights into the general organizational principles of nervous systems and their impact on neural activity . However , despite these considerable advances in connectomics , connectivity alone is clearly insufficient to understand the neural basis of behavior . Although network structure can certainly constrain neural activity , it does not uniquely determine it . Connectivity data must be integrated with neurophysiological and behavioral data in order to obtain a complete picture of neural function [25]–[28] . In addition , connecting a connectome to behavior requires a much finer-grained analysis of connectivity than is usually done . In addition to calculating such global network properties as degree distributions and clustering coefficients , the specific interneurons and functional pathways that connect the relevant sensory neurons to the relevant motor neurons must be identified and the electrophysiological properties of those components and connections must be characterized . The nematode worm Caenorhaditis elegans is an ideal organism in which to explore in detail the link between neural connectivity and behavior . C . elegans has been an important model system for biological research in a variety of fields including genomics , cell biology , developmental biology , and neuroscience [29]–[33] . Among its many experimental advantages are its short life cycle , compact genome , stereotypical development , ease of propagation , and simplicity of the neuromuscular system . The complete cell lineage , which is invariant between animals , has been established [29] . Most importantly for neuroscience , the C . elegans connectome for the hermaphrodite , comprising 302 neurons and over 7000 connections , is by far the most complete to date [1] . Yet , despite its relatively simple nervous system , C . elegans displays a large repertoire of behavior including locomotion , foraging , feeding , touch withdrawal , and taxes involving smell , taste , and temperature [32] , [34] , [35] . In addition , the worm exhibits more complex behaviors such as mating , social feeding , learning and memory [36]–[41] . A variety of techniques exist for characterizing and manipulating these behaviors , including automatic visual tracking [42]–[45] and the use of microfluidics to finely control the structure of artificial soil-like environments [46] , [47] . Given the availability of a nearly complete data set on its connectome and the fact that many of its behaviors have been well characterized , the major remaining obstacle to detailed analyses of the neural basis of behavior in C . elegans is a neurophysiological one . Until recently , electrophysiological analysis in C . elegans has been difficult due to its small size and pressurized body . However , substantial progress is now being made using whole-cell patch-clamp techniques [48] , calcium imaging [49] , and optogenetics [50] , and optical and electrophysiological recordings in C . elegans are becoming routine [51]–[54] . In addition , electrophysiological studies in the closely-related but larger nematode Ascaris can be used to make inferences about C . elegans electrophysiology [55] , [56] . Unfortunately , we are still a long way from knowing even which synaptic connections in this nervous system are excitatory or inhibitory , let alone the magnitudes of such connections or their time courses . Indeed , the shortage of electrophysiological data has been the main reason that few neuroanatomically-grounded models of C . elegans behavior have been undertaken , despite the fact that its connectome has been known for over 25 years ( e . g . , [57] , [58] ) . In order to address the current lack of electrophysiological data to match the comprehensive connectome data for C . elegans , one can turn to stochastic optimization techniques such as evolutionary algorithms applied to brain-body-environment models of a behavior of interest [59]–[61] . In this approach , the model is constrained to known neuroanatomy and the unknown electrophysiological parameters are evolved such that the behavioral performance of the entire model is optimized to match that of the animal . Since different runs of the evolutionary algorithm can produce different solutions with nearly the same behavioral performance , the result of this process is not a unique model , but rather an ensemble of possible models [62] . Clusters of similar solutions can then be identified within this ensemble and representative members from each cluster can be analyzed in detail as to how the observed behavior arises from the interaction between the neuroanatomically-constrained evolved neural circuit and the model body and environment in which it is embedded . The insights gained from these analyses can then be used to design experiments that distinguish between the various possibilities , focusing experimental effort where it is most crucial . The results of such experiments can in turn be used to further constrain subsequent evolutionary optimizations . To demonstrate the utility of this approach , we focus here on salt klinotaxis , a form of chemotaxis in C . elegans . Klinotaxis is defined as a strategy for moving up a gradient through gradual changes in orientation directed towards the source [63] . Salt chemotaxis [64] is one of the most studied spatial orientation behaviors in the nematode . Orientation to salt is important for C . elegans because the bacteria on which it feeds release salt into the surrounding medium as a natural part of their metabolism [35] . Salt chemotaxis also exhibits plasticity , both in the form of habituation to high salt concentrations [65] and taxis reversal after association of salt with an aversive stimulus [66] , [67] . Moreover , the sensory neurons involved in chemotaxis have been identified [68]–[70] . The behavior itself has an interesting substructure , consisting of at least two distinct strategies: klinokinesis , and the more recently discovered klinotaxis . Klinokinesis is defined as a biased random walk [71] , [72] . A number of models of klinokinesis have previously been constructed [73]–[76] . As an orientation behavior , klinotaxis fundamentally involves brain-body-environment interactions , since the salt distribution detected by chemosensory neurons drives the motion of the body , which in turn changes the perceived salt distribution . Klinotaxis is a particularly interesting spatial orientation behavior because ( unlike klinokinesis ) it exhibits state-dependence: the reactions to sensory input depend on the worm's internal state at the time of the stimulus . In this paper , we construct a neuroanatomically-grounded model of C . elegans klinotaxis by building on a previous sensorimotor model that did not include interneuronal pathways [77] . First , we identify a minimal klinotaxis circuit by systematically searching the C . elegans connectome for pathways linking chemosensory neurons to the neck motor neurons responsible for steering and then pruning the resulting network based on both experimental considerations and several simplifying assumptions . We then run a large set of evolutionary searches for the electrophysiological parameters of this minimal circuit that optimize a measure of chemotactic performance . Although this measure does not specifically reward klinotaxis , we find that a significant fraction of these searches successfully produce klinotaxis in a way consistent with both the nematode and the previous model . Next , we analyze in some detail the mechanisms by which one of the best evolved circuits operates , providing insight into how the observed sensorimotor transformations are actually implemented interneuronally . We then enlarge our analysis to characterize the similarities and differences between this mechanism and other solutions observed in the ensemble . Finally , we propose a series of experiments that can be performed to determine which of these alternatives the worm itself may be using .
In order to identify candidate klinotaxis networks , we mined the C . elegans connectome using the chemosensory neurons as the root set and the neck motor neurons as the target set . We began with the maximal network , connecting all chemosensory neurons to all neck motor neurons . We then constrained this network based on experimental evidence and simplifying assumptions until we arrived at a minimal but neuroanatomically-grounded klinotaxis network . The maximal network contains all paths between chemosensory and neck motor neurons . The C . elegans chemosensory system enables it to detect a wide variety of volatile and water-soluble cues , with a total of 8 pairs of amphid neurons that are exposed directly to chemicals in the environment: ADF , ADL , ASE , ASG , ASH , ASI , ASJ , ASK [78] , [79] . A total of 113 of the 302 C . elegans neurons are motor neurons [1] . As klinotaxis involves modulation of the side-to-side headswings , we were interested in the motor neurons that innervate the muscles in the head . We therefore only considered the 10 head and neck motor neuron classes: RIV , RIM , RMG , RMF , RMH , RMD , RME , SMB , SMD , and URA [80] . Figure 1A shows all of the paths between these two sets of neurons . Without additional constraints , the network connecting those two sets contains 90 . 72% of all neurons and 97 . 95% of all the chemical synapses and gap junctions in the connectome dataset . The klinotaxis network is clearly contained within the maximal network , but is likely to involve a smaller subset of neurons . There are several ways to constrain the maximal network . One of the simplest and most effective is to limit the length of the paths because information is likely to be lost after traveling through many neurons due to nonlinearities and noise . From the maximal network we knew the longest path between chemosensory neurons and neck motor neurons was 7 . As we limited the length of the paths , the size of the network ( as measured by the number of neurons and chemical and electrical connections ) was reduced dramatically ( Figures 1B and 1C ) . How do we decide what path length to constrain the network to ? We considered a network to be fully-connected if signals from every sensory neuron could reach every motor neuron . Within the context of klinotaxis , this is an important criterion because it ensures that all information from the environment can be used to modulate motor neurons on both sides of the worm: dorsal and ventral . For any network , there is a minimal path length that meets the fully-connected requirement . For the network connecting all chemosensory neurons to all neck motor neurons that minimum was length 5 , which included still 87 . 74% of all neurons and 92 . 98% of all the chemical synapses and gap junctions in the connectome dataset . In order to further reduce the complexity of the network , we constrained the root and target set of neurons based on experimental results . The sensory neurons required for many chemosensory responses have been defined by killing identified neurons with a laser microbeam , and testing the operated animals for their behavioral capabilities . Studies have shown that chemotaxis to sodium and chloride ions are mediated mainly by the ASE sensory neurons [68] . Simultaneous ablation of all amphid and phasmid neurons except ASE spares chemotaxis , indicating that the role of ASE in water-soluble chemotaxis is unique [68] . There have been no studies in the motor neurons involved in the gradual turning observed during forward locomotion in klinotaxis . However , from studies of locomotion [81] , we know SMB motor neurons set the amplitude of sinusoidal movement . Modulating the amplitude of sinusoidal movement at the timescale of head sweeps ( see Methods ) during forward locomotion can lead to gradual turning . This gradual turning is a likely candidate for producing the curvature in the translational direction of the worm ( i . e . , the direction of movement , see Methods ) . The unconstrained network connecting ASE to SMB is still rather large , with 88 . 41% of all neurons and 93 . 76% of all chemical synapses and gap junctions in the connectome . We reduced this network to the minimal fully-connected one by once again constraining the path length . Constraining the network to paths of length 3 ( the minimum consistent with the fully-connected criterion ) reduces it to only 23 ( 7 . 61% ) neurons and 276 ( 3 . 78% ) chemical synapses and gap junctions . This allows us to test how much of the behavior can be accounted for by the most direct paths only , with indirect paths added in subsequent iterations of the model . There is , however , a further simplification that we can make to the network . Two neurons can be connected by one or more chemical and electrical synapses . We refer to the total number of chemical and electrical synapses between two neurons as the contact number . We simplified the network one step further by setting a threshold for the number of contacts between two neurons . The assumption is that the better-connected paths are more likely to have stronger interactions . When we constrained the network to paths with 2 or more contacts , we obtained a network that contained only two interneuron classes: AIY and AIZ ( Figure 2 ) . We refer to this network as the minimal klinotaxis network . Any further simplification breaks the fully-connected criterion . The actual klinotaxis network falls between the two extreme networks: minimal ( Figure 2 ) and maximal ( Figure 1A ) . There are three main reasons why the minimal network is worth studying in more detail . First , the network is fully-connected: each of the sensory neurons can affect all of the motor neurons . Second , while the sensory neurons have been well identified , the circuits of interneurons that process sensory information are much less well characterized . The two interneurons that have been shown to be involved in klinotaxis , AIZ [63] and AIY [82] , are included in the minimal network . This is important because it was not deliberately taken into consideration in the selection of the network; rather it emerged from the experimental constraints and simplifying assumptions . Finally , starting with the minimal network allows us to test the minimum neuroanatomy necessary to produce the behavior . As soon as the network fails in some respect and as more experimental data becomes available , the constraints can be relaxed and more components included in a systematic way . To identify neuroanatomically constrained neural networks for klinotaxis in C . elegans , we generated a population of 100 different networks using an evolutionary algorithm . Networks evolved reliably after 300 generations . Out of the 100 evolutionary runs , 17 failed to produce networks capable of efficient chemotaxis ( chemotaxis index lower than 0 . 5 , see Methods ) . Of the rest , we focused only on the highest-performing subpopulation , namely those networks having a chemotaxis index ( CI ) of at least 0 . 75 ( n = 27 ) . When tested over a longer assay , this subpopulation had an average CI of 0 . 87 . All further analysis was limited to this high-performance subpopulation . Networks were evolved in chemical gradients with conical shapes , where the chemical concentration falls as a linear function of the Euclidean distance to the peak . To test for generalization we measured chemotaxis index and reliability in a Gaussian-shaped chemical gradient ( see Methods ) , which resembles the gradients used in laboratory tests of chemotaxis in C . elegans [64] . The measures in the Gaussian gradient closely matched those obtained in the conical gradient ( Table 1 ) . This experiment shows that evolved networks are not specialized for the shape of the gradient; instead , they embody a more general solution to the task of klinotaxis , making them appropriate for further study . In order to determine the mechanism by which model worms reach the peak of the gradient , we first observed how the trajectories of virtual worms varied as a function of the model's random initial bearing ( i . e . , the angle difference between the direction of translational movement and the direction of the peak of the gradient ) and then analyzed whether the trajectories met the two criteria for klinotaxis set out in previous work [77] . Figure 3 shows that model worms made a smooth turn until they were oriented in the direction of steepest ascent . Thus the output of the model was consistent with both real worm tracks during klinotaxis [63] and the previous theoretical model [77] . Klinotaxis has two necessary conditions: ( C1 ) The organism continuously adjusts its orientation toward the line of steepest ascent; ( C2 ) The adjustments in orientation are made on the basis of comparisons of the stimulus at a single point on the body as this point is swept from side to side over time . To ascertain whether networks met C1 , we plotted track-curvature , quantified in terms of turning bias ( see Methods ) , as a function of instantaneous bearing relative to the gradient peak ( Figure 4A ) . According to C1 , the algebraic sign of the turning bias should always be opposite to the sign of bearing . Figure 4A shows that this was indeed the case . To ascertain whether the optimized networks met C2 , we plotted turning bias as a function of the amplitude of the gradient in the direction normal to the worm's translational movement ( Figure 4B ) . This plot revealed that turning bias increased linearly with the amplitude of the gradient normal to the worm , as expected for a simple causal relationship between the concentration differences during head sweeps and turning bias . Furthermore , on average , turning bias was not affected by the model worm's movement in the translational direction ( black points , Figure 4C ) . This finding suggests that turning bias in the model is controlled by changes in concentration sensed during the side-to-side head sweeps , as required by C2 . Each of the data points in the gradient in the translational direction ( black points , Figure 4C ) corresponds to the average over two distinct bearings . For example , +90 and −90 degrees both have 0 translational gradient . Their corresponding turning biases are nonzero , equal in magnitude but opposite in sign . So although the translational gradient does not influence the turning bias on average , when we studied the different cases more systematically we found some information in the translational direction: the magnitude of the turns were larger for negative translational gradients than for positive translational gradients ( gray points , Figure 4C ) . There is a significant difference in the turning bias of the model worm when moving up the gradient at an angle than when moving down the gradient at that same angle ( see red points , Figure 4C ) . Therefore , the magnitudes of the corrections are larger when the worm is heading away from the peak than when the worm is heading towards the peak . Although this is not a requirement of klinotaxis , it is an efficient component to the strategy exploited by the evolved model worms . The sinusoidal relationship between turning bias and bearing , together with the linear relationship between turning bias and the normal component of the gradient , are qualitatively similar to the relationships observed in studies of klinotaxis in real worms [63] . This similarity is significant for two reasons . First , as we did not explicitly include selection criteria in the evolutionary algorithm to approximate these features , this similarity is an emergent property of the evolved networks . Second , the resemblance suggests that the model employs a klinotaxis strategy similar to the one used by real worms , making the model presented here especially appropriate for the generation of testable predictions concerning how the biological network functions . If we consider only the transformation that occurs between the sensors and motors , it is possible to compare the results of the neuroanatomically-grounded model with the previous simplified model . In order to do this , we studied how changes in concentration are transformed into changes in motor responses , as reflected by the worm's orientation , using single stepwise changes in concentration of different magnitudes at different points in the locomotion phase ( increments in concentration: upsteps , Figure 5A; decrements in concentration: downsteps , Figure 5B ) . Orientation responses were expressed in terms of turning bias , which was computed over a complete cycle of locomotion following the concentration step . We observed that turning bias varied as a sinusoidal function of the phase of locomotion at which the concentration change occurred ( Figures 5A and 5B ) . This function had extrema near phases 0 and π , where the instantaneous velocity vector ( v , see Methods ) diverges most from the unbiased translational vector ( u , the worm's direction of movement in the absence of external input , see Methods ) , and minima near phases of π/2 and 3π/2 , where the instantaneous velocity vector diverges least from the unbiased translational vector . In the context of klinotaxis on a gradient , the instantaneous velocity vector at the time of an upstep signals the direction of the peak implied by such a step , whereas the instantaneous velocity vector at the time of a downstep signals the direction opposite to the peak . Thus , the model worm corrects its orientation relative to discrepancies between its velocity vector and the direction of the peak throughout the locomotion phase . The amplitude of the orientation response was proportional to the amplitude of the concentration step ( Figure 5 ) . This proportionality is important when the changes in concentration produced during dorsal and ventral sweeps have the same sign but different magnitudes . The sensorimotor transformation in Figure 5 is qualitatively similar to the previous theoretical model [77] , which suggests the principles of operation of the neuroanatomical network are consistent with the simpler model . A neuroanatomically-grounded model allows us to go beyond overall sensorimotor transformations to examine the interneuronal implementation of klinotaxis . In this section we analyze in some detail one of the best evolved networks ( Figure 2 ) , a representative of many of the rest of the high-scoring subpopulation . The network's performance depends on the parameters that it evolved , but the solution is not brittle: there is a graceful degradation of the performance as parameters are independently perturbed away from the evolved values ( see Figure S1 ) . In order to understand how changes in concentration travel through the network , we studied how the synaptic potential ( henceforth , output ) of the neuron changed as a function of step changes in concentration of different sign and magnitude over the full spectrum that the model worms experience during klinotaxis runs ( Figure 6 ) . The dynamics of the chemosensory neurons follow directly from their definition ( see Methods ) : ASER and ASEL react only to downsteps and upsteps in concentration , respectively ( Figures 6A1 and 6A2 ) . The connections between the chemosensory cells ASEL and ASER and the first interneuron class ( AIY ) are excitatory and inhibitory , respectively ( Figure 2 ) . Therefore , upsteps in concentration move the membrane potential of both AIY cells upward and downsteps in concentration move the membrane potential of both AIY cells downward ( henceforth , we will refer to the membrane potential as the activation of the neuron , with positive-going changes as increases in activation and negative-going changes as decreases in activation ) . How each AIY cell reacts to changes in concentration is a function of their bias parameter in relation to the strength and sign of the incoming chemical synapses from the chemosensory neurons . Because of the nonlinearity of the input-output relation ( see Methods , Eq . 3 ) , each AIY cell can only respond to changes in concentration within a certain range ( henceforth , sensitivity ) . Also , given that the parameters of the network are not constrained to be left/right symmetric , the range of sensitivities of the two AIY cells can be different . In this network , AIYL is sensitive to small changes in concentration , positive or negative ( Figure 6B1 ) ; whereas AIYR , due to a strong negative bias parameter ( Figure 2 ) , is only sensitive to large increases in concentration ( Figure 6B2 ) . Crucially , the ranges of sensitivities of the two cells are complementary , such that together they cover a broader range of the possible stimuli than individually . The gap junction between cells drives the activation of the neurons closer together: the lower the resistance , the bigger the effect . The effect , however , is not always noticeable in output space due to the nonlinearity of the input-output relation: changes to the activation of the neuron can be masked by the saturation of the input-output relation . This is the case for the gap junction between the AIY cells in this network ( Figures 6B1 and 6B2 ) . Indeed , blocking the gap junction does not alter the dynamics of the output of the two interneurons significantly . From this experiment we conclude that the sensitivities of the AIY cells depend mainly on the incoming chemical synapses from ASE . Neuroanatomically , AIZ cells only receive a chemical synapse from the AIY cell directly upstream . When we combine this with AIZ's own nonlinear response , we would expect the range of sensitivities to different steps in concentration to be a reduced set from AIY's . But this is not what we observe ( Figure 6C ) . Unlike in the AIY interneurons , the gap junction between these AIZ cells plays an important role . The effect can be seen in AIZR best: even though AIYR is not sensitive to downsteps ( Figure 6B2 ) , AIZR is sensitive to them ( Figure 6C2 ) . This information is transferred not via the chemical synapses downstream , but via the lateral gap junction with AIZL . The AIZ gap junction plays a crucial role in redistributing and broadening the sensitivities to the changes in concentration between the left and right cells . Unlike interneurons , motor neurons additionally receive an oscillatory input . Therefore , how they react to step changes in concentration depends on the phase of the locomotion cycle ( Figure 6D–E ) in which a change occurs . In order to understand their operation , we first consider their dynamics in the absence of sensory input . Figure 7 shows the steady-state input-output ( SSIO ) curve of the left and right motor neurons . The oscillatory input drives the motor neurons around the red trajectory , with dorsal and ventral cells out of phase . The key feature of the motor neurons is that the sensitive part of the SSIO curve is such that when one of the dorsal motor neuron is in the sensitive area of the curve , the ventral one is not , and vice versa . Even though the SSIO curves for the left ( Figure 7A ) and right ( Figure 7B ) pair of motor neurons are different , they share the same principles: ( a ) the sensitive region is biased with respect to the oscillatory range , and ( b ) increases in concentration move the input towards the most saturated part of the range; decreases in concentration move the input towards the sensitive part of the range . This is consistent with the principles observed in the motor neurons of the previous sensorimotor-only model [77] . It is the asymmetry in the location of the sensitive area in the SSIO curves of the motor neurons that allows for state-dependence in the network ( Figure 7 ) . We analyze first the left pair of motor neurons . When the concentration increases , AIZL's activation also increases , and for some range of magnitudes the neuron output increases as well ( blue traces Figure 6C1 ) . Given the inhibitory connection to the motor neurons ( Figure 2 ) , SMBDL and SMBVL receive less input as a consequence . As the dorsal and ventral neurons are out of phase , one is in the sensitive region of its SSIO curve and the other is not . Therefore , the output of one of the motor neurons decreases and the other one stays the same ( compare blue trace in Figure 6D1 to Figure 6E1 ) . This decreases the difference between the output of the dorsal and ventral motor neurons , which ultimately decreases the worm's turning . When the concentration decreases , AIZL's activation also decreases ( red traces , Figure 6C1 ) , and for some range of these changes in concentration the neuron output decreases as well . In this case , the motor neurons receive more input , and as a consequence of the bias in sensitivity , the output of one of the motor neurons increases and the other stays saturated ( compare red trace in Figure 6D1 to Figure 6E1 ) . This increases the difference between the output of the dorsal and ventral motor neurons , which ultimately increases the worm's turning . Despite the differences in evolved parameter values , the effect is similar in the right motor neurons ( Figures 6D2 and 6E2 ) . When the concentration increases , AIZL's activation also increases , and for some range of magnitudes the neuron output increases as well ( blue traces , Figure 6C2 ) . Given the excitatory connection to the motor neurons ( Figure 2 ) , SMBDR and SMBVR receive more input as a consequence . As the dorsal and ventral neurons are out of phase , one is in the sensitive region and the other is not . Therefore , the output of one of the motor neurons increases and the other one stays the same ( compare blue trace in Figure 6D2 to Figure 6E2 ) . This decreases the difference between the output of the dorsal and ventral motor neurons , which ultimately decreases the worm's turning . When the concentration decreases , AIZR's activation also decreases ( red traces , Figure 6C2 ) , and for some range of these changes in concentration the neuron output decreases as well . In this case , the motor neurons receive less input , and as a consequence of the bias in sensitivity , the output of one of the motor neurons decreases and the other stays saturated ( compare red trace in Figure 6D2 to Figure 6F2 ) . This increases the difference between the output of the dorsal and ventral motor neurons , which ultimately increases the worm's turning . In order to understand the range over which each neuron is sensitive to changes in concentration , we visualized the difference in output between the no input and input conditions as a function of the magnitude and polarity of stepwise changes in concentration ( Figure 8 ) . In AIY , the sensitive regions of the neuron output for the left and right cells are different ( Figure 8A ) : AIYL is most sensitive to small negative and positive steps whereas AIYR is most sensitive to larger positive steps . We can also use this analysis to study the role of the gap junction , by blocking it and comparing the changes in sensitivity to the unblocked condition . We observed almost no change in the sensitivities of the left and right cells when the gap junction is blocked ( dashed curves , Figure 8A ) . In contrast , for AIZ , the sensitive regions of the neuron output for both the left and right are similar ( Figure 8B ) : both are sensitive to small upsteps and downsteps , though AIZL is still most sensitive to small negative and positive steps and AIZR is biased towards larger positive steps . The difference in the range of sensitivity for the AIZ cells when the gap junction is blocked is dramatic ( dashed curves , Figure 8B ) . The surfaces in Figures 8C and 8D allow us to visualize how the sensitivities change as a function of locomotion phase , in addition to the magnitude and polarity of the step change in concentration . As for the interneurons , the main differences between the left and right pair of motor neurons is in the range over which they are sensitive to changes in concentration . The left pair of motor neurons is most sensitive to small downsteps and upsteps ( Figure 8C ) , whereas the right pair is more sensitive to large upsteps ( Figure 8D ) . This difference stems from the combination of sensitivities of the AIZ cells upstream ( Figure 8B ) and the dynamics of the motor neuron ( Figure 7 ) . The dorsal and ventral , left and right motor neurons add together to affect the dorsal and ventral muscles , respectively . Therefore , the different ranges of sensitivity are ultimately combined at the level of the dorsal and ventral muscles . The final issue in analyzing the mechanism of klinotaxis in this model circuit is to consider the behavioral effects of motor activity on the orientation of the body . How do the step changes in concentration result in changes in the translational direction ? ( cf . Figure 5 ) . In order to answer this question , we analyzed the orientation responses produced by single stepwise changes in concentration ( upsteps , Figure 9A; downsteps , Figure 9B ) . Upsteps activate the ON cell , which reduces turning angle . The effect of the turning angle reduction on the worm's translational direction is dependent on the phase of locomotion . We consider two representative phases . First , an upstep at the midpoint of a ventral/dorsal head sweep ( Figure 9A1 , points a/c ) : turning angle is decreased for approximately the duration of a head sweep ( 2 . 1 secs , cf . Fig . 6 ) . This persistent reduction in turning angle attenuates the ensuing dorsal/ventral turn; with the result that the worm's translational velocity vector rotates ventrally/dorsally ( red/green trajectory , vector ua/uc vs . u ) . Both the dorsal rotation at point a and the ventral rotation at point c are appropriate orientation responses because the model worm turns in the direction of its instantaneous velocity vector at the time of the increase in concentration . Second , an upstep at the ventral/dorsal maximum ( point b/d ) : no rotation occurs because the decrease in turning angle attenuates parts of the dorsal and ventral turns almost equally ( blue/yellow trajectory ) . The absence of rotation at points b and d is appropriate because the model worm's instantaneous velocity vector at the time of the increase in concentration was the same as its translational direction . Downsteps activate the OFF cell , which increases turning angle . The analysis is the same as with upsteps , except that the turning increases instead ( Figure 9B1 ) . Crucially , the ventral rotation at point a and the dorsal rotation at point c are appropriate orientation responses because the model worm turns away from the direction of its instantaneous velocity vector at the time of the decrease in concentration ( red/green trajectory , vector ua/uc vs . u ) . To obtain a more complete understanding of how the simple rules for predicting changes in turning angle lead to correct orientation responses , we expanded the analysis of Figures 9A1 and 9B1 to include steps in concentration not only at points a–d , but also at the points in between ( upsteps , Figure 9A2; downsteps Figure 9B2; cf . Figure 5 ) . This mechanism depends on three basic principles . ( 1 ) The two motor neurons are biased such that when one motor neuron is sensitive to sensory input , the other is not . ( 2 ) The signs of connections from sensory neurons to motor neurons are adjusted with respect to motor neuron bias such that ON cell activation reduces the curvature of the worm's path , whereas OFF cell activation increases the curvature of the worm's path . ( 3 ) The dynamics of sensory responses are adjusted so that changes in curvature are transient , lasting for approximately the duration of a head sweep . As a result , changes in curvature cause the worm's path to veer toward concentration increases and away from concentration decreases , unless the worm's head is moving in the direction of the gradient peak at the time the concentration change is encountered . These principles are similar to those found in the previous sensorimotor-only model [77] . The novelty of the analysis here lies in the implementation of the mechanism at the interneuronal level . The interneurons on the left and right side of the network show a certain amount of shared information about the changes in concentration , but also some degree of specialization: some changes in concentration are sensed by the left side of the network only and some changes in concentration are sensed by the right side of the network only . This feature allows the network to extend the coverage of the range of possible changes in concentration . Finally , the gap junctions between the interneurons can play a key role in distributing the sensitivities . The result of evolutionary optimization is not a unique model , but rather an ensemble of possible models . Given the underconstrained nature of optimization ( due to the lack of electrophysiological data and the possibility of variability in both the available experimental data and behavior across individuals ) , understanding the structure of this ensemble is a key aspect of the approach we have taken in this paper . Studying an ensemble of models with different underlying parameters and similar behaviors provides opportunities to explore different possible mechanisms that could be operating in the worm . How does the mechanism that evolved in the network analyzed generalize to the rest of the high-scoring population ? We examined the similarities and differences between , on the one hand , the evolved electrophysiological parameters and the interneuronal dynamics of the best evolved network and , on the other hand , the rest of the high-scoring subpopulation .
In this paper , we used evolutionary algorithms to set the unknown electrophysiological parameters of a minimal salt klinotaxis circuit extracted from the C . elegans connectome such that a simple brain-body-environment model of the worm exhibited efficient chemotaxis . We first analyzed in detail the operation of a high-performing solution and then explored the extent to which similar principles were operating throughout the ensemble of high-performing solutions . Due to the underconstrained nature of the problem , the particular parameter sets found by different runs of the evolutionary algorithm varied widely even when producing very similar chemotaxis behavior . However , several broad classes of patterns observed in the ensemble of high-performing solutions suggest new experiments that need to be carried out in order to select between these possibilities and further refine the model . First , a more systematic analysis of turning as a function of the gradient in the translational direction in the worm is required . In our models , we observed larger magnitude turns for negative translational gradients than for positive translational gradients across all successful networks ( Figure 4C ) . Although this was not a criteria set forth in our definition of klinotaxis , it is an efficient supplement to the strategy , exploited by the evolved model worms . Essentially , the magnitudes of the corrections are larger when the worm is heading away from the peak than when the worm is heading towards the peak . This has not yet been analyzed in the worm . Second , ablations of individual AIY and AIZ cells ( as opposed to both members of a class simultaneously ) should be performed . The variance in klinotaxis performance for AIY-ablated model worms was higher than for AIZ-ablated model worms in the successful population . Our analysis revealed that the distribution of the information about the sensory stimuli was better distributed across AIZ cells and AIY cells , where sometimes all of the information resided in only one of the cells , and none of the information in the other one . This is influenced by the neuroanatomy of the circuit , where AIY cells have access to information from both the ON cell and the OFF cell , enabling an individual cell to integrate most of the necessary information to perform the task . AIZ cells , on the other hand , receive already integrated information from the AIY cells , and have to distribute the information to other AIZ cell , via the gap junction , in order to have an effect on all four motor neurons . So far klinotaxis experiments involving AIY and AIZ have been performed only while ablating the left and right cells simultaneously [63] . Unfortunately , simulated ablations of the whole class do not make sense in this minimal model because there are no alternative pathways , thus making it difficult to compare the existing experimental results to simulated ablations . Nevertheless , the results of the model are consistent with the ablation experiments in the real worm , where killing AIY also has higher variance than killing AIZ [63] . The variation , of course , could be attributed to different reasons in the model and the worm . The variance in the simulation data arises from the different parameters of each of the successful klinotaxis networks . There are at least two possible sources of variation in the worm data: experimental ‘noise’ created by variability in the observational and experimental techniques , and natural variability in the worms [31] . Third , a more refined set of ablation experiments could also test which pattern of AIY sensitivities that we observed in our model ensemble is utilized by C . elegans . Analysis of the dynamics of model AIY interneurons revealed three types of successful networks within the population: networks where only one AIY cell was sensitive to the full range of changes in concentration , networks where left and right cells had a different range of sensitivities to changes in concentration , and networks where both cells covered the full range of changes in concentration . Ablation experiments to individual AIY cells could distinguish between all three scenarios in the worm . In the first scenario , ablating one of the AIY cells would lead to a major decrease in performance , whereas ablating the other AIY cell would have no effect on performance . In the second scenario , killing either AIY cell individually would not decrease performance entirely , but the behavioral deficiency between ablating left and right cells would be significantly different . In the third scenario , the behavioral deficiency when ablating left and right cells would be similar . The majority of networks exhibit a different range of sensitivity over the sensory input in the left and right cells . This is similar to what has been observed in ASE [69] . Our analysis could help distinguish the AIY cells further . A study of the resulting behavioral pattern of left and right ablations could allow us to infer the difference in the range of sensitivities between the two cells . Networks without sensitivity to upsteps produce less efficient spiral tracks towards the peak , but just as reliably . On the other hand , networks without sensitivity to downsteps produce efficient paths straight towards the peak , but only for some orientations . Fourth , any additional physiological analysis of the relationship between changes in concentration and interneuronal activity would help to narrow down the arithmetic sign possibilities in this circuit . From our analysis , we know the paths from ASEL to the motor neurons and from ASER to the motor neurons are antagonistic: an upstep in concentration increases the net input to the neck motor neurons , and a downstep in concentration decreases the net input , or vice versa . Crucially , we know from our analysis which of the two it ought to be based on how the sensitive region of the motor neurons is biased with respect to the oscillatory range . Essentially , the polarities of the connections must be such that an increase in concentration shifts the input towards the most saturated part of the range and a decrease in concentration shifts the input towards the sensitive part of the range . Because there are three chemical synapses between the chemosensory neuron and the motor neuron , there are a total of 8 possibilities . Of course , direct electrophysiological study of these connections in the worm would be ideal , but other experimental possibilities exist . For example , characterizing the sensitivity of the motor neurons with respect to the locomotory oscillation would help to narrow down the set of possible polarities that can result in successful klinotaxis to half . Fifth , blocking individual gap junctions between the two interneuron classes , AIY and AIZ , would provide insight into the relationship between number of contacts and functionality . Our analysis revealed that the functional role of the AIZ gap junction was more crucial than the role of the AIY gap junction in the successful population . The anatomy of the chemical synapses in the network is likely to be playing a key role . AIY has access to the full range of input from the incoming chemical synapses of ASER and ASEL , whereas AIZ cells depend only on the nonlinear output of the AIY cells upstream from it . This was also reflected in the evolved strength of the gap junctions in the successful population , with a median ratio of 0 . 69 between the strength of the AIY/AIZ . This result is roughly consistent with the corresponding ratio derived from the known neuroanatomy in the worm: the AIY gap junction has one contact whereas the AIZ gap junction has two contacts [1] . Not all of the experiments we have proposed are equally feasible . The first experiment could be performed using detailed behavioral data from chemotaxis assays . Experiments 2 through 4 require individual cellular ablations that are currently possible . The last experiment , testing the role of the gap junction and the way in which information is shared in the interneurons , AIY and AIZ , is not currently feasible in C . elegans . It is important to reiterate that the current minimal model is not the ultimate model of the actual klinotaxis circuit , only a useful starting point for the modeling-experimental cycle . There are several directions in which our minimal klinotaxis model could be extended . First , as limitations of this minimal model are encountered , additional interneuronal pathways should be considered . For example , Figure 13 shows the circuit obtained by relaxing the contact threshold constraint in our C . elegans connectome search; it contains an additional 13 neurons and 18 feedforward paths between ASE and SMB . Similar extensions could be obtained by relaxing the path length constraint . Second , as additional experimental observations are made , neurons should be added or deleted from the network . For example , a recent study shows that RIA encodes head movement [83] . A subsequent model could explore the paths between ASE and the neck motor neurons , through RIA . Third , more biological realism can be introduced to the model as necessary to account for new experimental results . As more neurophysiological information becomes available about C . elegans neurons , neuromodulation , and the relationship between contact number and strength of interaction [48] , [69] , [84] , [85] , more biophysically-grounded neural models can be employed . In addition , more realistic models of the body musculature could be employed [86] , [87] . Finally , a neuroanatomically-grounded model of C . elegans klinotaxis could serve as a springboard for other future modeling efforts , including the interaction of klinotaxis with klinokinesis [64] , [71] , its integration with locomotion [87] , associative learning [66] , [88] , and its relationship with other taxes such as odortaxis and thermotaxis [78] , [89] . In the long run , a model such as the one we have described here may represent an initial step along a path to the ultimate goal of having a brain-body-environment model of a complete animal . In this paper , we have shown how a stochastic optimization technique such as evolutionary algorithms can be used as a kind of semi-automated hypothesis generator . By combining known neuroanatomical constraints from the C . elegans connectome with reasoned simplifications of its body and environment , optimization can fill in missing electrophysiological parameters in plausible ways so as to produce worm-like klinotaxis . Since our knowledge of any biological system is always partial , this methodology can be applied more generally: optimization can be used to explore the possibilities for what is unknown in ways consistent with what is known . A key feature of this approach is that the result is not a unique model , but rather an ensemble of models that are consistent with current knowledge of the system of interest . By studying the structure of this ensemble , one can formulate new experiments that can distinguish between the various classes of possibilities . The results of these experiments can then be used as additional constraints for subsequent optimizations in an iterative cycle of model refinement . In this way , productive interactions between modeling and experiment can begin very early in the lifecycle of a biological modeling project , when very little data is available , and carry through to a mature project when the system has been very well-characterized experimentally .
Neuronal connectivity data for C . elegans was assembled by White et al . , [1] from 5 animals , and later revisited [2] , [24] . For each neuron , data exists for the total number of chemical synapses . There is also information about the synapse type: gap junction , where there is no directionality; an unambiguous chemical synapse from one neuron to another , also called a monadic synapse; and a joint chemical synapse between one neuron and more than one recipient , which can be dyadic or triadic . The C . elegans connectome data set is not 100% complete . Connectivity data for 39 of the 302 neurons is partially missing , including the most posterior 21 of the 75 motor neurons . Current theoretical and experimental studies are aimed at estimating and reconstructing missing data [90] , [91] . The klinotaxis network focuses on neurons that are in the head and neck , which is where the data is most complete . While it is possible for the missing data to change the results , there is no reason to wait for its full reconstruction to begin to develop the methods of analysis to link the connectome to behavior . In order to search the connectome , we developed code that finds paths connecting two sets of neurons in the C . elegans connectome . Using existing online tools ( e . g . , [92] , [93] ) , it is possible to manually examine the connections between pairs of neurons . However , no tool was available to systematically search the connectome for all pathways connecting two sets of neurons that satisfy a flexible set of search criteria . Our code recursively performs a breadth-first search of the C . elegans connectome database from a Root Set of sensory neurons to a Target Set of motor neurons subject to a set of constraints . At each step of the search , two constraints were applied . A Depth Limit constraint terminated the search at a specified pathway length . A Contact Threshold constraint only considered connections that involved more than a given number of chemical synapses or gap junctions . Changes in salt concentration were encoded by ON and OFF chemosensory cells [69] using an instantaneous function of a derivative operator applied to the recent history of attractant concentration [94]: ( 1 ) where c ( t ) is the concentration at time t , and N and M are the durations of the two intervals over which the concentration is averaged . In response to a concentration step of infinite duration at , yields a linear rise to a peak at , and a linear decay to base line at ; accordingly , N and M are referred to as the “rise time” and the “decay time” of the sensory neurons . In the case of the OFF cell , , the signs were inverted so that decreases in concentration yielded positive activations . In both ON cells and OFF cells negative activations were set to zero . Interneurons were modeled as passive , isopotential nodes according to: ( 2 ) where y represents the membrane potential ( or neuron activation ) relative to the resting potential ( thus y can assume positive and negative values ) , is the time-constant , the first sum term is the input from the chemical synapses , the second sum term is the input from the electrical synapses , and the third term represents external input to the neuron . The model assumed chemical synapses release neurotransmitter tonically and that steady-state postsynaptic voltage is a sigmoidal function of presynaptic voltage [56]: ( 3 ) where is the synaptic potential or output of the neuron . The chemical synapse has two parameters: is a bias term that shifts the range of sensitivity of the output function , and represents the strength of the chemical synapse . We can interpret the strength as the product of the number and size of the chemical synapses . The importance of electrical synapses has been shown in several C . elegans behaviors , including locomotion and touch-withdrawal behaviors [95] , [96] . Electrical synapses are generally described as rectifying ( current passes preferentially in one direction ) or non-rectifying ( current is passed equally efficiently in both directions ) . Unfortunately , there is no concrete evidence about the nature of electrical synapses in C . elegans . Until more evidence is available , and in line with previous models [97] , the model assumes electrical synapses in C . elegans are nonrectifying , with as a conduct conductance between cell i and j ( >0 ) . Neck motor neurons were modeled similar to the interneurons , except with self-connections . Biophysically , self-connections can be interpreted as the voltage dependence of inward currents underlying the graded regenerative potentials that are characteristic of several C . elegans neurons , including the neck motor neurons [48] , [84] . The neck motor neurons also receive an additional input from an oscillatory component , . ( 4 ) where represents the strength of the connection from the oscillatory component . Because the cellular mechanism by which oscillations are generated during locomotion in C . elegans is unclear , we did not explicitly model this mechanism; instead , we represented its effect as a sine wave , , with T = 4 . 2 sec , the duration of a one cycle of locomotion on agar [73] . The dorsal and ventral motor neurons receive out of phase input from the oscillatory component . In sinusoidal locomotion ( without slip ) , each body segment follows the one anterior to it . The worm was therefore represented as a single point ( x , y ) with instantaneous velocity v . The angular direction of movement μ was measured relative to the positive x-axis ( Figure 14A ) . The biomechanics of locomotion were represented in idealized fashion , with two main assumptions . ( 1 ) Neck muscle length was proportional to motor neuron output . ( 2 ) The turning angle ( Figure 14B ) was proportional to the difference in muscle length . After combining constants of proportionality , this gives: ( 5 ) where , and are activations of the dorsal and ventral neck motor neurons , is the strength of the connection from motor neurons to muscles . It follows that the model worm's position is updated as: ( 6 ) where v is a constant speed of 0 . 022 cm/s [73] . To include pirouettes , the model worm's orientation was randomized with an average frequency of 0 . 033 Hz , which matches the baseline frequency of pirouettes in real worms [71] . In analysis , pirouette frequency was set to zero . We did not explicitly model the mechanism responsible for generating the oscillations for forward thrust; instead , we represented its effect as a sine wave . Movement in real worms cannot occur without the thrust generated by undulations [98]; to implement this constraint , the velocity of the model worm was set to zero unless undulations were present . The gradient during a typical salt chemotaxis assay has a Gaussian shape [64] . In the context of evolution , however , Gaussian gradients are problematic because local steepness is systematically related to distance from the gradient peak . To avoid this problem , we used conical gradients of varying steepness during evolution . Accordingly , attractant concentration was proportional to the Euclidean distance from the gradient peak , ( 7 ) where determines the steepness of the gradient . The parameters of the model were evolved using a genetic algorithm [99] . The optimization algorithm was run for populations of 60 individuals . We evolved the following parameters ( ranges are shown in brackets ) : [1 , 3]; , , , and [−15 , 15]; [0 , 15] , N and M [0 . 1 , 4 . 2] . Network parameters were symmetrical across the dorsal/ventral midline . Parameters were encoded in a 20-element vector of real-values between [−1 , 1]; when needed , parameters were linearly mapped to their corresponding ranges . Each time the algorithm was run , individuals were initialized by random selection from the range of each parameter . Populations were evolved for 300 generations . At the end of a run , the parameters of the best performing individual were stored for later analysis . The algorithm was run 100 times ( using different random seeds ) , yielding 100 distinct networks . Fitness was evaluated in simulated chemotaxis assays . At the start of each assay , the model worm was placed with a random orientation at a point 4 . 5 cm from the peak of the gradient and motor neuron activations were randomized over the range [0 , 1] . Gradient steepness α was randomized over the range [−0 . 38 , −0 . 01] . The fitness score was quantified in terms of a chemotaxis index CI defined as the time average of the distance to the peak of the gradient , ( 8 ) where h ( t ) is the Euclidean distance to the peak , h ( 0 ) is the model worm's initial distance from the peak ( 4 . 5 cm ) , and T is the total simulated assay time ( 500 sec ) . For simplicity , negative CI values were set to zero . The fitness of an individual was defined as the average CI over 50 assays . A common measure of chemotaxis performance in C . elegans salt chemotaxis assays is the proportion of worms that reach the gradient peak [35] , which we will refer to as reliability . In our simulations , we defined the peak to be a region enclosed by a circle with a radius of 0 . 1 cm centered on the peak . A locomotion cycle consists of alternating ventral ( solid trajectory , Figure 14C ) and dorsal ( dashed trajectory , Figure 14C ) head sweeps . The principal orientation vector used was the direction of translation , defined by any pair of points on a trajectory separated by a phase difference of , i . e . one cycle of locomotion ( Figure 14C ) . The vector 90 degrees counter-clockwise from the direction of translation is the normal direction , which was used to quantify the gradient as sensed by the model worm over a single head sweep . The angle between the line of steepest ascent and the direction of translation is the model worm's bearing . The turning bias was defined as the sum of the turning angle over one cycle of locomotion . The evolved model neurons are sensitive to sensory input over a certain range . The coverage of a neuron was defined as the proportion of sensory input , over a specified range , where the output of the model neuron was substantially different from the output of the neuron over nearby stimuli . The range was determined by the sensory input observed during a usual klinotaxis run , ±0 . 02 . The average output of the neuron was recorded for different steps in concentration over that range , in intervals of 5×10−4 . Each point was considered ‘covered’ only if the average output of the neuron for that input was sufficiently different ( greater than 1×10−4 ) than the average output during the previous step in concentration . | Maps of the connections between neurons are being assembled for several organisms , including humans . But connectivity alone is insufficient for understanding the mechanisms of behavior . Nowhere is this more obvious than in the nematode C . elegans , where the nearly complete connectome has been available for over 25 years yet little is known about the neural basis of most of its behavior . Here we combine known neuroanatomical constraints from the C . elegans connectome with a simplified body and environment , and use optimization techniques to fill in the missing electrophysiological parameters in plausible ways so as to produce worm-like behavior . We focus on one spatial orientation behavior , where the reactions to sensory input depend on the worm's internal state at the time of the stimulus: salt klinotaxis . By exploring the possibilities for what is unknown in ways that are consistent with what is known , we generate an ensemble of hypotheses about the neural basis of this behavior . Studying the structure of this ensemble , we formulate new experiments that can distinguish between the various hypotheses . This methodology is likely to accelerate the discovery and understanding of the biological circuitry underlying the behavior of interest , before a complete electrophysiological characterization is available . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"neural",
"networks",
"neuroscience",
"animal",
"behavior",
"computational",
"neuroscience",
"theoretical",
"biology",
"biology",
"connectomics",
"computer",
"science",
"computer",
"modeling",
"neuroanatomy",
"computational",
"biology",
"evolutionary",
"biology",
"behavioral"... | 2013 | Connecting a Connectome to Behavior: An Ensemble of Neuroanatomical Models of C. elegans Klinotaxis |
Oral cholera vaccines ( OCVs ) are being increasingly employed , but current killed formulations generally require multiple doses and lack efficacy in young children . We recently developed a new live-attenuated OCV candidate ( HaitiV ) derived from a Vibrio cholerae strain isolated during the 2010 Haiti cholera epidemic . HaitiV exhibited an unexpected probiotic-like activity in infant rabbits , preventing intestinal colonization and disease by wild-type V . cholerae before the onset of adaptive immunity . However , it remained unknown whether HaitiV would behave similarly to other OCVs to stimulate adaptive immunity against V . cholerae . Here , we orally immunized adult germ-free female mice to test HaitiV’s immunogenicity . HaitiV safely and stably colonized vaccinated mice and induced known adaptive immune correlates of cholera protection within 14 days of administration . Pups born to immunized mice were protected against lethal challenges of both homologous and heterologous V . cholerae strains . Cross-fostering experiments revealed that protection was not dependent on vaccine colonization in or transmission to the pups . These findings demonstrate the protective immunogenicity of HaitiV and support its development as a new tool for limiting cholera .
The bacterial pathogen Vibrio cholerae causes the severe human diarrheal disease cholera , a potentially fatal illness characterized by rapid-onset of fluid loss and dehydration . Recent estimates place the global burden of cholera at ~3 million cases per year , and over 1 . 3 billion people are at risk of this disease [1] . V . cholerae proliferates in the small intestine and produces cholera toxin ( CT ) , which leads to water and electrolyte secretion into the intestinal lumen [2] . The O1 serogroup of V . cholerae causes virtually all epidemic cholera . This serogroup includes two serotypes , Inaba and Ogawa , whose LPS structures differ by a single methyl group on the terminal O-antigen sugar [3] . Serologic and epidemiologic studies have established the existence of extensive serotype cross-reactivity and -protectivity , although immunogenicity and protection is highest to the homologous serotype [4–7] . Toxigenic O1 strains are divided into two major biotypes , classical and El Tor , but the former has not been isolated in over a decade and is thought to be extinct [8] . Ongoing evolution of El Tor V . cholerae has given rise to variant El Tor strains , which are distinguishable from earlier strains by a variety of features , including the expression of non-canonical ctxB alleles that may impact disease severity in afflicted patients [4 , 9 , 10] . These contemporary strains , such as the ctxB7-expressing V . cholerae strain responsible for the 2010 Haitian cholera epidemic , are thought to be the globally dominant cause of cholera [10–12] . Currently , serogroup O139 isolates only cause sporadic disease [13] . Notably , antibodies ( or immune responses ) targeting the O1 O-antigen do not protect against O139 challenge and vice versa [14–16] . Oral cholera vaccines ( OCVs ) have recently become widely accepted as a tool for cholera control [17] . Vaccines are a potent method to combat cholera due to their ability to both directly and indirectly reduce disease and transmission [18] . Killed multivalent whole-cell OCVs , such as Shanchol , have shown promise both to prevent disease in endemic regions and as reactive agents to limit cholera during epidemics [19] . However , killed OCVs tend to be less effective at eliciting protective immunity in young children ( <5 years old ) , who are most susceptible to cholera [20 , 21] . Additionally , these vaccines typically require two doses over the span of several weeks , although recent studies suggest that a single dose may still lead to moderate protection [20 , 22 , 23] . There is no live-attenuated OCV licensed for use in cholera-endemic regions . The only clinically available live-attenuated OCV is Vaxchora ( CVD103-HgR ) , which is derived from a classical O1 Inaba V . cholerae strain and was approved by the US FDA in 2017 for use in travelers [24] . In contrast to killed OCVs , live vaccines , such as CVD103-HgR and the El Tor-derived vaccine Peru-15 , elicit more potent immune responses in young children [25 , 26] , potentially because they more closely mimic natural infection than killed OCVs . In particular , live vaccines can produce antigens in vivo that are not expressed in the in vitro growth conditions used to prepare killed vaccines; furthermore , the inactivation processes used to formulate killed vaccines can destroy antigenic epitopes [27] . In addition to the requirement for multiple doses of some OCVs for optimal protection , all current live and killed OCVs are thought to be accompanied by a post-vaccination lag in protection during induction of anti-V . cholerae adaptive immunity . The shortest reported time to protective efficacy is 8 days post-vaccination , a delay that could hamper reactive vaccination campaigns designed to limit the spread of cholera outbreaks [28] . We recently created HaitiV , a new live-attenuated OCV candidate derived from a variant El Tor O1 Ogawa V . cholerae clinical isolate from the 2010 Haiti cholera outbreak . HaitiV harbors many genetic alterations that render it avirulent and resistant to reversion while preserving its robust capacity for colonization of the small intestine [29] . In an infant rabbit model of cholera [30] , intestinal colonization with HaitiV conferred protection against lethal wild-type ( WT ) V . cholerae challenge within 24 hours of vaccination , a timescale inconsistent with the development of adaptive immunity and suggestive of a “probiotic”-like mechanism of protection . Here , using a mouse model of V . cholerae intestinal colonization , we show that oral administration of HaitiV to female mice elicits serum vibriocidal antibodies and protects their pups from lethal challenge with virulent V . cholerae . Thus , HaitiV has the potential to provide rapid probiotic-like protection as well as to elicit long-lasting immune protection from cholera .
All bacteria were grown in Luria-Bertani ( LB ) broth supplemented with the relevant chemicals at the following concentrations: streptomycin ( Sm , 200μg/mL ) , kanamycin ( Km , 200μg/mL ) , carbenicillin ( Cb , 50μg/mL ) , sulfamethoxazole/trimethoprim ( SXT , 80 and 16μg/mL ) and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside ( X-gal , 60μg/mL ) . For growth on plates , LB + 1 . 5% agar was used . All V . cholerae strains in this study , except for PIC158 and PIC018 , were spontaneous SmR derivatives of the wild-type . Bacteria were stored as -80°C stocks in LB with 35% glycerol . S1 Table lists the strains used in this study . The CT deletion strain in the H1 V . cholerae background ( HaitiWT ) was generated by allelic exchange as previously described , with an additional selection step to enhance the efficiency of obtaining a stable single crossover strain [29] . Briefly , HaitiWT was conjugated with SM10λpir E . coli bearing the suicide plasmid pCVD442-ctxAB-KmR , containing sacB as well as a kanamycin resistance cassette from pKD4 sandwiched by homology arms targeting the ctxAB operon ( locus tags N900_RS07040 –N900_RS07045 ) . Single crossovers were selected on LB+Sm/Cb/Km agar plates . To select for a double crossover , verified single crossovers were grown in LB + Cb/Km for 4 hours at 37°C and then passaged in LB+10% sucrose overnight at room temperature . Sucrose-resistant ( sacB-negative ) , KmR and CbS colonies were then conjugated with SM10λpir E . coli bearing pCVD442-ctxAB ( no KmR cassette ) and clean KmS double crossovers generated via an identical protocol . The ΔctxAB deletion was verified by colony PCR with internal and flanking primers . 4-week old germ-free ( GF ) female C57BL/6 ( Massachusetts Host-Microbiome Center ) or Swiss-Webster ( Taconic Farms ) mice were housed in a BL-2 facility for the duration of the experiment . On Day 0 , 2 , 4 , 6 , 14 , 28 , 42 and 56 , mice were anesthetized with isoflurane and orally gavaged with 109 CFU of an overnight culture of either HaitiV or CVD103-HgR in 100μL 2 . 5% Na2CO3 . Mice were weighed at every immunization and once every 4–5 days between boosts . At each weighing , fresh fecal pellets were plated on LB + Sm to enumerate shed bacteria . At Day 7 , 14 , 28 and 42 post first immunization , blood samples were obtained from each mouse by tail vein incision . A Day 1 blood sample was collected from the Swiss-Webster cohort and the single-dose C57BL/6 cohort . Blood was clotted at room temperature for 1 hour , centrifuged at 20000 x g for 5 minutes and the supernatant ( serum ) stored at -20°C for analysis . Vibriocidal antibody quantification was performed by complement-mediated cell lysis using PIC018 ( Inaba ) or PIC158 ( Ogawa ) V . cholerae as the target strain as previously described [31] . Seroconversion was defined as ≥4x increase in titer over the baseline measurement . The characterized mouse monoclonal antibody 432A . 1G8 . G1 . H12 targeting V . cholerae O1 OSP was used as a positive control for the vibriocidal assay . Titers are reported as the dilution of serum causing a 50% reduction in target optical density compared to no serum control wells . Anti-cholera toxin B subunit ( CtxB ) and anti-O-specific polysaccharide ( OSP ) responses were measured by previously described isotype-specific ELISAs [32 , 33] . Briefly , 96-well plates ( Nunc ) were coated with 1 μg/mL solution of bovine GM1 monosialoganglioside ( Sigma ) in 50mM carbonate buffer overnight . Next , 1μg/mL CtxB in 0 . 1% BSA/PBS purified from the classical Inaba strain 569B ( List Biological Laboratories ) was layered onto the GM1-coated wells . Wells were blocked with a 1% BSA/PBS mixture after which 1:50 dilutions of the mouse serum samples were loaded into each well . Goat anti-mouse IgA , IgG or IgM secondaries conjugated to HRP ( Southern Biotechnology ) were then added at a concentration of 1μg/mL in 0 . 1% BSA/0 . 05% Tween/PBS and incubated for 90 minutes . Detection was performed by adding an ABTS/H2O2 mixture to the wells and taking an absorbance measurement at 405nm with a Vmax microplate kinetic reader ( Molecular Devices Corp . , Sunnyvale , CA ) . Plates were read for 5 min at 30 s intervals , and the maximum slope for an optical density change of 0 . 2 U was reported as millioptical density units per minute ( mOD/min ) . Results were normalized using pooled control serum from mice previously immunized against V . cholerae and reported as ELISA Units as previously described [32] . Anti-OSP responses were measured and reported similarly to anti-CtxB responses , only instead of CtxB , purified OSP:BSA from either PIC018 or PIC158 ( 1 μg/mL ) was used to coat plates as previously described [34] . Additionally , OSP ELISAs were carried out with 1:25 dilutions of the serum samples . The infant mouse survival challenge was adapted from previous reports to optimize the dosage for HaitiWT and to include more frequent monitoring intervals [31 , 32] . Pregnant dams were singly housed at E18-19 for delivery . At P3 ( third day of life ) , pups were orally inoculated with 50μL LB containing 107 CFU of a directly diluted 30°C overnight culture of V . cholerae and returned to their dam . Infected pups were monitored every 4–6 hours for onset of diarrhea and reduced body temperature . Once symptoms appeared , monitoring was increased to every 30 minutes until moribundity was reached , at which point pups were removed from the nest and euthanized by isoflurane inhalation followed by decapitation for dissection and CFU plating of the small intestine on LB + Sm/X-gal . Pups that were alive at 48 hpi were deemed protected from the challenge . Cross-fostering was performed by transferring up to half of a litter between dams on the first day of life ( P1 ) . Fostering was maintained for at least 48 hours before infection to fully replace the milk from the original dam . We excluded rejected pups from analyses due to our inability to attribute mortality to infection alone . Statistical analyses were performed with Prism 8 ( Graphpad ) . To analyze whether immune responses were significantly changed over time , a one-way ANOVA was performed . Due to missing values from paired measurements as a result of insufficient serum sample volumes , antibody titers were analyzed with a mixed-effects model one-way ANOVA using the earliest sample ( Day 1 or Day 7 ) as the control and post hoc tests performed with a Dunnett’s multiple comparison test . Survival curves were analyzed with the log-rank ( Mantel-Cox ) test and CFU burdens were compared with the Mann Whitney U test . A p-value <0 . 05 was considered statistically significant . This study was performed in accordance with the NIH Guide for Use and Care of Laboratory animals and was approved by the Brigham and Women’s Hospital IACUC ( Protocol #2016N000416 ) . Infant ( P14 or younger ) mice were euthanized by isoflurane inhalation followed by decapitation . When required , adult mice were euthanized by isoflurane inhalation followed by cervical dislocation .
While the infant rabbit model enables investigation of the progression of a V . cholerae-induced diarrheal disease that closely mimics human cholera [30] , it is not appropriate to study vaccine immunogenicity because newborn animals lack a fully developed immune system . Instead , we used adult GF mice to study HaitiV immunogenicity . In contrast to normal adult mice , which are resistant to V . cholerae intestinal colonization , oral inoculation of GF mice with V . cholerae results in stable intestinal colonization without adverse effects [35–37] . In the GF model , serum markers of immunity , such as vibriocidal titers , can be measured , but challenge studies are not possible due to the persistent colonization of the vaccine strain and the resistance of adult mice to diarrheal disease . Here , we further developed a variation of the GF model [38] . Besides measuring serum markers in the orally vaccinated adult mice , neonatal pups ( which are sensitive to V . cholerae induced diarrheal disease ) born to these mice were subjected to challenge studies to evaluate vaccine protective efficacy . We established two cohorts of orally immunized adult female GF mice . In the first cohort , a small pilot study was set up to compare the immunogenicity of HaitiV and a streptomycin-resistant derivative of CVD-103HgR . This cohort consisted of 4-week-old Swiss-Webster GF mice that were immunized with either vaccine strain ( n = 3 per group ) . Cohort 2 consisted of a set of 4-week-old C57BL/6 mice that were all immunized with HaitiV ( n = 7 ) . We generally followed the multi-dose oral immunization scheme previously used in this model , which included eight doses of 1x109 CFU vaccine over eight weeks [36 , 37] . After this vaccination regimen , the mice in cohort 2 were mated and vaccine-induced protective immunity was assessed in the progeny ( Fig 1A ) . Based on fecal CFU , all animals in both cohorts were stably colonized with high levels of either vaccine strain ( Fig 1B ) . No adverse effects of long-term colonization with HaitiV or CVD103-HgR were noted , and all mice gained weight over the course of the study ( Fig 1B ) . Fecal shedding and presumably intestinal colonization of HaitiV in cohort 2 was eliminated after these dams were used to cross-foster pups born to specified-pathogen free ( SPF ) control mice ( described below ) , suggesting that a normal microbiota can outcompete HaitiV . Serum samples from the immunized mice were used to quantify antibodies targeting several V . cholerae factors thought to play roles in protection from cholera . One of these metrics , the vibriocidal antibody titer , is a validated correlate of protection in vaccinated humans [39–42] . In cohort 1 , all mice immunized with HaitiV or CVD-103HgR seroconverted within 2 weeks and developed vibriocidal titers consistent with those reported in human studies for live OCVs ( Fig 2 ) [41 , 43] . Furthermore , HaitiV and CVD-103HgR elicited comparable vibriocidal titers . In cohort 2 , HaitiV immunization of C57BL/6 mice also induced high vibriocidal titers to Ogawa and Inaba target strains ( Fig 2C ) . Isotype-specific levels of antibodies targeting Ogawa and Inaba OSP , and CtxB were also measured since they also likely contribute to immunity to cholera [39] . Although we did not measure Day 1 titers in cohort 2 , measurements from naïve GF C57BL/6 mice and baseline measurements from cohort 1 , and Day 1 of HaitiV-inoculated C57BL/6 mice in a later cohort ( S2 Fig ) showed undetectable levels of vibriocidal antibodies ( Figs 2A and S2 ) . The cohort 2 mice developed strong anti-Ogawa and anti-Inaba OSP responses ( Fig 3 , S2 Table ) . The anti-Ogawa OSP titers were generally higher than those targeting Inaba OSP , likely reflecting the fact that HaitiV is an Ogawa strain . All mice in cohort 2 also developed high levels of anti-CtxB IgA , IgG and IgM antibodies ( Fig 4 , S1 Table ) . The 100% seroconversion rate and general increase over time of all three humoral immune responses measured ( vibriocidal , anti-CtxB and anti-OSP antibodies ) reveals that orally delivered HaitiV can elicit V . cholerae-specific immune responses . To assess the protective efficacy of HaitiV in this model , we challenged the neonatal progeny of HaitiV-immunized or control dams with lethal doses of different wild type V . cholerae strains . This assay has been used to study passive immunity elicited by cholera vaccines , but has not been characterized in vaccinated GF mice [31 , 32 , 44] . Initially , we optimized this assay with litters from SPF C57BL/6 control mice . Three or four-day old pups were inoculated with 107 or 108 CFU of HaitiWT , the virulent strain from which HaitiV was derived , and returned to their dams for monitoring ( Fig 5A ) . Infected pups from both groups rapidly developed signs of dehydrating diarrheal disease , including accumulation of nest material on their anogenital regions , lethargy , skin tenting and hypothermia . All infected pups died by 48 hours post inoculation ( hpi ) , with a median time to moribundity of ~23–26 hpi ( Fig 5B ) . At the time of death , all pups were heavily colonized , with >107 CFU/small intestine ( Fig 5C ) , and had swollen ceca , another hallmark of productive cholera infection in mammalian models [30 , 45] . Since there were no significant differences in survival or bacterial loads in mice challenged with either 107 or 108 CFU , the smaller dose was used in subsequent experiments ( Fig 5B ) . Diarrhea and death in this model were entirely dependent on CT; infant mice inoculated with 107 CFU HaitiWT ΔctxAB or HaitiV were completely healthy at 48 hpi , despite sustained intestinal colonization ( Fig 5C ) . We next mated HaitiV-immunized animals from cohort 2 with age-matched GF male mice , thereby preserving their colonization with HaitiV . When challenged with HaitiWT , none of the 16 pups born to HaitiV-immunized dams developed signs of diarrhea or died by 48 hpi; in stark contrast , all pups born to non-immunized dams died within ~30 hpi ( Fig 6A , left ) . There was a marked ~5 , 000-fold reduction in the intestinal load of HaitiWT in pups born to immunized versus control dams ( Fig 6A , right ) . The pups of the immunized dams remained healthy for at least 2 weeks post-challenge , even though there were still detectable but very low levels of HaitiWT in their intestinal homogenates ( S1 Fig ) . Thus , oral immunization with HaitiV elicits an immune response that provides potent protection in nursing pups from diarrheal disease , death and V . cholerae intestinal colonization . Pups of HaitiV-immunized dams were similarly challenged with heterologous V . cholerae strains , to test the serotype and serogroup specificity of protection engendered by oral immunization with HaitiV . The additional challenge strains included an O1 Inaba strain ( N16961 ) that has been used as the challenge strain in several human volunteer cholera studies [43 , 46] and the serogroup O139 strain MO10 , which was isolated during the 1992 O139 outbreak in India . Most pups from HaitiV-immunized dams were protected from N16961 V . cholerae challenge ( 7/10 survival at 48 hpi , Fig 6B ) . Despite the clinical protection , there was a much less dramatic reduction in the intestinal burden of N16961 ( ~20-fold ) compared to that observed with HaitiWT challenge , indicating serotype-specific responses play an important role in limiting colonization . Surprisingly , pups challenged with MO10 also exhibited some protection , but there was no concomitant reduction in the intestinal burden of this O139 strain ( Fig 6C ) . Together , these observations demonstrate that animals can exhibit protection from death despite relatively robust colonization , suggesting that protection from disease may result from immunity targeting factors such as CtxB , in addition to those that impede colonization . Since our earlier studies indicated that HaitiV itself can mediate rapid protection against cholera independent of an adaptive immune response , it was important to investigate whether pups nursed by HaitiV immunized dams were colonized with the vaccine strain . Extensive plating of intestinal samples from the >50 pups used for survival assays ( limit of detection = 50 CFU/small bowel ) did not reveal any HaitiV CFU in the pups reared by HaitiV-shedding dams . Thus , vaccine strain transmission and its probiotic effects are almost certainly not the explanation for the potent protection observed in nursing pups . Cross-fostering experiments were undertaken to investigate the likely passive nature of the protection . P1 pups born to SPF dams were transferred to and reared by HaitiV-immunized dams and then challenged 2 days later with HaitiWT ( Fig 5A , between P1-P3 ) . All pups crossed-fostered by immunized dams were protected ( 100% survival at 48 hpi ) and nearly all had marked reductions ( ~1 , 000 fold ) in their intestinal HaitiWT burdens ( Fig 7A ) . These observations mirror the challenge studies presented above ( Fig 6A ) , indicating passive immunity from milk accounts for the protection that HaitiV-immunized dams bestow to their progeny . Conversely , when pups born to HaitiV-immunized dams were cross-fostered by SPF ( non-vaccinated ) dams , all succumbed to HaitiWT challenge , albeit with an increase in median survival time ( by ~6-hour ) and had high HaitiWT intestinal burdens ( Fig 7B ) . The modest extension in survival time in these mice may be due to trans-placentally derived immunity or residual milk from the HaitiV-immunized dam . Although prior work with OCVs in GF mice suggested that multiple boosts were required to maximally induce immune responses , our observation of the prolonged colonization in multiply-vaccinated animals led us to test whether a single oral dose of HaitiV could also stimulate protective immune responses [37] . A singly-vaccinated group of female GF C57BL/6 mice ( n = 4 ) was established to investigate this possibility . Like our studies with the multi-dose regimen , a single dose of HaitiV led to sustained colonization in the mice ( S2 Fig ) . HaitiV induced vibriocidal antibody titers comparable in magnitude to those from serially immunized mice ( Fig 2C ) . Litters from singly-immunized mice were also completely protected from disease resulting from HaitiWT challenge , phenocopying pups from the first cohorts ( S2 Fig ) .
Vaccines for cholera are being increasingly embraced as public health tools for prevention of endemic cholera and limiting the spread of cholera epidemics [17] . Killed OCVs are efficacious in endemic populations , but live OCVs promise to be more potent , particularly in young children [19] . Here , we showed that the live OCV candidate HaitiV induces vibriocidal antibodies and other immunological correlates of protection against cholera in GF mice and leads to protection against disease in their offspring . Protection in this model was dependent on passively acquired factors in the milk of immunized dams and not transmission or colonization of HaitiV . Although our relatively small cohort sizes precluded rigorous statistical comparisons of immune responses in the immunized mice , oral administration of even a single dose of HaitiV elicited detectable vibriocidal antibodies in all animals . These observations provide strong data establishing HaitiV’s immunogenicity . Additionally , the comparable vibriocidal titers elicited by HaitiV and CVD-103HgR , a live OCV licensed by the FDA for travelers , bodes well for HaitiV’s immunogenicity in humans . Combining the immunogenicity data presented here with our finding that HaitiV can protect from cholera prior to the induction of adaptive immunity [29] , suggests that HaitiV may function as a dual-acting agent , providing both rapid-onset short-term protection from disease while eliciting stable and long-lasting immunity against cholera . Data from the challenge experiments ( Fig 6 ) are consistent with the prevailing notion that serogroup , and to a lesser extent serotype are major determinants of protection against V . cholerae challenge [39 , 47] . Although it is thought that exposure to Inaba strains is more cross-protective than exposure to Ogawa strains , the relative potency of Inaba versus Ogawa vaccines in eliciting dual protection against both O1 serotypes requires further definition , as it has been suggested that both Ogawa and Inaba vaccine strains are good candidates for development [4–7] . A mixture of Ogawa and Inaba serotypes either as distinct strains or one bivalent strain ( serotype Hikojima ) may be beneficial in broadening the breadth of the immune response to HaitiV [48 , 49] . The modest protection that HaitiV immunization provided against V . cholerae O139 was unexpected . The epidemiology of the original O139 outbreak and experimental studies in rabbits demonstrate a lack of cross-protection between the two serogroups [14 , 15 , 39] . Notably , although pups born to HaitiV-immunized dams and challenged with MO10 survived longer than pups born to non-immunized dams , there was little difference in the MO10 intestinal colonization between these groups ( Fig 6C ) . The discrepancy between clinical protection and relatively robust colonization suggests that HaitiV stimulates immune responses to V . cholerae factors , like CT , that may contribute to disease but not directly to colonization . The capacity of live OCVs to induce immune responses to in vivo-expressed antigens , including CtxB , is a property that heightens the appeal of live vs killed OCVs [27 , 50] . Although GF mice enabled us to test the protective efficacy of a candidate live OCV , their absence of the microbiota and resulting improper immune development are important caveats to consider . The GF model does not recapitulate the competitive microbial environment that live OCVs will encounter in the human host . We observed similar prolonged shedding patterns for both CVD-103HgR and HaitiV in the GF mice ( Fig 1 ) , yet CVD-103HgR is known to be shed by human volunteers at a low frequency for a short period [25 , 51] . Thus , our findings likely overestimate HaitiV’s capacity to colonize the human intestine . The observation that exposure of HaitiV-immunized dams to SPF-derived pups during the cross-fostering experiments led to the elimination of detectable HaitiV in feces supports the prediction that this vaccine will not stably colonize humans . Since we employed a multi-dose vaccination schedule here , it remains an open question whether transient exposure of naïve mice to HaitiV will also stimulate protective immunity , as has been shown in the context of vaccination with V . cholerae outer membrane vesicles [52 , 53] . The streptomycin-treated mouse model of V . cholerae colonization , which allows for temporary intestinal colonization , may also be useful to investigate the duration of colonization required for immunity [54] . Ultimately , the capacity of HaitiV to colonize the intestine and the relationship between colonization and protective immunity will need to be defined in human volunteers . The immunogenicity of live OCVs in mice has only been investigated in GF animals because adult mice with intact microbiota are refractory to intestinal V . cholerae replication and colonization . However , previous studies of live OCVs in GF mice only analyzed immune correlates of protection and not protection against challenge [36 , 37] . The combination of the neonatal survival assay with the oral GF vaccination model builds on existing knowledge of these mice to assay both the immunogenicity and protective efficacy of live OCV candidates [38] . This model may be a useful addition to existing approaches that probe the molecular bases of vaccine-mediated mucosal protection against pathogens , a topic with significant translational potential that remains poorly understood [53 , 55] . A recent report employing a similar maternal-infant transmission model in the context of intraperitoneally-delivered heat-killed Citrobacter rodentium highlights the versatility of assessing vaccine protective efficacy using the infant progeny of immunized animals as readouts [56] . The broad availability of genetically engineered mice and the relative ease of GF-derivation provides a powerful opportunity to leverage both host and bacterial genetics to explore how live-OCVs can be optimized to better defend against this ancient pathogen . | Oral cholera vaccines are increasingly used as public health tools for prevention of cholera and curtailing the spread of outbreaks . However , current killed vaccines provide minimal protection in young children , who are especially susceptible to this diarrheal disease , and require ~7–14 days between vaccination and development of protective immunity . We recently created HaitiV , a live-attenuated oral cholera vaccine candidate derived from a clinical isolate from the Haiti cholera outbreak . Unexpectedly , HaitiV protected against cholera-like illness in infant rabbits within 24 hours of administration , before the onset of adaptive immunity . However , HaitiV’s capacity to stimulate adaptive immune responses against the cholera pathogen were not investigated . Here , we report that HaitiV induces immunological correlates of protection against cholera in adult germ-free mice and leads to protection against disease in their offspring . Protection against disease was transferable through the milk of the immunized mice and was not due to transmission or colonization of HaitiV in this model . Coupling the immunogenicity data presented here with our previous observation that HaitiV can protect from cholera prior to the induction of adaptive immunity , we propose that HaitiV may provide both rapid-onset short-term protection from disease while eliciting stable and long-lasting immunity against cholera . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"vibrio",
"immunology",
"tropical",
"diseases",
"microbiology",
"vaccines",
"bacterial",
"diseases",
"vibrio",
"cholerae",
"neglected",
"tropical",
... | 2019 | Oral immunization with a probiotic cholera vaccine induces broad protective immunity against Vibrio cholerae colonization and disease in mice |
B-1 cells play a critical role in early protection during influenza infections by producing natural IgM antibodies . However , the underlying mechanisms involved in regulating this process are largely unknown . Here we found that during influenza infection pleural cavity B-1a cells rapidly infiltrated lungs , where they underwent plasmacytic differentiation with enhanced IgM production . This process was promoted by IL-17A signaling via induction of Blimp-1 expression and NF-κB activation in B-1a cells . Deficiency of IL-17A led to severely impaired B-1a-derived antibody production in the respiratory tract , resulting in a deficiency in viral clearance . Transfer of B-1a-derived natural antibodies rescued Il17a-/- mice from otherwise lethal infections . Together , we identify a critical function of IL-17A in promoting the plasmacytic differentiation of B-1a cells . Our findings provide new insights into the mechanisms underlying the regulation of pulmonary B-1a cell response against influenza infection .
Highly localized infection in respiratory tract is a defining feature of influenza infection . An appropriate induction of both innate and adaptive immune responses at this site is necessary for virus elimination and host recovery . B cell response is triggered primarily in respiratory tract , which is essential for antiviral immune response against influenza infections by opsonization of pathogens , activation of complement receptor-mediated phagocytosis and promotion of other immune defenses [1–4] . Influenza virus-binding antibodies are provided by two sources , B-1 cells and conventional B-2 cells . Due to the low frequency of antigen-specific B-2 cells at the onset of infection and a general requirement for simultaneous T cell help , early induction of natural antibody response by B-1 cells is critical . B-1 cell-associated antigen receptors are biased with respect to BCR repertoire and preferentially recognize conserved epitopes present on common pathogens [3 , 5] . In addition , B-1 cells are known to secrete most natural antibodies spontaneously at very low level and do so in the apparent absence of antigen challenge [5 , 6] . Thus , this class of lymphocytes provides efficient immune surveillance by producing natural neutralizing IgM antibodies before isotype class-switched , high affinity-maturated IgG can be produced by B-2 cells [7–11] . B-1 cells are enriched in the pleural and peritoneal cavities where pulmonary or intestinal infections usually occur [5] , which consist of two functionally specialized subpopulations , CD5+ B-1a and CD5- B-1b cells [12–14] . B-1a cells produce most of the natural antibodies and can also participate in innate responses upon antigen stimulation [15–18] . Previous studies have shown the localized accumulation of B-1a cells in mediastinal lymph nodes ( MedLN ) during influenza infection [3] . However , it remains to be investigated whether B-1a cell response occurs in the lung and if so , what molecular mechanisms regulate this process during infection . IL-17A has been identified as a pro-inflammatory cytokine and participates in chronic inflammation and autoimmune diseases via its effects on a broad range of target immune cells [19–22] . Either deficiency or blockade of IL-17A signaling diminishes antibody responses [23–26] . Early studies have shown that IL-17A-mediated signaling is critical for early control of pulmonary bacterial infections [27] . We previously reported that IL-17A deficient ( Il17a-/- ) mice exhibited impaired viral clearance and more severe immunopathological changes after H5N1 influenza virus infection when compared with wild type ( WT ) controls [28] . However , it has remained unclear whether IL-17A deficiency affects B-1 cell response during influenza infection . Here , we show for the first time that airway exposure to influenza causes migration of pleural B-1a cells to lungs for further differentiation into plasma cells with enhanced production of protective IgM antibodies , a process critically regulated by IL-17A-mediated NF-κB activation and Blimp-1 induction . Our findings provide new insights into natural antibody response by B-1a cells and its regulatory mechanisms .
In H1N1 influenza virus-infected mice , significantly up-regulated IL-17A expression was detected in lung tissue as early as 2 days post-infection ( dpi ) ( S1A Fig ) . Among immune cell subsets present in the lung tissue , IL-17A-producing γδT cells were detected by flow cytometric analysis in naïve mice ( S1B and S1C Fig ) , and the size of this IL-17A+ γδT population significantly increased at 2 dpi and peaked by 5dpi ( S1D and S1E Fig ) . In contrast , very few IL-17A+ CD4+ T cells were detected in naïve lungs , and the size of this cell population did not increase until 5dpi ( S1C and S1E Fig ) . Therefore , IL-17A+ γδT cells represent the major source of IL-17A produced at early stage of pulmonary influenza infection . To assess a protective function of IL-17A against influenza infection in vivo , we found that H1N1 virus-infected Il17a-/- mice exhibited significantly reduced survival rate compared with WT controls ( Fig 1A ) . In addition to the reduction in body weight , a much higher viral burden was detected in H1N1-infected Il17a-/- mice ( Fig 1B and 1C ) , suggesting a deficiency in viral clearance in Il17a-/- mice . Further histological analysis revealed substantially increased severity of lung damage in Il17a-/- mice , characterized by pronounced inflammatory destruction and leukocyte infiltration ( Fig 1D and 1E ) . In WT mice , intra-nasal administration of H1N1 virus induced a rapid anti-virus IgM response in bronchoalveolar lavage fluid ( BLF ) whereas local IgG levels did not increase significantly until 7 dpi ( Fig 1F ) . In contrast with WT mice , this early IgM response was severely impaired in Il17a-/- mice ( Fig 1G ) . Notably , we also detected markedly reduced phosphorylcholine ( PC ) -specific IgM production in BLF of H1N1-infected Il17a-/- mice ( Fig 1G ) . Since previous studies found that PC-specific natural antibodies were exclusively produced by B-1a cells in naïve mice [29–31] , these data indicate that IL-17A deficiency might lead to impaired B-1a cell response during influenza infection . To determine whether B-1a cells infiltrate in the lung tissue in response to influenza infection , we examined the kinetic changes of pulmonary B-1a cells in H1N1 virus-infected WT mice by flow cytometry ( Fig 2A ) . As early as 2 dpi , a marked increase of CD19+IgM+CD43+CD5+ B-1a cells was found in the lung tissue and peaked at 5 dpi ( Fig 2A and 2B ) . Accumulation of infiltrated B-1a cells was also observed in the lung tissue of Il17a-/- mice ( S2A and S2B Fig ) . Their presence in lung tissue was further confirmed with histological examination ( Fig 2C ) . To determine the cellular source of infection-induced IgM , we enumerated the frequencies of IgM-producing cells in lung tissue by ELISPOT analysis . Remarkably , we detected a significant amount of infiltrated virus-specific IgM-producing B-1a cells , which was 15-fold more than virus-specific IgM-producing B-2 cells in the lung tissues by ELISPOT analysis ( Fig 2D and 2E ) . These results suggest that IgM-producing B-1a cells in lung tissue serve as the predominant source of influenza virus-binding IgM at early stages of infection . After establishing a key role of B-1a cells in early virus-specific antibody production in lung tissue and observing significantly impaired B-1a cell response in Il17a-/- mice , we next determined whether timely induced natural antibodies by B-1a cells are critical for the survival of Il17a-/- mice . Previous studies showed that B-2 and B-1b cells , but not B-1a can be efficiently generated when bone marrow is transplanted to adult mice [18 , 32] . In this study , we constructed irradiation chimeras with B-1a cell depletion , which was achieved by reconstituting the full body-irradiated mice with bone marrow cells only , whereas transfer of bone marrow cells together with pleural cavity B-1 cells allowed full regeneration of B-2 , B-1a and B-1b cell populations in irradiated mice ( Fig 2F and S3A and S3B Fig ) . Upon reconstitution , serum levels of virus-specific and PC-specific IgM were very low in mice receiving bone marrow cells alone ( S3C Fig ) . Transferring serum from the chimera mice with fully reconstituted B-1a cells into H1N1 virus-infected Il17a-/- mice was performed at 1 dpi . These Il17a-/- mice exhibited much higher survival rates when compared to Il17a-/- mice that received serum from the chimera mice without B-1a cells or Il17a-/- mice with no serum transfer ( Fig 2G ) . These data demonstrate that impaired B-1a cell responses largely account for the reduced survival of H1N1-infected Il17a-/- mice . To rule out the possibility that Il17a-/- mice have intrinsic defect in natural antibody production from B-1a cells , we next examined the mice at naïve state . Detailed analysis of un-challenged WT and Il17a-/- mice found no significant differences in total numbers of B-1 cell subsets in either plural or peritoneal cavities ( S4A and S4B Fig ) . Moreover , similar levels of total IgM , virus-specific and PC-specific IgM were detected in BLF and serum of naïve mice ( S4C and S4D Fig ) . Therefore , these results collectively suggest that IL-17A is essential for early induction of natural antibody from B-1a cells during H1N1 infection , but not for their normal development or function at the naïve state . Next , we performed cytospin preparations on sorting-purified B-1a cells from lung tissue or pleural cavity . B-1a cells in lung tissue were found to be morphologically distinguishable from B-1a cells in the pleural cavity and had a more differentiated plasma cell appearance ( Fig 3A ) , indicating that B-1a cells underwent plasmacytic differentiation after migration into lung tissue . Although B-1a cells from lung tissue of infected Il17a-/- mice revealed an apparent plasmacytic morphology , they were morphologically distinguished with B-1a cells from lung of infected WT mice by exhibiting reduced cytoplasm-to-nucleus ratio ( Fig 3A ) . Since CD138 expression has been associated with plasmacytic differentiation of B-1a cells [33] , we examined the levels of CD138 expression on B-1a cells by flow cytometry . As shown in Fig 3B , 3C and 3D138 expression was similar on pleural B-1a cells between WT and Il17a-/- mice from naive and infected mice , but markedly upregulated on B-1a cells in lung tissues from infected mice . Further analysis reveals a much higher level of CD138 on B-1a cells from lung of WT mice than that of Il17a-/- mice . To examine whether IL-17A affects B-1a differentiation and antibody production in vivo , we evaluated the antibody producing capacity of B-1a cells from lung tissue and pleural cavities of naïve or virus-infected WT and Il17a-/- mice . Spontaneous but negligible IgM production was detected in pleural B-1a cells ( Fig 3C ) , resulting in large numbers of pinhead-size ELISPOTs , without significant antibody secretion detected in the supernatant ( Fig 3D and 3E ) . Also , no differences in frequencies of IgM producing cells or IgM made per cell were observed between WT and Il17a-/- mice either before or after infection ( Fig 3D and 3F ) . However , markedly increased IgM was detected in culture supernatant of sorting-purified B-1a cells from lung tissues ( Fig 3D–3F ) . In accordance with this result , much larger and heavier antibody-forming spots were observed by ELISPOT analysis ( Fig 3C ) . The increased amounts of IgM secreted per B-1a cell from lung tissue was detected based upon the concentrations of IgM in culture supernatants and correlated spot frequencies detected by ELISPOT ( Fig 3F ) . Thus , these data suggest that pulmonary B-1a cells become high-rate immunoglobulin-producing plasma cells after migration into lungs of infected mice . We then compared the antibody production by B-1a cells from WT and Il17a-/- mice . B-1a cells from lung tissue of Il17a-/- mice showed significantly decreased frequencies of IgM-producing cells and reduced levels of antibodies in culture supernatants ( Fig 3C–3E ) . When IgM levels were correlated with spot frequencies detected by ELISPOT , the amount of IgM secreted per B-1a cell from Il17a-/- mice was only 2/3 of that from WT controls ( Fig 3F ) . Together , these data suggest that IL-17A deficiency impairs B-1a plasmacytic differentiation during influenza infections . Flow cytometric analysis revealed that IL-17 receptors A ( IL-17RA ) and C ( IL-17RC ) were expressed at high levels on B-1a cells from pleural cavities of WT mice ( Fig 4A and 4B ) . To examine whether IL-17A directly affects B-1a cells , we found that B-1a cells markedly increased their antibody production when treated with IL-17A in culture ( Fig 4C ) . ELISPOT analysis confirmed the increased numbers of antibody-producing B-1a cells after IL-17A treatment ( Fig 4D and 4E ) . We also detected up-regulated levels of aid , irf-4 and xbp-1 transcripts in B-1a cells upon IL-17A treatment ( Fig 4F and S1 Table ) . Moreover , up-regulation of Blimp-1 , IRF4 , and XBP-1 at both mRNA and protein levels was detected in IL-17A-treated B-1a cells ( Fig 4F and 4G and S5 Fig ) . Notably , IL-17A enhanced the processing of NF-κB1 precursor p-105 and increased the nuclear translocation of p-65 in B-1a cells ( Fig 4H ) . Together , these data demonstrate a direct function for IL-17A in promoting B-1a cell differentiation and antibody production . As the existence of multiple binding sites for NF-κB was predicted in the promoter of prdm-1 gene that encodes the transcriptional factor Blimp-1 ( Fig 5A and S1 Table ) , we performed the chromatin immunoprecipitation ( CHIP ) assay to determine whether IL-17A signaling could elicit this response . Indeed , NF-κB bound to multiple sites in the prdm-1 gene promoter following IL-17A treatment . Moreover , amplification with primers for predicted sites 4 , 8 , 9 , 10 , 12 in the prdm-1 promoter showed increased levels of products ( Fig 5B ) . Furthermore , we observed increased nuclear translocation of NF-κB/p65 upon IL-17A treatment by confocal microscopy ( Fig 5C ) .
The ability of B-1 cells to produce natural IgM antibodies is an important part of the innate immune system . Many studies have characterized B-1 cells as first-line effectors of host defenses prior to the development of adaptive humoral and cellular immune responses [2 , 3 , 10 , 34] . Current investigations have mainly focused on the development and homeostasis of B-1 cells [13 , 14] , but much remains to be determined about the regulatory mechanisms underlying B-1 response against infections . In this study , we have found that the B-1a subset preferentially and rapidly immigrates into the lungs of H1N1-infected mice . Recent studies have shown that IL-17A plays a crucial role in promoting germinal center formation and antibody production by B-2 cells [23 , 24 , 35] , but a function of IL-17A in regulating B-1 cell responses has not been established . Here , we demonstrate that B-1a cells express functional surface receptors for IL-17A while IL-17A promotes B-1a cell differentiation via NF-κB activation and Blimp-1 induction . Moreover , IL-17A drives the differentiation of pulmonary B-1a cells into high-rate IgM producing plasma cells in H1N1-infected mice . Of particular importance , B-1a cell-derived natural antibodies can rescue Il17a-/- mice from otherwise lethal infections , indicating a critical role of IL-17A in regulating B-1a response against H1N1 infections . Tissue specific micro-environmental factors may favor the plasmacytic differentiation of B-1a cells and need to be identified . Possibly relevant , B cell-activating factor ( BAFF ) , a TNF family cytokine produced by macrophages and dendritic cells , regulates the survival of peritoneal B-1 cells [36] . Organs such as lung and gut were previously thought to be non-immune but now appear to actively shape immune functions of a broad range of immune cells [37 , 38] . There is emerging evidence indicating lung as a potential site for lymphocyte education during the onset of diseases [37 , 38] . In the current study , up-regulated IL-17A expression was detected in lung tissues of influenza-infected mice as early as 2 dpi . Our detailed analysis has demonstrated a critical role of IL-17A in supporting plasmacytic differentiation of B-1a cells both in vivo and in vitro . Early studies found that B-1 cells constitutively secrete small amounts of IgM and may be maintained in a “semi-activated” or “pre-plasma” cell state [39] . IgM antibody-production of B-1a cells is closely related with their egression from peritoneal and pleural cavities to other lymphoid organs [16 , 40] , where they can differentiate into plasma cells [41–43] . A similar transition may occur following influenza infection . Existing evidence indicates that B-1a cells can actively accumulate in the lung-draining lymph nodes following influenza infection [3] . Here we have further demonstrated the functional significance of local differentiation of B-1a cells in the lung tissue and its regulating signal . We show that lung-infiltrated B-1a cells were more plasma cell-like with respect to morphology and transcription profiles as compared to ones present in the pleural cavity . Moreover , the plasmacytic differentiation of B-1a cells contributes to increased natural antibody level that is critical for animal survival . Our current understanding of transcriptional regulation for plasmacytic differentiation comes mainly from the investigation of B-2 cells . Previous studies have identified a network of transcriptional factors that regulate plasmacytic differentiation . One principle molecule closely associated with this process of B-2 cells is Blimp-1 , the master regulator of plasmacytic differentiation [44 , 45] . Blimp-1 orchestrates a gene expression program that drives B cells to become plasma cells through the repression of genes involved in the B-2 cell proliferation , antigen presentation , germinal center reactions , and B-T cell-cell interaction [45] . Ectopic expression of Blimp-1 is sufficient to drive B-2 cells to differentiate into antibody-secreting cells [46 , 47] . Although several studies demonstrated that B-1 cells constitutively express low but detectable levels of Blimp-1 in steady state [48] , and antibody production of B-1a cells requires Blimp-1 [48 , 49] , the regulatory mechanisms underlying plasmacytic differentiation of B-1a cells , particularly the involvement of Blimp-1 during this process , remain to be elucidated . We have observed that up-regulated Blimp-1 expression at both mRNA and protein levels is closely associated with IL-17A-induced differentiation of B-1a cells both in vivo and in vitro . Thus , it is possible that plasmacytic differentiation of B-1a cells requires similar regulatory mechanisms involving Blimp-1 as compared with B-2 cells . Nuclear factor-κB ( NF-κB ) was first described as a transcription factor in B cells that binds to the enhancer element controlling immunoglobulin kappa light chain expression [50] . Highly activated and constitutive levels of NF-κB were reported in B cells [51] whereas its decreased expression led to cell death or growth arrest [51–53] . Considering that the steady state tonic signaling in B-1 cells in the absence of specific stimulation represents a major difference from B-2 cells [54–57] , it is reasonable to speculate that the threshold levels of NF-κB activation in B-1 cells maintain the natural antibody production . In support of this hypothesis , we detected increased nuclear translocation of NF-κB p65 in B-1a cells upon IL-17A stimulation by Western blot analysis . Moreover , multiple binding sites of NF-κB on the promoter of prdm-1 gene were confirmed by CHIP analysis , consistent with recent findings that NF-κB binding to the promoter of Prdm-1 directly induces Blimp-1 transcription and expression during plasmacytic differentiation of B cells [58 , 59] . Together , our results reveal a novel function of IL-17A in activating the NF-κB-Blimp1 axis for B-1a cell differentiation . The adaptive immunity requires the cognate interaction between T and B cells and clonal expansions to generate antigen specific response and memory . Despite their relatively low frequency in the secondary lymphoid tissues , the properties of B-1 cells that secrete antibodies with repertoire that is enriched for highly poly-specific to microbial antigens provide a unique advantage for their pivotal role in first-line protection [8 , 9 , 49] . One striking benefit of innate B-1a response is its rapid and effective response to control the initial infection [2 , 3 , 34 , 60] . The proximity of pleural cavity to the lung provides pleural B-1a cells the advantage to respond quickly to pulmonary infections . Based on the in vivo and in vitro analyses , we have shown that influenza infection triggers a series of rapid events in the lung where B-1a cells become IgM secreting plasma cells under the influence of IL-17A . Of particular importance , the IL-17A-mediated B1-a response is closely correlated with animal survival from H1N1 infection , which may suggest a potential therapeutic target for the treatment of influenza infections .
Female Il17a-/- mice on C57BL/6 background and C57BL/6 WT control mice between 6–8 weeks of age were used . Il17a-/- mice were obtained from Dr . Yoichiro Iwakura [61] at the Institute of Medical Science , The University of Tokyo , Japan . And C57BL/6 mice were purchased from the Jackson Laboratory ( Bar Harbor , ME , USA ) . All the mice were housed in specific pathogen-free laboratory animal unit of the University of Hong Kong , and were given free access to food and water . For H1N1 influenza virus challenge experiments , mice were housed in biosafety level-2 individual ventilation cages ( IVCs ) and given free access to food and water . Experiments were followed with the standard operating procedures in a biosafety level-2 laboratory and were approved by the Institutional Animal Ethics Committee , The University of Hong Kong . The 50% lethal dose ( LD50 ) of A/PR/8/34 was determined in C57BL/6 mice after serial dilution of the viral stock from embryonated hens’ eggs , and LD50 does of A/PR/8/34 were adopted in viral challenge experiments . After anesthetized with isoflurane , mice were intranasally ( i . n . ) challenged with 30μl virus diluted in PBS . Weight loss , signs of illness and survival were monitored for 14 successive days . Mice were sacrificed at the indicated time points for examination . All animal experiments were approved by the Committee on the Use of Live Animals in Teaching and Research ( CULATR ) at the University of Hong Kong ( CULATR project number: 2735–12 and 3681–15 ) , following the Code of Practice for Care and Use of Animals for Experimental Purposes established by the Animal Welfare Advisory Group , Agriculture , Fisheries and Conservation Department , and approved by the Government of the Hong Kong Special Administrative Region . Influenza type A virus , H1N1 strain A/Puerto Rico/8/1934 , was propagated in the allantoic cavity of 10-day-old embryonated hens’ eggs at 37°C with 65% humidity for 48 hours as previously described [28 , 62] . Allantoic fluid was collected and stored in aliquots . To prepare inactivated virus , the allantoic fluid was concentrated and purified in a 10–50% sucrose gradient by centrifugation at 25000g , 4°C for 2 hours . 0 . 25% formalin ( v/v ) was used to inactivate the purified virus at 4°C for 7 days . Further purification was performed with Amicon Ultra Membrane ( 4208 ) ( Millipore , Billerica , MA , USA ) , with a molecular weight cutoff at 30 kDa . The products were re-suspended in phosphate-buffered saline ( PBS ) . Inactivation of the virus was confirmed by the absence of cytopathic effects and detectable hemagglutination ( HA ) in the supernatant of two consecutive 50% tissue culture infectious dose ( TCID50 ) assays by the method of Reed and Muench [62 , 63] . Female C57BL/6 mice between 6–8 weeks of age were used to generate B-1a eliminated mice . Briefly , mice were full-body irradiated with 956 cGy of Caesium . To construct mice without B-1a cells , eight hours after irradiation , 3x106 bone marrow cells from C57BL/6 mice were injected intravenously ( i . v . ) via the tail vein into irradiated mice . Control mice were generated by transferring both 3x106 bone marrow cells and 5x106 peritoneal cavity cells from C57BL/6 . Mice with B-1a cell depletion was analyzed 2 months after cell transfer . Infected Il17a-/- mice were i . v . injected 0 . 5ml of serum from naïve WT mice , irradiated WT mice reconstituted with BM and cavity cells , or irradiated WT mice reconstituted with only BM cells at 1 , 3 , 5 dpi , respectively . Mice were monitored for the survival rate for 14 successive days . Cells were incubated at 4°C with Fc-blocking reagent ( Biolegend ) before the addition of the appropriate fluorochrome-labeled mAbs . For multicolor flow cytometric analysis , cell samples were stained with the following monoclonal antibodies specific for following phenotypic markers: anti-B220-FITC ( clone RA3-6B2 ) , anti-B220-PE ( clone RA3-6B2 ) , anti-CD5-PE7 ( clone 53–7 . 3 ) , anti-CD43-PE ( S11 ) , anti-Gr1-FITC ( clone RB6-8C5 ) , anti-CD19-PerCp-Cy5 . 5 ( 6D5 ) , anti-IgM-APC ( clone RMM-1 ) , anti-CD11b-PE ( clone M1/70 ) , anti-IL-17RA-PE and the isotype-matched control antibody from Biolegend ( San Diego , CA , USA ) or BD Biosciences Pharmingen ( San Diego , CA , USA ) ; anti-IL-17RC-APC and the isotype-matched control antibody from R&D ( USA ) . Fluorescent stained cells were then analyzed with FACS Aria I flow cytometer ( BD Biosciences ) and analyzed with FlowJo software ( Tristar ) . Mice were sacrificed at the indicated time points post-infection , and tissues were inflated with 10% neutral buffered formalin for at least 24 hours before processing and embedding . Lung tissue was sectioned at 6-μm thickness and stained with hematoxylin and eosin for histopathological evaluation . Slides were examined in a blinded manner and scored with a semi-quantitative system as previously described [28] according to the relative degree of inflammation and tissue damage [64–66] . The cumulative scores of inflammatory infiltration , degeneration and necrosis provided the total score per animal . Lung infiltration of inflammatory cells was scored as follows: 0 , no inflammation; 1 , mild peribronchial and peribronchiolar infiltrates , extending throughout <10% of the lung; 2 , moderate inflammation covering 10–50% of the lung; 3 , severe inflammation involving over one-half of the lung . Degeneration was scored as follows: 0 , no degeneration; 1 , little vacuolar degeneration of bronchi and bronchiole epithelium cells , normal pulmonary alveoli; 2 , mild necrosis of bronchi and bronchiolar epithelium , mild alveoli damage; 3 , severe degeneration . Necrosis was scored as follows: 0 , no necrosis; 1 , mild necrosis with scant exudate; 2 , marked necrosis with abundant exudate; 3 , severe interstitial edema around blood vessels , apparent injured parenchyma and degenerated alveolar epithelial cells with greater infiltration of inflammatory cells . For immuno-fluorescent examination , tissues were embedded in OCT , and snap-frozen in liquid nitrogen . Cryo-sections ( 6 μm ) were stained with monoclonal antibodies specific for phenotypic markers and examined with confocal microscope Carl Zeiss LSM 710 . Slides were examined in a blinded manner . BLF was prepared by instilling 0 . 5 ml of sterile-filtered PBS through the trachea into the lung airways and aspirated with a syringe . Lavage fluid was centrifuged at 1 , 500 rpm for 5 minutes and collected supernatant was stored at -80°C for further examination . Total protein levels in BLF were determined by Bradford protein assay ( BIO-RAD , Hercules CA ) . Antibody concentrations in BLF were examined with ELISA assay . Lung tissues or PBS washed cells were homogenized in Trizol ( Invitrogen , Life technologies ) , following procedures as previous described [28] . Briefly , total RNA samples were prepared with an RNeasy Kit ( Qiagen , Hilden , Germany ) and reverse transcribed with SuperScript III First-Strand Synthesis SuperMix ( Invitrogen , Carlsbad , CA , USA ) . Real-time PCR was performed using Platinum SYBR Green qPCR SuperMix-UDG with ROX ( Invitrogen ) according to the manufacturer’s instructions with an Applied Biosystems Prism 7900HT real-time PCR system ( Foster City , CA , USA ) . Real-time PCR reactions were set up under the following conditions: 95°C for 2 min , 40 cycles of 95°C for 15 s and 60°C for 30 s . The threshold cycle ( CT ) of gene products was determined and set to the log-linear range of the amplification curve and kept constant . Relative expression of genes was calculated as 2ΔΔCT with normalization to the corresponding internal genes . To determine the copy numbers of the H1N1 NP viral RNA in infected lungs , Total 2ug RNA was reverse transcribed , and 4ul of reverse transcribed products was subjected to real-time PCR analysis . The serial diluted pHW2000 plasmid constrcted with viral NP gene ( Genebank: KT314335 . 1 ) of the H1N1 influenza A virus ( A/Puerto Rico/8/1934 ) was used as quantitative standard . Single cell suspensions of mouse splenocytes , lymph node or lung were obtained from fresh tissue samples . Mouse B-1a or B-2 cells were sorting-purified with a BD FACSAriaI cell sorter . Total lung cells were isolated as described [67] . Briefly , mice were sacrificed and perfused with PBS via injection into the right ventricle , which flushed blood vessels in the lungs . Lung tissue was harvested for digestion with type II collagenase and DNase I ( Merck , Whitehouse Station , NJ , USA ) for 1 hour at 37°C . After red blood cell lysis with ACK buffer , the cell number was enumerated . Frequencies of various immune cell populations were examined by immuno-staining and flow cytometric analysis . Samples including BLF , serum or supernatants from cell culture were collected for measuring the production of IgM using a colorimetric sandwich ELISA . Standard capture antibody , phosphorylcholine ( PC ) -BSA ( Biosearch Technologies ) or purified influenza virus , biotinylated detection antibody and horseradish peroxidase ( HRP ) -conjugated streptavidin were used . The reactions were developed after the reaction of 3 , 3' , 5 , 5'-Tetramethylbenzidine ( TMB ) ELISA substrate solution ( Fisher Scientific ) and read in a Microplate Absorbance Reader ( TECAN , Austria ) at OD450nm absorption . Total IgM was quantitatively determined with serial diluted IgM standard ( Biolegend ) . Arbitrary units of A/Puerto Rico/8/1934-specific or PC-specific IgM titers , defined as U/ml , were determined by comparision to hyper-immune sera from WT mice . Binding of that serum at 1x104 dilution was set as 1U . Coating was performed in a 96-well filtration plate ( cat . No . MAHAS4510 ) with 100μl of 5μg/ml goat anti-mouse IgM , purified inactivated H1N1 influenza virus A/PR/8/34 , or PC-BSA in coating buffer ( or PBS ) , and incubated at 4°C overnight . Plates were washed and then blocked with RPMI 1640 with 10% fetal bovine serum ( R10 ) at room temperature for 1 hour . Purified cells ( 0 . 2–0 . 02x106 ) were seeded into wells and incubated overnight at 37°C . Plates were washed thoroughly before the addition of goat anti-mouse IgM-AP ( 1:1000 ) diluted in 1% BSA-PBS overnight at 4°C . After washing , plates were developed by adding BCIP/NBT solution ( Sigma-Aldrich ) . Spot formation was monitored visually and stopped immediately by gently washing the plate . Single cell suspensions were washed and diluted in 100 μl of RPMI-1640 medium with 10% fetal bovine serum ( FCS ) . Cytospin preparation was performed at 500 rpm for 2 minutes in a Shandon CytoSpin III Cytocentrifuge ( Thermo Scientific , USA ) . The slides were fixed in cold acetone at room temperature for 5 to 10 minutes before Wright’s staining . To perform the Wright’s staining of cytospin-prepared cells , sufficient quantity of Wright Stain Solution ( Electron Microscopy Sciences , USA ) was placed upon the smear for 3 minutes at room temperature . After washing the stained smear , the film was allowed to dry in the air and mounted with mounting medium . B-1a cells purified from C57BL/6 mice were stimulated with 20ng/ ml of IL-17A . At 1 , 2 , 4 , 6 , 8 and 24 hours post-stimulation , B-1a cells were collected for CHIP assay based on the manufacturer’s instruction ( CHIP assay kit , Beyotime , China ) . Briefly , Cells were cross-linked with 1% formaldehyde , and lysed with SDS lysis buffer . Equal amount of proteins were immunoprecipitated with anti-p65 or anti-normal mouse IgG overnight . The immuno-complex was captured by protein A/G agarose beads for 2 hours , washed and eluted with elution buffer . After reverse cross-linking of protein/DNA complexes in 0 . 2M NaCl at 65°C for 5 hours , DNA was purified according to the manufacturer’s instruction ( DNA Purification Kit , Beyotime , China ) . Real-time PCR was conducted to detect the putative NF-κB binding sequences in the promoter of prdm-1 gene . Data in this study were indicated as mean with standard error . Statistical comparisons were calculated by the Student’s t-test . To construct the survival curve of H1N1 influenza virus-infected mice , the Kaplan-Meier analysis method was adopted in the analysis . A value of P<0 . 05 was considered statistically significant for all the data . The UniProt ( http://www . uniprot . org/ ) accession numbers for genes and proteins discussed in this paper are: mouse Blimp-1 , Q60636; mouse AID , Q9WVE0; mouse IRF4 , Q64287; mouse XBP-1 , O35426; mouse IL-17A , Q62386; mouse HPRT , P00493 . | Influenza infection is highly localized in respiratory tract where immune response is triggered to provide protection from primary infection . Although natural IgM antibodies produced by B-1a cells have long been recognized as first-line protection against influenza , it remains unclear whether B-1a cell response occurs in the lung and what molecular mechanisms regulate this process . We show that airway exposure to influenza causes migration of B-1a cells to lungs for further differentiation into plasma cells with enhanced production of protective IgM antibodies . IL-17A critically regulates this process by driving differentiation of B-1a cells to high-rate IgM producing plasma cells in situ . Thus , IL-17A is a key factor in the local inflammatory milieu that modulates early humoral immunity afforded by B-1a cells . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2016 | IL-17A Promotes Pulmonary B-1a Cell Differentiation via Induction of Blimp-1 Expression during Influenza Virus Infection |
Kinesin is a family of molecular motors that move unidirectionally along microtubules ( MT ) using ATP hydrolysis free energy . In the family , the conventional two-headed kinesin was experimentally characterized to move unidirectionally through “walking” in a hand-over-hand fashion by coordinated motions of the two heads . Interestingly a single-headed kinesin , a truncated KIF1A , still can generate a biased Brownian movement along MT , as observed by in vitro single molecule experiments . Thus , KIF1A must use a different mechanism from the conventional kinesin to achieve the unidirectional motions . Based on the energy landscape view of proteins , for the first time , we conducted a set of molecular simulations of the truncated KIF1A movements over an ATP hydrolysis cycle and found a mechanism exhibiting and enhancing stochastic forward-biased movements in a similar way to those in experiments . First , simulating stand-alone KIF1A , we did not find any biased movements , while we found that KIF1A with a large friction cargo-analog attached to the C-terminus can generate clearly biased Brownian movements upon an ATP hydrolysis cycle . The linked cargo-analog enhanced the detachment of the KIF1A from MT . Once detached , diffusion of the KIF1A head was restricted around the large cargo which was located in front of the head at the time of detachment , thus generating a forward bias of the diffusion . The cargo plays the role of a diffusional anchor , or cane , in KIF1A “walking . ”
Time dependent structural information is of central importance to understand detailed mechanisms of biomolecular systems . In particular , biomolecular machines dynamically transit many structurally and chemically distinct states making cycles in state space , by which they fulfill their functions . Unfortunately , no single experimental technique provides sufficient spatio-temporal resolution for them . X-ray crystallography and others provide structural information at high resolution , but this is primarily static . Biochemical and single molecular experiments tell us kinetic and dynamic behaviors , but their spatial resolution is limited . To fill the gap among them , molecular dynamics ( MD ) simulations have been playing important roles . Yet , due to their size and long time scale involved , atomistic MD cannot cover an entire cycle of molecular machines at the moment [1] . To overcome this limitation , recently , we initiated to use structure-based coarse grained MD ( CGMD ) methods [2] , [3] to mimic the cycle of machines for the case of F1-ATPase and others [4] , [5] . Notably , most of these machines contain more than one ATPase domains and their coordinated dynamics are crucial to understand the mechanisms [6] , [7] , [8] , [9] . This is an interesting issue , but at the same time , makes the cycle unavoidably complicated . Thus , for the simplicity and clarity , it is good to study those that contain only one ATPase domain and that have much of crystallographic information . In this sense , a single-headed kinesin , KIF1A , is an ideal target system , for which here we performed CGMD simulations mimicking an entire ATP hydrolysis cycle . Kinesin is a family of molecular motors that move unidirectionally along microtubule ( MT ) using ATP hydrolysis free energy [10] . In the family , the conventional kinesin , kinesin-1 , was experimentally characterized to move toward the plus ends of MT processively with discrete 8-nm steps per one ATP hydrolysis reaction , where the coupling between ATP hydrolysis reactions and 8-nm steps is rather tight [11] , . The conventional kinesin is a two-headed motor and has been shown to “walk” in a hand-over-hand fashion by coordinated motions of the two heads [6] , [9] , [17] , [18] . In this sense , it is a surprise that even though KIF1A , a member of kinesin family , is a single-head motor , it still can move processively and directionally along MT , as observed by single molecule experiments [19] , [20] , [21] . In particular , the mechano-chemical coupling of KIF1A is loose: KIF1A can move back and forth stochastically with an average biased towards the forward direction , with step sizes in multiples of 8-nm . This is in contrast to conventional kinesin that seldom shows backward steps without a large load and that shows a uniform step size of 8-nm per one ATP hydrolysis [19] , [20] , [21] . Thus , KIF1A must use a different mechanism from the conventional kinesin to achieve the overall unidirectional motions . How KIF1A , with only one head , can generate the unidirectional movements driven by ATP-hydrolysis reaction is unclear in terms of structural dynamics , which we address in this paper by structure-based CGMD . Various molecular simulations have been applied to kinesin as well as other molecular motors [4] , [8] , [22] , [23] , [24] . For the molecular simulations of KIF1A movements , structural information on nucleotide-dependent conformational change is indispensible . X-ray crystallography provides KIF1A structures in two major conformations; ATP and ADP bound forms [25] , [26] , [27] . The two forms share the overall fold of the head domain with some changes . One crucial change is in the helix α4; its orientation relative to the rest of the head is rotated by about 20 degrees between the two forms ( Fig . 1A , blue for ATP-form and red for ADP-form ) . Another major change is in the so-called neck-linker region , which is the C-terminus of the head domain: the neck-linker is ordered and tightly docked to the core of the head in the ATP form ( magenta in Fig . 1B ) , while it is disordered and thus invisible in the ADP form . This neck-linker docking/undocking has been implicated as a source of the power-stroke in the kinesin family [28] , [29] . The K-loop ( L12-loop ) and L11-loop , which flank the α4 helix , also show some changes between the two forms ( Fig . 1A ) . Cryo-electron microscopy ( cryo-EM ) of the KIF1A-MT complex together with X-ray structures of building blocks led to structural models for the KIF1A-MT complex in the two major forms of KIF1A ( the ATP and ADP forms ) [30] , [31] . The modeled complexes show that , in both of the forms , the key interaction sites of KIF1A with MT is the α4 helix , which fits to a groove located between α-tubulin and β-tubulin . In the two forms of KIF1A , the orientation of the α4 helix relative to MT is mostly unchanged , which leads to the 20-degrees rotation of the core domain relative to the long axis ( z-direction in this article ) of MT depending on the bound nucleotide states ( Fig . 1C ) : In the ATP-form , the core adopts the “upright” docking ( blue in Fig . 1C left ) , while in the ADP form the core is rotated about 20 degrees and adopts the “tilted” docking ( red in Fig . 1C right ) [31] . This core rotation has been suggested to be important for KIF1A movement [26] , [31] . The KIF1A-MT complex models provide a clue for the processivity of KIF1A . Regardless of the nucleotide states , the positively-charged K-loop of KIF1A is close to the negatively charged E-hooks , disordered C-terminus regions of α/β-tubluin [26] , [31] . Thus , the long-ranged electrostatic attractions between K-loop and E-hooks are assumed to prohibit the KIF1A from completely leaving from MT . This idea is supported by a mutational experiment , in which charge reduction of K-loop decreased the processivity [19] . The ATP hydrolysis cycle and its correlation with KIF1A head motion have been investigated previously [19] , [20] , [21] , [26] ( see Fig . 2A ) . The ATP form of the KIF1A head binds strongly to MT ( T-phase in Fig . 2A ) , whereas the direct contact of the ADP-form of KIF1A to MT is weak . Thus , after the ATP hydrolysis and Pi release , the KIF1A head can detach from MT ( still loosely bound to MT via electrostatic interactions of the K-loop and E-hooks ) . The detached head starts diffusion along MT under the constraints generated by the interaction of the K-loop and E-hooks . After the long one-dimensional diffusion along MT , KIF1A can finally find a binding site located at the groove between α- and β-tubulins ( D-phase in Fig . 2A ) . The contact between tubulin and KIF1A induces ADP dissociation from KIF1A leading to the nucleotide free state . In this state , KIF1A binds MT tightly ( Φ-phase in Fig . 2A ) . At the final stage , ATP binding induces neck-linker docking and the rotation of the core ( T-phase in Fig . 2A ) . The above knowledge , however , does not tell us the mechanism of how KIF1A can generate directional movement towards the plus end of MT . To address the mechanism of directional movement , we designed and conducted a series of molecular simulations employing structure-based CG protein models . The structure-based CG protein models have proven to be useful to study mechanical aspects of kinesin [8] , and other molecular motors [4] , . Based on the energy landscape view of proteins [32] , [33] and structural data for the two forms of KIF1A , we set up single and/or two-basin energy landscapes of KIF1A for every phase of an ATP cycle [34] , [35] . Then , the ATP hydrolysis cycle was mimicked by dynamically switching the energy functions of KIF1A in different phases of the cycle ( Fig . 2B ) [4] , [36] . While the full-length KIF1A has a rather long C-terminal tail , we here concentrate on a truncated KIF1A ( C351 ) that was used in in vitro single-molecule assays [19] , [20] , [21] . We note that , by employing the structure-based CG simulations , our purpose here is not to conduct a single simulation that most accurately approximate the real molecular system , as some parameters in the CG simulations are not accurately derived from atomic interactions . Instead , taking advantage of the speed of the structure-based CG simulations , we systematically conduct a series of simulations for a broad range of these parameter values . These comparative computer experiments are useful for a mechanistic understanding .
We designed a simulation system for one ATP hydrolysis cycle of KIF1A that induces KIF1A motions along MT . The simulation system contains 7 protein subunits: a KIF1A molecule that moves dynamically and three copies of tubulin αβ dimers that were fixed in the form of a segment of single protofilament of MT ( Fig . 1D ) . All the proteins were modeled at a one-bead-per-residue resolution ( each amino acid was represented by a bead located at the Cα position ) . For KIF1A , we employed structure-based CG models that concisely represent the energy landscape , which is a globally funnel-like shape where the bottom of the funnel can have more than one basin [33] . We focused on a truncated KIF1A C351 ( unless otherwise mentioned ) since the motility of this type KIF1A is intensively investigated in the single-molecular assay of Okada et al [19] , [20] , [21] . Conformational changes of KIF1A upon chemical reactions were simulated by the multiple-basin model [35] , while the long-time dynamics that do not involve chemical reactions , such as diffusion process , were simulated by a single-basin perfect-funnel ( i . e . , Go ) model [34] , [37] ( see below and Materials and Methods for more details ) . Protein dynamics was simulated by stochastic differential equation , i . e . , the Langevin equation ( see below and Materials and Methods for more details ) . The crystal structures of ATP-bound KIF1A ( designated as KIF1A ( T ) hereafter ) and ADP-bound KIF1A ( KIF1A ( D ) ) are available from the Protein Data Bank and were used in the CG models as reference structures of the corresponding states . For the KIF1A-MT complex structures , the cryo-EM-based models for the ATP- and ADP-bound KIF1A-MT complexes are also available and were used ( we designate XT and XD respectively ) . These models explain the high and low affinities in ATP-bound and ADP-bound forms of KIF1A , respectively , by the number of direct contacts . The structure for nucleotide-free KIF1A ( KIF1A ( Φ ) ) is currently unavailable; we assumed that the neck linker is disordered based on experiments , and that the KIF1A ( Φ ) -MT complex structure XΦ except the neck linker to be the same as that of XT because both states have a similarly high affinity to MT . Using these complexes , we modeled the interactions between KIF1A and MT as a Go-like pair potential ( unless otherwise mentioned ) . In the current CG model , the interaction strength between KIF1A and MT is a key parameter . First , the interaction strength parameter had to be tuned so that KIF1A ( T ) can stably bind to MT while KIF1A ( D ) can detach from MT during the affordable simulation time . This tuning was easy because , as mentioned above , the modeled complex structures of KIF1A-MT have more residue-contacts in the ATP form than in the ADP-form . A more delicate tuning was necessary for the affinity of KIF1A ( D ) to MT because KIF1A ( D ) is expected to detach from MT and later reattach . Obviously , a too weak interaction does not lead to attachment of KIF1A to MT , whereas a too strong interaction does not allow the detachment from MT . Via many preliminary runs , we found a certain range of the interaction strength parameters that satisfy these conditions ( described in the next subsection ) . Our simulation started from the XT . KIF1A was bound to the central tubulin αβ dimer ( Fig . 1D ) . We simulated the KIF1A ( T ) state for 5×105 τ , where τ is the unit of time in CG-simulation , using the multiple-basin potential with two basins: a stable basin at XT and a meta-stable basin at XD structures ( Fig . 2B top ) . The unit of time τ can be mapped to ∼0 . 128 ps in real time scale based on the diffusion constant of the KIF1A head ( see Materials and Methods for the detail information ) . Then , we induced the conformational change to the ADP-bound form by switching the potential so that the XD structure becomes more stable than XT ( see the second row and left cartoon of Fig . 2B ) . With this setting , we simulated the system for 4×106 τ , which is long enough to complete the conformational change to ADP-form . For many samples , KIF1A ( D ) detached from MT during this period . We note that , throughout the simulations , a constraint potential was applied that represents long-range loose interactions between the K-loop and E-hooks , by which KIF1A cannot move far away from MT ( see Materials and Methods for details ) . Then , we conducted a long simulation ( 2×108τ ) with the single-basin Go potential for the XD ( the second row and central cartoon in Fig . 2B ) . The switch from the multiple-basin potential to the single-basin Go potential saves computer time and is done solely for technical reasons . During this period , many trajectories showed KIF1A re-attachment to MT . Once KIF1A attached to MT , we continued the run for another ∼1×107τ and then moved to the next stage . The next stage is a preparation to the subsequent conformational change to the nucleotide-free ( Φ ) state and uses the multiple-basin model with the stable basin at XD and the meta-stable basin at XΦ for 5×105τ . After that , corresponding to the release of ADP , we induced the conformational change to the nucleotide free form by switching the potential so that the XΦ structure is more stable than XD ( the third row right in Fig . 2B ) . Subsequently , for a long time dynamics , we used the single-basin potential for the Φ state for 1×107τ . Finally , ATP-binding is mimicked by switching the potential to the single potential for XT . We simulated the T state for ∼2×108τ , which completes the XTXDXΦXT cycle . We now analyze KIF1A movement during one ATP cycle . As in Fig . 2A , it is expected that KIF1A detaches from MT and attaches to MT both in the D-phase . Thus , modeling of the interaction between KIF1A ( D ) and tubulin is very delicate . Since the CG modeling is unavoidably less accurate , instead of deciding one “correct” interaction strength , we scanned the strength over a certain range . In a strong interaction case ( designated as [stand-alone/strong] , εgoKIF1A-MT = 0 . 225 ) ( Throughout the paper , the energy unit corresponds to kcal/mol ( ∼1 . 7 kBT = ∼6 . 95 pN . nm ) although the mapping is rather approximate ) , we saw KIF1A cannot detach from MT for 99 of 100 trajectories ( Fig . 3A ) within the simulated time . Whereas , with a weak interaction ( [stand-alone/weak] , εgoKIF1A-MT = 0 . 153 ) that was carefully tuned after trial-and-errors , we found that KIF1A can detach from MT and attach to MT ( the first three cases in Fig . 3B ) for 186 of 235 samples ( ∼80% ) . The rest 49 samples did not show detachment ( bottom in Fig . 3B ) . The first , second , and third cases in Fig . 3B illustrate the one forward step ( +8 nm ) , the zero-step ( 0 nm ) , and the one backward step ( −8 nm ) within one ATP hydrolysis cycle , respectively ( For an example of stand-alone KIF1A movements for [stand-alone/weak] , see Supporting Information Video S1 ) . Of the 186 cases that KIF1A detached from and attached to MT within one ATP chemical cycle ( TDΦT ) , the positions of KIF1A at the end of simulations were +8 nm ( the forward step ) for 44 cases , 0 nm ( zero-step ) for 92 cases , and −8 nm ( the backward step ) for 50 cases ( For statistics , Table 1 ) . We note that the system contained only 3 pairs of tubulin αβ's that correspond to kinesin binding sites of +8 nm , 0 nm , and −8 nm so that possibilities of two steps were out of the scope here . On average , no significantly biased move was observed . Apparently , this does not explain the in vitro single molecule experiments that found forward biased moves . The simulations above did not consider electrostatic interactions at all , which may have affected the results . Indeed , recent work by Grant et al reported forward bias of two-headed kinesin landing due to electrostatic interactions [22] . We thus added the electrostatic interactions between KIF1A and MT by the Debye-Huckel formula and repeated the same set of simulations for 80 samples for the case of [stand-alone/weak/DH] . We set the salt concentration of 50 mM , and put +1 charges to all the Lys , Arg , and His residues and −1 charges to all the Asp and Glu residues in the simulated system . Of 80 , 6 samples did not show detachment , 12 samples showed one forward-step ( 8 nm ) , 16 samples showed one backward step , and 46 samples returned to the original site ( see Fig . S1 ) . Thus , inclusion of the simple electrostatic interactions did not produce forward-biased movements although it changed the trajectories to some extents ( see Figs S2 , S3 , S4 , S5 , S6 ) . We further tried simulations with many different sets of parameters never finding biased motions . Our results is apparently inconsistent with the biased binding mechanism proposed in [21] . There can be two possibilities . 1 ) Some fine effect which is not included in our CG simulations , such as more accurate electrostatic treatment by Grant et al , is responsible for the forward biased binding . 2 ) The forward-biased binding is not realized . Further work is necessary to solve the issue . In struggling for search of models/situations that exhibit the forward biased move of KIF1A , we came up with a situation that a large cargo-analog is attached to the C-terminus of the neck-linker of KIF1A . The cargo-analog is modeled as a large sphere of ∼1 µm radius , and thus has very small diffusion constant . There are some in vitro experiments for myosin , as well as another kinesin mutant , that suggest the importance of diffusion anchor linked at the end of motor proteins for processive and directional movements [38] , [39] . Technically , we added a mass point with large friction coefficient to the C-terminus of the neck linker . With a large cargo-analog , we first used a strong interaction between KIF1A and MT ( [cargo/strong] , εgoKIF1A-MT = 0 . 225 , the same strength as the case of [stand-alone/strong] ) , and simulated one ATP cycle for 109 samples . We modeled the cargo as a sphere of radius 3000 times as large as the radius of an amino acid , which is ∼1 µm . Assuming the same density as amino acids , the mass of the cargo scales as 30003 times as large as that of an amino acid . The Stokes-Einstein law D = kBT/6πηr , where η is water viscosity: ∼0 . 8 m [Pa s] and r is the radius of the particle , gives that the diffusion constants Dcargo for the cargo is 3000 times smaller than the diffusion constant of an amino acid . ( See Materials and Methods for the detailed information ) . In the simulations , we found most samples either moved one-step forward ( 52 of 109 cases , an example in the upper panel of Fig . 4A top and Video S2 ) or re-bound to the original site ( 56 of 109 cases , the upper panel of Fig . 4A bottom ) , while almost no case of the backward step was found ( Table 1 ) . In ATP-bound state ( t<5105τ ) , KIF1A head kept binding to MT firmly and the cargo-analog did not move significantly at 4 nm in front of the head corresponding to the length of the neck-linker ( a snapshot in Fig . 4B top , a histogram in Fig . 5 left ) . Immediately after the ATP hydrolysis , KIF1A head detached from MT quickly . After the detachment , KIF1A head exhibited quasi-one dimensional diffusion along MT , while , due to the large friction , the cargo-analog did not move significantly . Thus , the fluctuation of the KIF1A head was restricted around the almost-fixed cargo located 4 nm in front ( Fig . 4B and Fig . 5 left ) . The cargo-analog played a role of an anchor ( or a cane ) . After some diffusion , the detached head finally re-bound to MT . Because of the limited range of diffusion , the re-binding site was either the forward site ( +8 nm ) or the original site ( 0 nm ) . After the attachment on MT , we changed the state of the system from ADP-state to Φ-state , which did not lead to any marked difference in the movement of the cargo or the head . After that , ATP binding to KIF1A induced the neck-linker docking that moved the position of the cargo-analog , which is about 8 nm in case of the forward step ( Fig . 5 left ) . Thus , after one ATP cycle ( TDΦT ) , the 8-nm or 0-nm displacements of the cargo-analog as well as the head were realized stochastically . With a weak interaction between KIF1A and MT ( [cargo/weak] , εgoKIF1A-MT = 0 . 153 ) , we still found clear forward bias ( the bottom panel of Fig . 4A ) although the details were different . In particular , due to a weaker interaction , the average time for the head diffusion increased , which resulted in larger exploration by one-dimensional diffusion and appearance of the one backward step ( 26 of 150 samples ) as well as the one forward ( 43 of 150 ) step , and the zero step ( 81 of 150 ) ( Fig . 5 middle ) . As noted before , our simulation system included only the three binding sites and thus two forward or backward steps were not realized by design . For comparison , Fig . 5 right shows the histogram of the move for the case of [stand-alone/weak] , confirming that no significant bias is observed . We now focus on the detachment process of the KIF1A head from MT after the ATP hydrolysis and Pi release . Upon the potential switch from ATP- to ADP-state at t = 5105 τ ( Fig . 2B top to the second row left ) , the decrease in the number of residue-contacts between KIF1A head and MT led to the reduced binding energy , which could induce the detachment of KIF1A head . Interestingly , with the strong interaction , the stand-alone KIF1A simulation showed the KIF1A head detachment with the probability 1% , whilst the simulation with the cargo-analog promptly induced the head detachment with the probability 100% ( Fig . 6A ) . Thus , clearly , the cargo-analog enhanced the KIF1A head detachment from MT . Even with the weak interaction between KIF1A and MT , the detachment probability was 79% for the stand-alone KIF1A ( Fig . 6A ) . With the strong interaction , we tested the detachment process with three cargo sizes ( and thus three frictions and masses ) ( the inset in Fig . 6A ) ; the small ( light-green ) , the middle-size ( red , the default one ) and the large ( purple ) cargoes correspond to the radii of 2000 , 3000 , and 4000 times of one amino acid , respectively . Technically , for given radii , masses and frictions were scaled according to Stokes-Einstein law in the same way as described . We see that KIF1A did not detach from MT with the probability 11% for the case of the small cargo , while the detachments probabilities were 100% for the system with the middle or the large cargo . Thus , the relatively large friction/mass cargo promoted the detachment of the KIF1A head . In the complex of KIF1A-MT , the α4 helix of KIF1A fits into a groove of MT . When the ATP hydrolysis occurs in the KIF1A head bound to MT , the head tends to make conformational change from ATP-form to ADP-form . With the constraint on the α4 helix , the conformational change would induce about 20 degree clockwise rotation of the head relative to the microtubule ( viewed from the top as shown in Fig . 1C left to right ) , which increases the distance between C-terminal of the head and the cargo rapidly . Then , a tag-of-war between the head and the cargo takes place . When the cargo is sufficiently large , the cargo is less mobile and wins the tag-of-war , thus finally pulling the KIF1A head out of MT . Fig . 6B illustrates time series of the binding energy for the strong interaction case . For the case of [stand-alone/strong] ( orange ) , the binding energy was weakened from ∼−30 kcal/mol in T-state to ∼-20 kcal/mol in D-state , but the latter was strong enough to hold the KIF1A head stably . For the large cargo case ( purple ) , upon ATP hydrolysis , KIF1A promptly detached from MT . For the cases of small ( light-green ) and the middle-size ( red ) cargoes , TD switch immediately weakened the binding energy to ∼−12 . 5 kcal/mol , which were followed either by detachment or by the relaxing to the binding energy ∼−20 kcal/mol in D-state ( light-green ) . This transient intermediate state with the binding energy ∼−12 . 5 kcal/mol corresponds to the frustration imposed by the immobile cargo . Similar behavior was seen in the case of the weak interaction ( Fig . 6C ) . We found it interesting to plot the trajectories in the plane ( zrelative , EB ) [zrelative: the relative position of the cargo ( zcargo-zhead ) , EB: the binding-energy] both for the cases with and without the cargo-analog ( Fig . 6D ) . Trajectories start from the right-lower area in ( EB , zrelative ) plane . With the large cargo ( red and blue ) , after the relaxation of the binding energy from the initial condition to 0 kBT ( the detachment ) , the cargo-analog moved . Whereas , without the cargo-analog , the C-terminus fluctuation occurred first and then KIF1A head may or may not detach from MT ( orange and dark-green ) . The difference comes from the different time scale for the mobility of the cargo-analog . Experimentally , the binding free energy of KIF1A head with MT was estimated from the dissociation constant as ∼−20 kBT in the ADP bound state [19] . In the current simulations , the binding energies in the D-phase are −35 kBT for the strong interaction case ( see Fig . 6B ) and −17 kBT for the weak interaction case ( Fig . 6C ) . Note that the experimental estimate is the free energy about the standard state , while the estimates from simulations are merely interaction energies . Thus these numbers should not be quantitatively compared . With the uncertainty in mind , perhaps , the real binding strength may fall in between the strong and the weak interaction cases . Next , we analyze the diffusion and the attachment processes of KIF1A head after the detachment in ADP-state ( Fig . 7 ) . The attachment rate for [cargo/strong] is larger than that for [cargo/weak] , as expected . Interestingly , the attachment rate for [cargo/weak] was much smaller than that for [stand-alone/weak] , probably due to the restricted motions anchored by the large cargo-analog . Thus , the large cargo-analog enhanced the detachment , but retarded the attachment . Fig . 7B shows a transient histogram for the z-coordinates of the KIF1A head and of the cargo-analog soon after the detachment from MT . With the cargo-analog ( Fig . 7B left and middle ) , its positions were nearly fixed , whereas the head fluctuated broadly ( ∼4 nm in both directions ) , which coincides with the length of neck linker . We note that , since we measure the diffusion after the detachment from MT , the histograms for [cargo/strong] and for [cargo/weak] are nearly the same: The diffusion process itself ( up to the attachment ) was not affected by the interaction strength . As the diffusion time increases , the cargo-analog slowly moves , which enables the head to reach the backward site , as well as the forward site . For the stand-alone case ( Fig . 7B right ) , C-terminus position diffused quickly after the detachment from MT , and the distributions of the C-terminus and the head were nearly symmetric about the starting point ( 0 nm ) . In the simulations , the average times τattachment for attachment of the KIF1A head to MT for the system cargo/strong and cargo/weak were ∼0 . 2×108 τ ( ∼2 . 5 µs ) and ∼0 . 5×108 τ ( ∼6 . 4 µs ) , respectively . A rough estimate of the diffusion length in this time scale is ∼1 . 2–1 . 9 nm , which is small . After ATP binding , the neck-linker docked to the head core . The neck-linker docking moves the cargo-analog by about +8 nm when the head landed to the forward site ( Fig . 8 ) . The docking rate depends on the size of the cargo-analog , as expected . Only in the cases of the weak interaction , after the attachment of the head onto MT , occasionally the head re-detached from and then re-attached to MT ( Fig . 9 ) . This extra processes , being not coupled with ATP cycle , did not produce significant bias in the KIF1A move . In the above simulations , the cargo was always placed at z∼4 . 25 nm based on the ATP-form reference structure , which may raise a concern that this specific initial positioning may affect the stepping statics . To address this concern , we performed the same type of one-ATP cycle simulations with various initial cargo positions z; z = 4 . 75 , 4 . 25 , 3 . 75 , 3 . 25 , 2 . 75 , 2 . 25 , 1 . 75 , and 1 . 25 nm especially for the cargo/strong case . This range corresponds to the range of cargo found at the end of original simulations ( see Fig . S7 A the upper panel which shows the distribution of the probability density for the relative position of the cargo at the end of original simulation ) . From each of these cargo ( initial ) positions , 10 simulations were conducted . From the initial position of the cargo: z>3 nm , we found clear forward-biased moves , whereas z<3 nm , the head seldom detached from MT ( the lower panel of Fig . S7A ) . Overall , by distributing the initial cargo positions , the forward bias is somewhat reduced on average . Importantly , however , we still clearly see , on average , forward-biased moves of KIF1A with the cargo . In this paper , we primarily focused on the specified molecular construct ( C351 ) used in in-vitro motility assay experiments [19] , [20] , [21] , in which the length of neck-linker except for His-tag is 22-residues . Here , to test robustness of our results , we investigated the stepping statics of another construct that has 5-residue longer neck-linker . The 5-residue segment is modeled as a flexible chain ( by Modeller ) . In a similar way to the above sub-section , we estimated the range of the cargo position ( Fig . S7B upper panel which shows the distribution of the probability density for the position of the cargo ) , and repeated simulations ( 10 runs each ) with the initial cargo position at z = 6 . 25 , 5 . 75 , 5 . 25 , 4 . 75 , 4 . 25 , 3 . 75 , 3 . 25 , 2 . 75 , 2 . 25 , 1 . 75 , and 1 . 25 nm . From the initial position of the cargo: z>4 . 5 nm , we found clear forward-biased moves , whereas z<4 . 5 nm , the head seldom detached from MT ( Fig . S7B lower panel ) . Thus , although the bias is weakened , we still see clear forward-biased moves of KIF1A with the cargo linked by 5-residue longer linker .
Conventional kinesin is dimeric and “walks” in a hand-over-hand fashion , akin to human walking by two legs . Extending the analogy to human walking , the current simulations suggest that the large cargo-analog can play the role of a cane for the walk of single-headed kinesin; with a cane , we can walk even with one leg . Although in this work , we focused on a truncated KIF1A which is used in the single-molecular assay of Okada et al [19] , [20] , [21] , we should note that the cellular function of KIF1A in vivo is markedly more complicated than the situation we considered here . Several experiments [40] , [41] showed that KIF1A may be dimerized by virtue of being bound to a single cargo-analog in some case . Our model does not straightforwardly apply to the dimeric KIF1A system in vivo . Next , we discuss in vitro experiments of related systems . First of all , forward-biased movements were observed for single-head kinesins , both KIF1A and a single-head construct of conventional kinesin mutant , with latex beads linked to C-terminus , where the size of beads are sub-µm to µm [13] , [21] . Thus the current simulations are perfectly consistent with these results . In a study of myosin VI , single-molecule experiments reported very similar phenomenon to our simulations [38] . A single-head construct of myosin VI did not show directional movements without beads/cargo . When a bead was attached to an end of the myosin head , it exhibited directional movements . It was argued that the bead played the role of diffusion anchor . Some other experiments in vitro are subtle . While a truncated single-headed kinesin ( K351 ) did not show marked processive movements , it exhibited processive and directional movements when fused with BDTC ( 1 . 3-S subunit of propionibacterium shermanii transcarboxylase ) [39] . Here , BDTC is much smaller than beads/vesicles and thus does not apparently correspond to the current simulations . Yet , the linked BDTC increased K351-BDTC affinity to MT which implies that BDTC attractively interacts with tubulin . Thus , the interaction of BDTC with MT may provide additional friction to the C-terminus of K351 , which is qualitatively the same as the role of the cargo-analog in our simulations . Next , we discuss the dynamics of KIF1A head and cargo-analog . In Video S2 and Fig . 4 , the cargo looks almost fixed at the beginning . First , we note that , although we used the cargo-analog of ∼1 µm-size , Video . S2 and the snapshot in Fig . 4B drew much smaller ball to “visualize” the cargo position . Thus a small ball is purely for graphics . Taking into account the ∼1 µm-sized cargo together with the Stokes-Einstein law and Maxwell-Boltzmann distribution , we see that the cargo does not move much: Based on the Stokes-Einstein law DKIF1A = 1 . 2×108 [nm2/s] , and Dcargo = 2 . 9×105 [nm2/s] , respectively , we estimate that , while the KIF1A head diffuses for ∼8 nm distance , the cargo diffuses only ∼0 . 4 nm which is rather small . Thus , it is physically reasonable that the cargo looks immobile in Video . S2 , and Fig . 4 . Furthermore , in Video S2 and Fig . 4A , the motions of KIF1A head look like Brownian dynamics with little effect of inertia , whereas the motions of the cargo-analog look under-damped oscillation . This difference can be understood by estimating the lifetime of the corresponding velocity correlations . We estimated that the lifetime of velocity correlation for the KIF1A head is ∼10 τ , and that for the cargo-analog is ∼108 τ . A characteristic time scale of the detached KIF1A head to find the adjacent binding site ( z = L or -L ) from the dissociation was ∼106 τ ( see Material and Methods ) . Therefore , within this time scale , the motions of KIF1A should be diffusive , while the motion of the cargo-analog is damped oscillation . Related to these arguments , we also note time scale for KIF1A head diffusion . As in Result , our estimate in simulations was τattachment∼2 . 5 µs–6 . 4 µs . Experimentally , the mean duration time of weak-binding state was estimated as τw = 7 . 5 ms [19] . However , τw obtained via an indirect estimate seems to include not only the diffusional searching time , but also the ADP release time . Since ADP release is a slow process , we do not know much on the diffusional search time . We next discuss the effect of neck-linker length on the enhanced forward-biased motions . After the neck-linker docking , the average positions of cargo for C351 and a 5-residue longer variant C351+5 were similar , while the distribution of the positions was broader for the C351+5 case than that for the C351 case . ( Fig . S7 A upper panel ) . Based on Fig . S7 , we can estimate that the average steps per one ATP cycle are about 2 . 7 nm for C351 and about 2 . 0 nm for C351+5 . Thus , a longer neck linker contributes to broadening of the distribution of the cargo position at the time of ATP hydrolysis , which results is gradual decrease in the forward biased movement of KIF1A . The full-length KIF1A has much longer neck linker . In the recent simulation study for the dimeric kinesin with a long tail domain and cargo [42] , the neck-linker docking itself did not bring the cargo to the forward position significantly . On top , the full length KIF1A tends to dimerize . So , our analysis focuses on the truncated KIF1A construct such as C351 and argument for the full-length system needs further analysis . Importantly , the current simulations showed that the linked cargo-analog is sufficient to induce the biased Brownian movement of KIF1A , but whether the linked diffusion anchor is necessary or not was not investigated . In vitro experiments , KIF1A exhibited forward-biased Brownian movements even when only a chromophore was attached [19] , [20] , [21] . Since the chromophore is much smaller than the cargo/bead , this does not correspond to the current simulations . Note that His-tags attached to C-terminus of KIF1A in vitro constructs may also contribute to additional interactions with MT since tubulin contains negatively charged C-terminus tails ( E-hooks ) on the surface . The result that some constructs with too short neck-linker did not exhibit directional movements suggests importance of a certain length of neck-linker between the head and the His-tag . Yet , we do not exclude the possibility of other mechanisms that induce the directional movements . In particular , recent computational work reported that electrostatic interactions between two-head kinesin heads and MT can provide modest bias to the forward direction [22] . Thus , the linked cargo-analog can be to enhance the forward bias . In the end , we briefly mention about possible experiments which can test the current simulation results . The direct test of the proposed mechanism is to perform a motility assay with a bead or dye attached to the core of the KIF1A head which is far from N- or C- termini and which is located on the surface opposite to the MT binding orientation . Even better way is to introduce two imaging probes; a bead in the C-terminus and a dye to the core of KIF1A head . We expect to see movements in asymmetric hand-over-hand fashion . Another easier but more indirect test is to investigate ion-strength dependence of the stepping statics for C351 . If the interaction between KIF1A and MT is weakened by a higher ion-strength , the re-binding time of KIF1A becomes longer , which results in weakening of the forward-bias .
To simulate an ATP cycle by the structure-based CG model , we need reference KIF1A-MT complex structures Xν for every states ν ( = T , D , Φ ) in the cycle ( T , D , and Φ correspond to ATP , ADP and nucleotide free state , respectively ) . The crystal structures of ATP-bound KIF1A ( KIF1A ( T ) ) and ADP-bound KIF1A ( KIF1A ( D ) ) are available from the Protein Data Bank and were used in the CG models as references . For the KIF1A-MT complex structures , we used the model structures 2HXF ( the pdb id ) for the KIF1A ( T ) /tubulin αβ complex XT , and 2HXH for the KIF1A ( D ) /tubulin αβ complex XD [30] , [31] ( see Fig . 1A , 1B , and 1C ) . Since motility-assay experiments heavily used a chimera protein C351 where the catalytic core of KIF1A was fused to the neck linker of conventional kinesin ( KIF5C ) [19] , [20] , [21] , we employed the same chimera C351 ( KIF1A-KIF5C ) , except for N-terminal T7-tag and C-terminal His-tag and Cys . Missing residues in the loop of KIF1A , including a part of the neck-linker region , were modeled by Modeller [43]: We constructed 200-samples and chose the model with the best Modeller score . These modeled loops were treated as flexible regions with reduced force constants ( see Coarse-grained model in Supporting Information Text S1 ) . For the nucleotide free state , no structure is available . Since KIF1A ( Φ ) constitutes the strong-binding state in a similar way to KIF1A ( T ) , we decided to use the same structure as the XT excluding the neck linker , which is known to be disordered in Φ-state [28] , [44] , [45] , [46] . We treated the neck liner in Φ-state as flexible regions . Both 2HXF and 2HXH contain missing residues in tubulin αβ at the so-called E-hook region , which were not explicitly modeled . Instead , we included the effect of E-hook in a simpler way ( see Coarse-grained model in Supporting Information Text S1 ) . The simulation system here contains a KIF1A molecule and three copies of tubulin αβ dimers ( Fig . 1D ) where all the tubulin molecules were fixed throughout simulations . The initial structure of simulations contained XT structure of KIF1A attached to the central tubulin αβ dimer at the form of 2HXF . The coordinates were set so that the MT protofilament is along the z-axis with the plus end being positive z , and KIF1A-binding surface of tubulin is roughly perpendicular to y-axis ( see Fig . 1D ) . The origin was defined by the position of Cα-atom of Phe94 of KIF1A at the initial structure ( roughly the center of mass of KIF1A ) . The period of the MT is about 8 nm along z-axis . We applied the structure-based CG models for the KIF1A-MT system [34] , [35] . KIF1A and three tubulin αβ dimers were represented by a set of beads , where each bead placed at the position of Cα atom represents one amino acid . ( See Supporting Information Text S1 for detail ) . To mimic an ATP hydrolysis cycle ( Fig . 2A ) , we employed a simulation protocol summarized in Fig . 2B . The dynamics of the KIF1A protein were simulated by the underdamped Langevin equation at a constant temperature T = 290 . 0 K with CafeMol [47] . The step size dt of the time integration is dt = 0 . 1 τ , where τ∼0 . 128 ( ps ) is the unit of time in CG-simulation . where vi is the velocity of the i-th bead and a dot represents the derivative with respect to time: t ( thus , vi = . ξi is a Gaussian white random force , which satisfies <ξi> = 0 and <ξi ( t ) ξj ( t′ ) > = 2mi γi kBT δij δ ( t -t′ ) 1 , where the bracket denotes the ensemble average and 1 is a 33 unit matrix . kB , is the Boltzmann constant , γi and mi are the friction coefficient and the mass for the one residue . ( See Supporting Information Text S1 for more detail ) . The units of CG-simulations are given as follows: The length unit is 0 . 1 nm . The energy unit is kcal/mol ( ∼6 . 95 pN . nm ) . The unit of mass can be defined as setting mi , the mass for an amino acid . We set mi = 10 , which is just a convention in CafeMol , which leads to the mass unit as 2 . 27510−26 kg . The friction coefficient γi for a residue is decided so that the diffusion constant of the KIF1A head in simulations roughly agrees with that ( 1 . 2×108 [nm2/s] ) by the Stokes-Einstein law: By setting γi = 0 . 1 for an amino acid , we obtained a reasonable diffusion coefficient of KIF1A head ( DKIF1A = 1 . 58×10−5 [nm2/τ] ) in simulation and the time unit in CG-simulation τ ∼0 . 128 ( ps ) . As for the default-sized cargo , we modeled its radius 3000 times as large as the radius of an amino acid , which is ∼1 µm . So , the mass mcargo of the cargo scales as mi×30003 , which gives 2 . 7×1011 . The friction coefficient γcargo for the cargo is 0 . 1×3000/30003 = 1 . 1×10−8 [1/τ] . In this paper , we used the underdamped Langevin dynamics as the equation of motions . We note that , even when we use the underdamped Langevin equation , it gives us overdamped motions when we investigate motions in time scales longer than the velocity correlation time . For an amino acid , the velocity correlation is given by 1/γi = 10 τ ( where the simulation time unit τ corresponds to ∼0 . 128 ps in real time scale ) . For the KIF1A head , we computed the lifetime of the velocity correlation of the center of mass , which was ∼10 τ , while that for the cargo was ∼108 τ . Thus , for the characteristic time scale tc ∼L2/2DKIF1A ∼106 τ of the KIF1A head to find the adjacent binding site ( z = L or -L ) from the dissociation , an amino acid and the KIF1A head behave as overdamped , while the cargo motion is underdamped . | It is one of the major issues in biophysics how molecular motors such as conventional two-headed kinesin convert the chemical energy released at ATP hydrolysis into mechanical work . While most molecular motors move with more than one catalytic domain working in coordinated fashions , there are some motors that can move with only a single catalytic domain , which provides us a possibly simpler case to understand . A single-headed kinesin , KIF1A , with only one catalytic domain , has been characterized by in vitro single-molecule assay to generate a biased Brownian movement along the microtubule . Here , we conducted a set of structure-based coarse-grained molecular simulations for KIF1A system over an ATP hydrolysis cycle for the first time to our knowledge . Without cargo the simulated stand-alone KIF1A could not generate any directional movement , while a large-friction cargo-analog linked to the C-terminus of KIF1A clearly enhanced the forward-biased Brownian movement of KIF1A significantly . Interestingly , the cargo-analog here is not merely load but an important promoter to enable efficient movements of KIF1A . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"physics",
"statistical",
"mechanics",
"theoretical",
"biology",
"biophysics",
"theory",
"biology",
"biophysics",
"simulations",
"biophysics"
] | 2013 | Structure-based Molecular Simulations Reveal the Enhancement of Biased Brownian Motions in Single-headed Kinesin |
Primate T-lymphotropic viruses type 1 ( PTLV-1 ) are complex retroviruses infecting both human ( HTLV-1 ) and simian ( STLV-1 ) hosts . They share common epidemiological , clinical and molecular features . In addition to the canonical gag , pol , env retroviral genes , PTLV-1 purportedly encodes regulatory ( i . e . Tax , Rex , and HBZ ) and accessory proteins ( i . e . P12/8 , P13 , P30 ) . The latter have been found essential for viral persistence in vivo . We have isolated a STLV-1 virus from a bonnet macaque ( Macaca radiata–Mra18C9 ) , a monkey from India . The complete sequence was obtained and phylogenetic analyses were performed . The Mra18C9 strain is highly divergent from the known PTLV-1 strains . Intriguingly , the Mra18C9 lacks the 3 accessory open reading frames . In order to determine if the absence of accessory proteins is specific to this particular strain , a comprehensive analysis of the complete PTLV-1 genomes available in Genbank was performed and found that the lack of one or many accessory ORF is common among PTLV-1 . This study raises many questions regarding the actual nature , role and importance of accessory proteins in the PTLV-1 biology .
The Primate T-lymphotropic virus type 1 ( PTLV-1 ) constitutes a group of deltaretroviruses infecting humans ( HTLV-1 ) or non-human primates ( STLV-1 ) . Phylogenetic studies led to the definition of PTLV-1 viral subtypes [1] . Viruses belonging to the African PTLV-1 subtypes ( i . e . subtypes b , d , e , f and g ) are found both in humans and non-human primates ( NHPs ) , and the human and simian strains are undistinguishable . Continuous zoonotic spillovers of these strains occur , following severe bites by infected non-human primates , or in the context of bushmeat hunting and handling [2–4] . Two subtypes are exclusively human: PTLV-1a is found in human populations scattered throughout the globe , while PTLV-1c is found in indigenous people of the Australomelanesian continent . In contrast , a group of STLV-1 has been described in macaques and great apes in Asia [5–7] . These viruses are genetically distant from other subtypes ( a fact that led some researchers to consider some of them as a separate STLV , STLV-5 ) [8] and have never been found in humans to date . HTLV-1 is estimated to infect at least 5–10 million people worldwide [1] . HTLV-1 is the etiological agent of many severe diseases , ranging from an aggressive lymphoproliferation , the adult T-cell leukemia/lymphoma ( ATL ) , to inflammatory syndromes , such as a neurodegenerative disease called HTLV-1 associated myelopathy or tropical spastic paraparesis ( HAM/TSP ) . Pathogenesis does not seem to be restricted to a certain HTLV-1 subtype; for instance , ATL cases have been reported in patients infected with HTLV-1a , -1b or 1c [9 , 10] . STLV-1 is also oncogenic . ATL have been reported in many simian species , from Macaques to Gorilla [11 , 12] . Deltaretroviruses are complex retroviruses . In addition to the canonical gag , pol , env retroviral genes , the PTLV-1 genome has a series of open reading frames ( ORFs ) encoding regulatory and accessory proteins [13–15] . Regulatory proteins are essential for viral expression and propagation both in vitro and in vivo . In contrast , accessory proteins are optional for viral expression in vitro , but required for viral persistence in vivo . There are three regulatory proteins in PTLV-1: Tax , Rex , and HBZ . PTLV-1 purportedly encodes four accessory proteins , named P12/P8 ( encoded by ORF I ) , P13 and P30 ( encoded by ORF II ) . This study reports the first STLV-1 genome from a virus infecting a bonnet macaque ( Macaca radiata ) , a macaque species from India . The virus replicated well in cell culture , and this material was used for sequencing , a full genome analysis of this divergent Asian STLV-1 strain was performed . While the canonical ORFs , as well as the ORFs encoding regulatory proteins were found conserved , the ORFs encoding the accessory proteins were absent or disrupted . This latter observation is intriguing as these proteins are purportedly essential for viral persistence in vivo . In order to determine if the absence of accessory proteins is specific to this particular strain , a comprehensive analysis of the complete genomes of different PTLV-1 subtypes available in GenBank was performed . Except for HTLV-1a strains , strains from other PTLV-1 subtypes all lacked at least one accessory ORF . This raises many questions regarding the actual role and importance of accessory proteins in the PTLV-1 biology .
The bonnet macaque Mra18C9 was housed in an animal rescue center and material was sent to us for routine diagnostic screening . Serological tests for PTLV-1 gave conflicting results . While the serum tested negative on an in-house ELISA , it reacted positively when tested by a local hospital laboratory . These conflicting results led us to perform additional serological testing using the INNO-LIA HTLVI/II assay ( Fujirebio Europe , Ghent , Belgium ) . The serum reacted strongly with PTLV-1/2 env gp46 and gp21 antigens , but did not react with any type-specific peptide . It was thus considered as an indeterminate sero-reactivity . As the serological tests were inconclusive , we performed a diagnostic PCR screening on the DNA isolated from PBMCs , by using a generic nested PCR assay able to amplify a fragment of the tax/rex region of both PTLV-I and -II . The PCR tested positive , and BLAST analysis on the 118 bp-long fragment revealed high nucleotide identity percentage ( >93% ) with STLV-1 identified from Formosan macaques ( Macaca cyclopis STLV108 , GenBank accession number: KM268809 ) , stump-tailed macaques ( M . arctoides MarB43 and marc1 , GenBank AY590142 and U76625 respectively ) , and long-tailed macaques ( M . fascicularis MFA-C194 , GenBank U59132 ) . The full-length STLV-1 genome infecting the bonnet macaque Mra18C9 was obtained by using a combination of the sequence-independent VIDISCA-454 technique , which generated 4 segments scattered along the STLV genome , and standard PCRs to bridge the sequence gaps and sequence the LTR . The flanking long terminal repeats ( LTRs ) comprise a TATA box , a polyadenylation signal , and three Tax-responsive elements ( TxRE ) type 1 . The TxRE-2 may not be fully functional as an insertion of two nucleotides is present at position 210–211 . Interestingly , the LTR was quite divergent from previously published PTLV-LTR sequences , as the nucleotide identity was lower than 80% ( Table A in S1 Text ) . The canonical retroviral ORFs ( gag , pro , pol , env ) are conserved , as well as the sequences necessary for the gag-pro and pro-pol ribosomal frame-shifts ( Fig 1 ) . Nucleotide and amino acid sequence comparisons confirm that Mra18C9 STLV-1 is a highly divergent PTLV-1 strain . At the nucleotide level , the gag , pro , pol , and env genes show not more than 80% identity with their counterparts from other PTLV , while their encoded proteins differ 10–15% in amino acid identity ( Tables A-B in S1 Text ) . The ORFs corresponding to the regulatory proteins Tax , Rex and HBZ are also present ( Fig 1 ) , and splice-acceptor and -donor sites are preserved . The tax gene sequence shares roughly 82% nucleotide identity , and the encoded Tax protein has a 88–90% amino-acid identity to other PTLV-1 sequences ( Tables A-B in S1 Text ) . Importantly , the PDZ-binding motif of Tax , which is necessary for the interaction of Tax with cellular factors , is mutated in Mra18C9 genome: the ETEV motif is changed into an ETEI motif . Phylogenetic analyses of the concatenated gag-pol-env-tax genes clearly demonstrated that Mra18C9 falls into the Asian STLV-1 group ( Fig 2A ) . Macaque STLV-1 strains all form long branches , and do not aggregate in a monophyletic group . Instead , they form a paraphyletic group . In order to better estimate the relative position of this strain , phylogenetic analyses were performed using the commonly studied LTR and env sequences ( Fig 2B and 2C ) . The different phylogenetic methods ( Neighbor-Joining , and a Bayesian approach ) gave very similar results . Results of the Bayesian analyses are shown in Fig 2 . The Mra18C9 strain was consistently found in a long branch among the macaque strains; the closest known sequences were the Macaca arctoides strains ( when considering either LTR or env sequences ) ( Fig 2B and 2C ) and the Macaca mulata strains MMU-R18 and R22 ( when considering the LTR ) ( Fig 2B ) . In conclusion , the Mra18C9 strain has a typical PTLV-1 organization , but this virus is highly divergent from other PTLV , suggesting a long and independent evolution . Sequence analysis suggests that the accessory proteins encoded by ORF-I may not be functional in the Mra18C9 strain . Indeed , the start codon of the open-reading frame is mutated ( ATG > 6821GCG ) , and multiple stop codons are found . Furthermore , the splice-acceptor site is mutated ( ATK 6380CAG/CAAC > Mra18C9 6719TAA/CAAC ) and may not be functional either . Thus , Mra18C9 does not encode the P12/P8 proteins . The splice-acceptor domain necessary for a P30 protein-encoding mRNA is also mutated ( ATK 6475TAG/CACT > Mra18C9 6814GGA/CGCT ) , suggesting that P30 may not be produced by the Mra18C9 strain . This is reinforced by the presence of an early stop codon in the ORF . Similarly , the splice-acceptor domain , necessary for a P13-encoding mRNA , is also mutated ( ATK 6872CAG/CAGG > MRA18C9 7223CAG/TTGG ) , and therefore the mRNA may not be synthesized . In addition , the ORF is also altered: the start codon is conserved but the stop codon is mutated ( TAA > 7549CAA ) . The resulting protein would then be much longer ( 137 amino-acid long putative protein instead of 87 aa-long ) , which can rigorously influence its functionality . Collectively , our analyses indicate that the STLV-1 Mra18C9 strain completely lacks the accessory genes , which have been reported in HTLV-1a strains . We wondered if the absence of accessory genes was specific to the Asian macaque STLV-1 strain isolated from M . radiata . For this purpose , an in silico analysis of PTLV-1 complete genomes available in GenBank was performed ( Table 1 and Tables C-F in S1 Text ) . Accessory proteins were conserved in HTLV-1a strains and in the available STLV-1b strain [12] . In contrast , the macaque STLV-1 strains lacked all of the accessory genes . The loss of accessory genes in MarB43 was previously reported [7] . All the other strains ( either HTLV-1b , c; or STLV-1e , f , g ) lacked at least 1 accessory protein . For instance , the HTLV-1b strains lack both P12 and P30 , and as previously mentioned HTLV-1c strains lack P12 [18] . In conclusion , although accessory genes have been found important for viral infection and persistence in HTLV-1a , many PTLV-1 strains lack one or more accessory genes .
PTLV-1 infects a wide range of non-human primates ( NHPs ) . We report the first strain infecting a bonnet macaque ( Macaca radiata ) . Complete genome analysis revealed that it is a highly divergent strain when compared to the currently known PTLVs . This would explain the conflicting results obtained by ELISA , as well as its low reactivity ( indeterminate profile ) on the commercial Western blot . Phylogenetic analyses positioned the Mra18C9 strain among other Asian macaque STLV-1 viruses . This confirms the large heterogeneity within the Asian PTLV-1 clade as was described previously [5 , 6] . In phylogenetic trees , Asian macaque STLV-1 strains form very long branches when compared to African ( including HTLV-1a ) strains . This could suggest that these viruses have evolved independently in Asia , with their simian host , for a very long period [19] . The introduction of STLV in Asian NHP has been dated at approximately 200 . 000 years ago [6 , 7] . Under this assumption , the viruses have coevolved for long with their host , and the phylogeny of Asian STLV-1 should mirror the evolution of Asian primates . However , this is not the case for macaque STLV-1 . First , they do not form a monophyletic group; instead they form a paraphyletic group . The strains are organized as a ladder , branching deeply next to the PTLV-1 root ( Fig 2A ) . One could argue that this particular topology results partially from genetic saturation . Moreover , among Asian STLV-1 there are only a few monophyletic groups corresponding to simian species ( Fig 2B and 2C ) . Apart from M . fuscata and M . arctoides STLV-1 , the other clades are composed of sequences of mixed origins , with STLVs from Pongides and Hylobatides that infect macaques . Even when focusing on sequences isolated from macaques , the distribution of the sequences does not follow the known Macaca phylogeny [20 , 21] . Together , this points to interspecies transmission ( between macaques or from macaques to orangutans or gibbons ) of such viruses , as others previously suggested [5 , 6] . Interspecies transmission of STLV-1 has been previously reported , although in the context of captivity [22 , 23]; the evidence of such transmission in natura is mostly inferred by phylogenetic analysis [6 , 24] . Thus , one could argue that the long branching is not only due to a long independent evolution in Asia , but also to an accelerated mutation rate for these strains , which could be related to frequent interspecies transmission . The canonical and the regulatory proteins are present in the genome of the Mra18C9 STLV-1 . The Mra18C9 STLV-1 strain is a functional , replicative virus ( at least in vitro , as it could be amplified on SupT1 cells ) . However , the strain may have an attenuated phenotype . First , the Tax-responsive element 2 ( TxRE2 ) seems disrupted due to a 2-nucleotide insertion [25] . This may lead to a lower basal transcription and reduced viral expression [26] . Second , the viral transactivator Tax has a mutated PDZ-binding motif ( PBM ) . The Tax PBM is essential for sustained proliferation both in vitro and in vivo [27 , 28] . A single mutation of the last amino acid of the PBM was shown to be sufficient to abrogate its function [29–31] . The pX region of the Mra18C9 STLV-1 strain lacks both ORF-I and ORF-II . We first hypothesized that the loss of these ORFs could render the virus less pathogenic . Indeed , accessory proteins have been shown to be important for viral persistence and pathogenesis in HTLV-1a . Mutations of accessory ORFs limit the replicative capacity of HTLV-1a in a rabbit model [32] . Similarly , in the closely related bovine leukemia virus , mutations of the homologue ORFs render the virus attenuated [33] . It was proposed that Australian HTLV-1c strains , because they lack the P12 protein , might be less oncogenic [18] . Nevertheless , ATL cases have been reported in HTLV-1c-infected individuals [9] . Moreover , while the absence of accessory proteins seems to be a general feature of macaque STLV-1 , ATL cases have been reported in naturally infected macaques [11] . Thus , even in the absence of accessory proteins , STLV-1 still present an oncogenic potential . Although STLV-1 is highly prevalent among Asian NHPs , and humans are in frequent contact with macaques and may acquire other retroviral infection , such as simian foamy viruses that are endemic in several species of Asians monkeys [34] , no Asian STLV zoonotic transmission has been reported so far . One could hypothesize that the absence of ORFI and ORFII in macaque STLV-1 can limit viral transmission and propagation . Indeed , HTLV-1a P12 and P30 have been found to be essential for viral replication in human and macaque dendritic cells [35] , which can play a key role in viral transmission . However , this proposition is nullified by our thorough analysis of the different PTLV-1 subtypes . Indeed , HTLV-1b , which is a very common virus in Central Africa , persists and propagates in humans despite the absence of P12 and P30 ( Table 1 ) . In conclusion , the role and importance of accessory proteins needs to be reconsidered in light of the analysis of the different PTLV-1 strains . Indeed , as most studies have focused on HTLV-1a , the 3 accessory proteins were found conserved and their function in viral persistence and transmission was believed to be essential [17 , 32 , 35] . This study indicates that some PTLV-1 can persist in the absence of one or many accessory proteins . This raises the question of a dispensable role of these proteins , or the presence of other accessory proteins yet to be identified in the other PTLV-1 subtypes .
The Bonnet macaque was housed in an animal rescue center ( animal shelter VZW , Belgium ) and material was sent to the BPRC for viral diagnostic screening . The BPRC is fully licensed by the Netherlands Food and Consumer Product Safety Authority ( belonging to the Ministry of Agriculture , Nature and Food Quality ) to work with animal products and perform diagnostic services for third parties ( Approval no . 1926950 ) . Sera were assayed for antibodies to PTLV-1 using an in-house developed serological test with purified , lysed HTLV-1 particles as coating antigen ( Advanced Biotechnologies Inc . , Eldersburg , USA ) [36] . Additionally , the serum from the bonnet macaque was also tested by ELISA in a local hospital laboratory ( DDL , Delft , The Netherlands ) , and a western blot ( WB ) analysis using the INNO-LIA HTLVI/II assay ( Fujirebio Europe , Gent , Belgium ) . Genomic DNA was isolated from whole blood using the QIAamp DNA blood mini kit ( QIAGEN Benelux B . V . , Venlo , The Netherlands ) . A 118 bp tax/rex gene fragment was amplified using the TR101/TR102 and SK43/SK44 nested primer sets , as previously described [37 , 38] . Furthermore , A pan-STLV PCR was performed with primers PH1F and PH2R , as described by van Dooren et al . [39] . The amplified fragment of 192 bp fully overlapped the 118 bp fragment from the first PCR . The virus discovery method VIDISCA-454 was used for the analysis of cell culture supernatant from the PTLV-infected SupT1 cell culture , as previously described [40] . In brief , PBMCs were isolated on a Ficoll and stimulated for 2 days with PHA ( 1 ug/ml final ) . Next , they were co-cultured with SupT1 cells until CPE was visible ( 2–3 weeks ) . The cell culture supernatant was centrifuged to remove cell debris and treated with TURBO DNase ( Ambion , Thermo Fisher Scientific , Breda , The Netherlands ) . Next , nucleic acids were isolated with a QIAamp Viral RNA Mini Kit ( QIAGEN Benelux BV , Venlo , the Netherlands ) and reverse-transcribed with SuperScript II ( Thermo Fisher Scientific ) using non-ribosomal random hexamers . Subsequently , second strand DNA synthesis was performed with 5 U of Klenow fragment ( New England Biolabs , Bioke , Leiden , The Netherlands ) . Double-stranded DNA was purified by phenol/chloroform extraction and ethanol precipitation and digested with Mse I restriction enzyme ( New England Biolabs ) . Adaptors with different Multiplex Identifier sequences ( MIDs ) were ligated to the digested fragments of the different samples . Before PCR amplification , the fragments were purified with AMPure XP beads ( Agencourt AMPure XP PCR , Beckman Coulter , Woerden , The Netherlands ) . A 28-cycles PCR with adaptor-annealing primers was performed . The program of the PCR-reaction was: 5 min 95°C , and cycles of 1 min 95°C , 1 min 55°C , and 2 min 72°C , followed by 10 min 72°C and 10 min 4°C . After purification with AMPure XP beads , the purified DNA was quantified with the Quant-it dsDNA HS Qubit kit ( Invitrogen , Carlsbad , CA , USA ) and diluted to 107 copies/μl . Samples were pooled and Kapa PCR ( Kapa Biosystems , Wilmington , MA , USA ) was performed to determine the quantity of amplifiable DNA in each pool . Subsequently , the Bioanalyser ( hsDNA chip , Agencourt ) was used to determine the average nucleotide length of the libraries . The pools were diluted until 106 copies/μl , titrated with beads ( DNA:beads ratio of 0 . 5:1 , 1:1 , 2:1 and 4:1 ) and used in an emulsion PCR according to the supplier’s protocol ( LIB-A SV emPCR kit ) . Sequencing was done on a 2 region GS FLX Titanium PicoTiterPlate ( 70x75 ) with the GS FLX Titanium XLR 70 Sequencing kit ( Roche , Woerden , The Netherlands ) . Sequence reads were analyzed using the blastn and blastp algorithms ( National Center for Biotechnology Information ) . PCR primers were designed on basis of the four fragments that were obtained from the VIDISCA as well as from the consensus LTR sequence derived from the alignment of other PTLV-1 ( Tables G-H in S1 Text ) . PCR reactions were performed in a final volume of 50 μl . Each amplification reaction was performed in 1x DreamTaq buffer containing 200 μM of each dNTP , 50 pmol of each primer , and 1 . 25 U DreamTaq DNA polymerase ( Thermo Fisher Scientific ) . The amplification reactions were performed for 35 cycles consisting of a 30 s denaturation step at 94°C , a 30 s annealing step at 55°C and an elongation step of 150 sec at 72°C . Amplicon purification and sequencing was performed essentially as described above , but by using a primer-walking sequencing strategy . Sequences were assembled with the SeqMan Pro software ( DNASTAR , Inc . , Madison , USA ) . The resulting contig was further analyzed using the MacVector software package ( MacVector , Inc . , Cambridge , UK ) . The complete genome of STLV-1 Mra18C9 has been deposited at GenBank with accession number MK639100 . LTR ( 701 bases ) and env ( 483 bases ) sequences were aligned together with most sequences available on GenBank using DAMBE [41] . An alignment of concatenated gag-pol-env-tax sequences ( 5820 bases ) was also generated with the sequences of complete PTLV-1 genomes available on GenBank . Phylogenetic trees resulted from analyses using the neighbor-joining method performed with the PAUP* v4 . 0b10 . The final alignment was submitted to the Modeltest program ( version 3 . 6 ) to select , according to the Akaike information criterion , the best model to apply to phylogenetic analyses . The selected substitution models were: Tamura-Nei for env and concatenated gag-pol-env-tax , and general time-reversible ( GTR + γ ) for the LTR . To test the robustness of the tree topologies , 1 , 000 bootstrap replicates were performed . Bayesian approaches were inferred with the MrBayes 3 . 2 . 7 program and robustness was tested with posterior probabilities . Both methods raised similar phylogenetic tree topology . The analysis was performed following the alignment of complete genomes available in GenBank ( Table C in S1 Text ) . Splicing acceptor and donor sequences were previously described for ATK [13] . Frameshift sites had been previously identified [42] . | Primate T-lymphotropic viruses type 1 ( PTLV-1 ) are complex retroviruses infecting both human ( HTLV-1 ) and simian ( STLV-1 ) hosts . It has been shown that the persistence and pathogenesis of these viruses depend on the expression of small , accessory proteins . A bonnet macaque ( a monkey present in India ) was found infected with STLV-1 . The genome was sequenced and found quite divergent from the other STLV-1 genomes previously described . Intriguingly , this virus does not encode accessory proteins . Analysis of other available sequences found that most strains lack at least one accessory gene . Thus the importance and the role of these proteins in the PTLV-1 biology should be revisited . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"taxonomy",
"vertebrates",
"animals",
"mammals",
"primates",
"phylogenetics",
"data",
"management",
"phylogenetic",
"analysis",
"genome",
"analysis",
"sequence",
"motif",
"analysis",
"molecular",
"biology",
"techniques",
"old",
"world",
"monkeys",
"research",
"and",
"an... | 2019 | Absence of accessory genes in a divergent simian T-lymphotropic virus type 1 isolated from a bonnet macaque (Macaca radiata) |
Tracking moving objects , including one’s own body , is a fundamental ability of higher organisms , playing a central role in many perceptual and motor tasks . While it is unknown how the brain learns to follow and predict the dynamics of objects , it is known that this process of state estimation can be learned purely from the statistics of noisy observations . When the dynamics are simply linear with additive Gaussian noise , the optimal solution is the well known Kalman filter ( KF ) , the parameters of which can be learned via latent-variable density estimation ( the EM algorithm ) . The brain does not , however , directly manipulate matrices and vectors , but instead appears to represent probability distributions with the firing rates of population of neurons , “probabilistic population codes . ” We show that a recurrent neural network—a modified form of an exponential family harmonium ( EFH ) —that takes a linear probabilistic population code as input can learn , without supervision , to estimate the state of a linear dynamical system . After observing a series of population responses ( spike counts ) to the position of a moving object , the network learns to represent the velocity of the object and forms nearly optimal predictions about the position at the next time-step . This result builds on our previous work showing that a similar network can learn to perform multisensory integration and coordinate transformations for static stimuli . The receptive fields of the trained network also make qualitative predictions about the developing and learning brain: tuning gradually emerges for higher-order dynamical states not explicitly present in the inputs , appearing as delayed tuning for the lower-order states .
Over the last decade , neuroscience has come increasingly to believe that sensory systems represent not merely stimuli , but probability distributions over them . This conclusion follows from two observations . The first is that the apparent stochasticity of the response , R , of a population of neurons inherently represents the likelihood of the stimulus s: R ∼ p ( r∣s ) [1] . The second is that certain common computations essential to the function of many animals require keeping track of probability distributions over stimuli , rather than mere point estimates . For example , primates integrate information from multiple senses by weighting each sense by its reliability ( inverse variance ) [5 , 6] . This framework has been used to hand-wire neural networks that integrate spatial information across sensory modalities and across time [2 , 7 , 8] . The more challenging problem faced by the brain , however , is to learn to perform these tasks . We have recently shown [4 , 9] that the problem of learning to integrate information about a common stimulus from multiple , unisensory populations of neurons can be solved by a neural network that implements a form of unsupervised learning called density estimation . Such a network learns to represent the joint probability density of the unisensory responses—to build a good model for these data—in terms of the activities of its downstream , multisensory units . For example [4] , an exponential family harmonium ( EFH ) [3] trained on the activities of two populations of Gaussian-tuned , Poisson neurons ( linear probabilistic population codes [2] ) that tile their respective sensory spaces ( visual and proprioceptive , e . g . ) will learn to extract the “common cause” of these populations , encoding the stimulus in its hidden layer . In this case , the unisensory information available on a “trial” can be characterized by two means ( best estimates ) and two variances ( inverse reliabilities ) ; and the estimate extracted by the hidden units of the trained network is precisely the inverse-variance-weighted convex combination that primates appear in psychophysical studies to use . Ecologically , however , the critical challenge is not typically to estimate the location of a static object , but to track the state of a dynamically changing environment . This task likewise requires reliability-weighted combination of information , in this case of the current sensory evidence and the current best estimate of the state given past information . But it is considerably more difficult , since its solution requires learning a predictive model of the dynamics , which is not explicitly encoded in the sensory reports . In the case of Gaussian noise and linear dynamics ( LDS ) , this recursive process is described by the Kalman filter , the parameters of which can be acquired with well-known iterative learning schemes . How the brain learns to solve this problem , however , is unknown . Here we propose a neural model that accomplishes this task . We show that by adding recurrent connections to an EFH similar to that used in [4] , the network can learn to estimate the state of a dynamical system . For concreteness , we consider the problem of tracking the dynamical state of the upper limb , a necessary computation for accurate and precise movement planning and control . In this case , the neural circuit corresponds to the posterior parietal cortex ( PPC ) , which appears to subserve state estimation [10 , 11]; and its inputs are taken to be a population of proprioceptive neurons . The network’s performance can be quantified precisely by restricting our view to linear-Gaussian dynamics , where the filtering and learning problems have known optimal solutions ( respectively , the Kalman filter and expectation-maximization , a maximum-likelihood algorithm ) . And indeed , performance approaches that optimum . We then extend the network to controlled dynamical systems . Under the assumption that the controls are provided by motor cortex , these too are observed only noisily by PPC , in the form of efference copy , which the network must then learn to interpret as motor commands . State estimation is again close to optimal . In addition , the network is neurally plausible in both its representation of stimulus probabilities [2] and in the unsupervised learning procedure , which relies only on pairwise correlations between firing rates of connected neurons [12 , 13] . Finally , the network makes two predictions about neural circuits that learn to perform state estimation: ( 1 ) During learning , position receptive fields will emerge before velocity receptive fields; or more generally , receptive fields will develop from lower- to higher-order states , especially when explicit information about the higher-order states is not in the inputs . ( 2 ) Filtering is implemented by tuning to past positions ( or more generally , lower-order states ) , rather than tuning directly to velocity ( or more generally , higher-order states ) .
More than one cortical area is thought to subserve object tracking . Since we have in this study focused on the task of tracking one’s own limbs , we consider posterior parietal cortex ( PPC ) , which is thought to be responsible for this task [10 , 11] . The computation may well be distributed across the PPC , but we focus on just one that has been particularly implicated [11] , Brodmann Area 5 . Our aim is to show that our neural network and its learning scheme are consistent with what is known about the connectivity of Area 5 , both interlaminar and inter-areal . In particular , we consider its connections with the primary motor area ( M1 ) and primary somatosensory cortex ( S1 ) . Our proposed implementation is speculative and not the only one possible; e . g . , we identify the “recurrent” units with another layer of Area 5 , but they might alternatively correspond to another area of PPC . Fig 9A summarizes the training procedure from an algorithmic perspective ( see Methods for details ) . In Fig 9B , as in Fig 9A , input comes from two sources . Feedforward , proprioceptive input ( R t θ ) from primary somatosensory cortex , S1 ( especially BA3a ) , projects to layer IV [23] . A copy of the efferent command ( R t u ) feeds back from M1 to layer I of Area 5 [23] . Layer II/III of Area 5 in turn projects forward to M1 [24] . Layer I is not believed to contain cell bodies [25] , so we take these to be the terminal branches of the apical dendrites of layer II/III cells ( which are also lightly labeled by anterograde tracers injected in M1 [23] ) . Within Area 5 , we propose that the temporally delayed recurrency ( Zt−1 ) of the rEFH is provided by the loop from layer II/III down to VI , then up to V , before modulating the activity of layer II/III neurons , consistent with the anatomy of Area 5 [25] . Layer IV and III , as well as V and III , also have reciprocal connections [25] , as required for the rEFH training procedure . The latter loop has in fact been hypothesized to give rise to rhythmic activity in rat parietal cortex [26] . According to the learning and filtering schemes of our model , the temporal flow of information is as follows . Sensory input ( r t θ ) and efference copy ( r t u ) arrive at , respectively , layer IV of BA5 and the feedback layer ( presumably VI ) of M1 . At the same time , a “copy” ( which could be any information-preserving transformation ) of activity from layer II/III of BA5 ( zt-1 ) passes down to layer V . Next , the spiking in these layers ( M1 layer VI , BA5 layer IV , BA5 layer V ) drives spiking ( zt ) in BA5 layer II/III . These responses encode , according to the model , the optimal estimate of the limb , and this information will ultimately become the temporally delayed recurrent activities identified above . For learning , however , it is also necessary that this activity drive spiking in M1 ( r ^ t u ) , BA5 layer IV ( r ^ t θ ) , and BA5 layer V ( z ^ t ) , through the reciprocal connectivity lately noted . A “copy” of the layer II/III activity ( zt ) is simultaneously propagated down to layer VI . Lastly , the activities in M1 , BA5 layer IV , and BA5 layer V again drive activity ( z ^ t ) in BA5 layer II/III . At the same time , the “copy” of layer II/III activity ( zt ) is communicated up to layer V .
We have shown that a neural network ( the “rEFH” ) with a biologically plausible architecture and synaptic-plasticity rule can learn to track moving stimuli , in the sense that its downstream ( “hidden” ) units learn to encode ( nearly ) the most accurate estimate of stimulus location possible , given the entire history of incoming sensory information ( Figs 1 and 2 ) . This requires learning a model of the stimulus dynamics . This is ( as far as we know ) the first biologically plausible model that has been shown to learn to solve this task . Moreover , the network learns the reliability of the sensory signal: the trained network leans more heavily on the internal model when the sensory signal is less reliable , and more heavily on the sensory signal when it is more reliable ( Fig 6 ) . We are particularly interested in tracking the state of one’s own limbs . Here , additional information about stimulus location is thought to be available in the form of a “copy , ” relayed to the posterior parietal cortex , of the efferent motor command [21] . And indeed , when such signals are available to our network , it learns to make use of them appropriately to track the arm more precisely—in spite of the fact that none of the incoming signals is “labeled” according to its role ( Fig 3 ) . Although an expectation-maximization ( EM ) algorithm can sometimes learn a Kalman filter that noticeably outperforms the best rEFH on these data , it usually does not ( Fig 4 ) . That is , learning in the rEFH is more robust than EM in the sense that the variance in performance across models trained de novo is smaller , albeit at the price of a bias towards worse models . Finally , and surprisingly , the downstream neurons of the trained network track a moving stimulus by encoding its position at various time lags ( Fig 5 ) . The earliest implementation of dynamical state estimation ( “filtering” ) in neural architecture comes from Rao and Ballard [27] . Their model , like ours , assigns a central role to recurrent connections , but as predictive coders rather than simply delayed copies of previous neural states . Likewise , the network connectivity is acquired with an unsupervised and local learning rule , a variant on EM . However , the authors do not train their network on moving objects or moving images , presumably because convergence of the neural state under their learning scheme is slow compared with any plausible stimulus dynamics . Instead , the connectivity is acquired on static images . Performance on state-estimation tasks is not tested . Several groups have hand wired neural networks to act as state estimators [7 , 8 , 28] . Although these papers do not address our central concern , the learning problem , it is nevertheless useful to compare the resulting architectures with our rEFH . For example , Beck and colleagues constructed a neural network to implement the Kalman filter on linear probabilistic population codes , as in this work , and showed its performance ( measured in information loss ) to be nearly optimal . From analytical considerations , the authors showed that the required operations on neural firing rates are weighted summation ( as in our network ) and a quadratic operation ( that acts like a divisive normalization in the steady state ) . In our rEFH , on the other hand , the only nonlinearities are elementwise: interaction between inputs is always in the form of a weighted sum . That the rEFH can nevertheless filter ( nearly ) optimally is possible because we do not require , as they do , that the output population encode information in the same way as the inputs ( sc . , that the posterior distribution over the stimulus have linear sufficient statistics; see S4 Text ) . This critical difference provides the basis for an experimental discrimination between the respective models . Likewise , filters have been hand wired into attractor networks [28] and spike-based ( rather than rate-based ) networks [8] . The latter in particular argues that the precise arrival time of spikes contains information about the stimulus , rather than the average rate across time , as in in our model . An approach that does include learning comes from Huys , Zemel , Natarajan , and Dayan [29 , 30] . The authors formulate the problem in terms very similar to ours , but they allow more general dynamical systems generated by Gaussian processes , and the basic unit of information is spikes rather than spike counts ( although approximations that ignore precise arrival times lose little information [29] ) . The most significant difference with our work is that the authors learn the parameters of their network with a supervised , non-local rule , which they do not consider to be a biological mechanism . But again the comparison is instructive . We are able to formulate an unsupervised rule because we approach the filtering problem indirectly: Natarajan and colleagues require the posterior distribution , conditioned on hidden-unit activities , to be factorizable over hidden-unit spikes ( so that a third layer can consider those spikes separately ) , and then force it to match the true filtering distribution by directly descending the KL divergence between them [30] . We , on the other hand , force the network to be a good model of its incoming data—which , when some of those data are past hidden-unit activities , achieves the same end . In the machine-learning literature , Hinton and colleagues have proposed three variants on a theme quite similar to ours [31–33] , although different in important ways . Most importantly , in all three , the past hidden-unit activities are treated by the learning rule as ( fixed ) biases rather than as input data; i . e . , they cannot be modified during the “down pass” of contrastive-divergence training . That these activities ought to be treated as data , we argue more rigorously in a forthcoming work . The earliest variant [31] , the “spiking Boltzmann machine , ” is , like ours , a temporal extension of the restricted Boltzmann machine that is trained with the contrastive-divergence rule . Hidden units are directly influenced by past hidden-unit activities , as with the rEFH , but possibly from temporal distances τ that are greater than one time step ( contra the rEFH ) . However , the weights from a particular “past” hidden unit at various delays ( e . g . , from z t − n τ i , n ∈ { 1 , 2 , . . . } to zt j ) are constrained to be identical up to a fixed ( not learned ) exponential decay . The motivation was to model the influence of past spikes in a biologically plausible way: Whereas in our rEFH , the ( one-time-step delayed ) past hidden activities are maintained in a separate population of neurons ( Fig 9 ) , in the “spiking Boltzmann machine” their effect on current hidden units is interpreted simply as the decaying influence of their original arrival . This makes it plausible , unlike in the rEFH , to include influences at delays greater than one time step . On the other hand , it necessitates treating those effects as biases rather than data . It is difficult to judge the limitations this imposes on the model , since the authors do not quantify its performance . However , they do investigate more thoroughly performance of a similar , but more powerful network . The “temporal restricted Boltzmann machine” ( TRBM ) [32] is a spiking Boltzmann machine without the constraint that the weights decay exponentially backwards in time; instead , they are learned freely and independently for all time . The order of the dynamical system that can be learned by this network turns out , unlike ours , to be tied to τ: TRBMs with τ = 1 ( like the rEFH ) can learn only random-walk behavior ( first-order dynamics ) [33] . This can ( presumably ) be overcome by including connections back as many time steps as the order of the system to be learned , but it is not obvious what biological mechanism could maintain copies of past activities at distant lags , or determine a priori how many such lags to maintain . The same authors show that this problem can be alleviated with a variant architecture , the “recurrent temporal RBM” ( RTRBM ) [33] , but it requires a non-causal learning rule ( backpropagation through time ) , again making it a poor model for neural function . For neither model do the authors precisely quantify its filtering performance; we do in a forthcoming study . Our simulations demonstrate three things: First , the rEFH is capable of learning to “track” moving stimuli , i . e . to estimate their dynamical state , and nearly as well as an optimal algorithm , as has been seen behaviorally in humans [34] . In fact , the network learns to encode the full posterior distribution over the stimulus , rather than just its peak: although we did not show it directly , it must , since the variance of this ( Gaussian ) distribution is required to combine properly the previous best estimate with the current sensory information . And rather than relying on a fixed estimate of sensory reliability , the network learns to take into account instantaneous changes in it ( Fig 6 ) . Second , the network does not require a special architecture or ad hoc modifications . It is , rather , identical , up to the choice of input populations , to the network and learning rule in our previous work [4] . Thus , if the input populations are proprioceptive and recurrent units , it will learn to estimate dynamical state; if they also include efference copy , it will learn the influence of motor commands on stimulus dynamics . If they are proprioceptive and visual reports of a common stimulus , it will learn to perform multisensory integration; if a gaze-angle-reporting population is also present , to transform the visual signal by that angle before integrating ( “coordinate transformations” ) ; if the stimulus distribution is non-uniform , to encode that distribution [4] . ( We have shown elsewhere , in terms of information theory , why this is the case [9] . For further discussion of the relationship between the static and dynamical computations , see S2 Text . ) Thus , the network provides a very general model for posterior parietal cortex , where some combination of all of these signals is often present . Third , the model makes some predictions about the encoding scheme , receptive fields , and connectivity of cortical areas that track objects . As with all models , we take certain elements of ours to be essential and others to be adventitious . That learning in posterior-parietal circuits can be well described as a form of latent-variable density estimation , for example , is central to our theory; but the precise form of the learning rule ( “one-step contrastive divergence” ) , although plausible , is not . Our theory requires that sensory neurons encode distributions over stimulus position , but the representation scheme need not be probabilistic population codes of the Pougetian variety [2] . Here we list three predictions that do follow from essential aspects of the network . The network learns to track by encoding past positions . This is a non-obvious scheme ( it is not , e . g . , the one used by the Kalman filter ) and apparently results from the fact that only position information is reported by the sensory afferents . It is possible that such receptive fields ( Fig 5A ) are in fact found in MSTd of monkeys that have been trained to track moving objects [22] . Now , in many circuits , velocity is detected at early stages . But even when velocity is directly reported by the inputs to an rEFH , tuning to past positions still appears , albeit with lower prevalence ( see S3 Text ) . More generally , we predict that higher derivatives ( e . g . , acceleration ) , especially those not directly available in sensory input , will be encoded via delayed , lower derivatives ( e . g . , velocities ) —as long as those higher-order states have lawful dynamics . During learning , receptive fields for position emerge before those for velocity . This is a necessary consequence of density estimation on recurrent units . A similar proviso attaches: where velocity information is directly reported , it is acceleration-coding that will emerge over time . The use of delayed , feedback connections in neural circuits is a mechanism for learning dynamical properties of stimuli . Under this prediction , primary sensory areas that process information with very little temporal structure—e . g . , smell—will lack the dense feedback found in , e . g . , visual areas . Alternatively , the recurrency might be identified with interlaminar , rather than interareal , structure , as we have hypothesized ( Fig 9B ) —which would explain why piriform cortex only needs three layers . More generally , our investigation was motivated by two main ideas . The first is that populations of neurons , in virtue of their natural variability , encode probability distributions over stimuli ( rather than point estimates ) [1 , 2] . Encoding certainty or “reliability” is a necessity for optimal integration of dynamic sensory information , since it determines the relative weight given to ( a ) current sensory information and ( b ) the prediction of the internal model . But rather than explicitly encoding the reliability of the stimulus location—e . g . , via neurons that are “tuned” to reliability , as other neurons are tuned to location itself—this reliability is identified with the inverse variance of the posterior distribution over the stimulus , p ( s∣r ) , conditioned on the population activity [2] . This distribution arises as a natural consequence of the ( putative ) fact that neural responses are noisy , and can therefore be characterized by a likelihood , p ( r∣s ) [1] . If reliability were not encoded this way , our learning scheme would not work: it would have no way of knowing what to do with those reliabilities , which would be to it indistinguishable from ( e . g . ) the location of another stimulus . The second idea is that higher sensory areas , like posterior parietal cortex and MSTd , can encode more precise distributions over the location ( e . g . ) of a stimulus than that provided by their sensory afferents at any given moment in time . This is due , essentially , to the continuity of the physical world: at successive moments in time , objects tend to remain near their previous positions . More precise localizations can consequently be achieved by a form of averaging that , because objects do move , accounts for the predictable changes in position from moment to moment . This requires learning a model of those predictable changes . The rich statistical structure of the sensory afferents—including efference copy of motor commands that may be influencing the evolution of the stimulus to be tracked , as when tracking one’s own limbs—makes it possible to learn the model from those inputs alone . This unsupervised learning is a much more efficient approach than trying to use the few bits of information that may be available in the form of reward: very few rewards can be reaped before an animal can control its own limbs . In the special case of linear dynamics and Gaussian noise , these two problems—learning a dynamical model , and filtering in that model—have known algorithmic solutions: an expectation-maximization algorithm and the Kalman filter , respectively . Rather than try to map operations on vectors and matrices directly onto neural activity and learning rules , we have taken a more general approach , showing how a rather general neural-network architecture that tries to build good models for its inputs can learn to solve the problem , if those inputs are suitably chosen: temporally delayed recurrent activity from downstream units must be among the inputs . Our network learns by a local , Hebbian rule operating on spike-count correlations , although it remains to relate these to more specific biological learning rules , like STDP .
We describe the most general dynamical system and observation model to be learned: a controlled , second-order , discrete-time , stochastic , linear dynamical system , whose “observations” or outputs come in the form of linear probabilistic population codes [2]; cf . Fig 3A . The uncontrolled model of the section Uncontrolled dynamical system ( Fig 1A ) is a special case ( see below ) . We interpret the plant to be a rotational joint , so distance is in units of radians; and the control to be a torque , hence in Joules/radian . The primary rationale for our choice of dynamics and observation model was to show what kinds of computational issues the recurrent , exponential-family harmonium ( rEFH ) can overcome—issues which it must overcome if it is to be a good model for the way cortex learns to solves the problem . In particular , it might appear that the rEFH can learn relationships only between its current inputs and the previous ones , since its recurrent inputs are from the previous time step only ( see Fig 9A ) . Therefore , we let the inputs report position only , but make the ( hidden ) dynamics second-order: velocity , as well as position , depends on previous position and velocity . If the rEFH can learn to associate only current and previous inputs , it can learn only first-order dynamics from these data . Furthermore , to clearly distinguish models that have learned second-order dynamics from those that have learned only a first-order approximation , we let the true dynamics be a ( damped ) oscillator ( first-order systems cannot oscillate ) . Although the demonstration is in terms of positions and velocities , the point is more general: if the rEFH can learn second-order dynamics from position reports , it can learn higher temporal dynamics from lower-order data more generally . The controlled , single-joint limb obeys: p ( θ t + 1 | θ t , u t ) = N ( A θ t + b u t + μ θ , Σ θ ) , ( 1 ) where the vector random variable Θt consists of angle and angular velocity . The control signal ( torque ) has itself first-order dynamics: p ( u t + 1 | u t ) = N ( α u t + μ u , σ u 2 ) , ( 2 ) making the combined system third-order . The initial state and control are also normally distributed: p ( θ 0 , u 0 ) = N ( ν 0 , ϒ 0 ) . ( 3 ) The current ( time t ) joint position and control are noisily encoded in the spike counts of populations of neurons , whose Gaussian-shaped tuning curves ( fi ) smoothly tile their respective spaces , proprioceptive ( angle ) and control ( torque ) . Spike counts are drawn from ( conditionally ) independent Poisson distributions: p ( r t θ | θ t , g t θ ) = ∏ i Pois [ r i , t θ | g t θ f i ( C θ t ) ] , p ( r t u | u t , g t u ) = ∏ i Pois [ r i , t u | g t u f i ( h u t ) ] , ( 4 ) with C = [1 0] and h = 1 . Here the gt are “gains , ” scaling factors for the mean spike count [2 , 4] . Because the signal-to-noise ratio increases with mean for Poisson random variables , these gains essentially scale ( linearly ) the reliability of each population . Therefore , in order to model instant-to-instant changes in sensory reliability , the gains of each population were chosen independently and uniformly: p ( g t θ ) = U ( 6 . 4 , 9 . 6 ) , p ( g t u ) = U ( 6 . 4 , 9 . 6 ) . ( 5 ) Since the discrete time interval for a single draw from Eq . 4 is 0 . 05 s ( see below ) , these gains correspond to maximal firing rates between 130 and 192 spikes/second , reasonable rates for neurons in cortex . The joint distribution of the states , controls , their observations , and the gains is the product of Eqs 1–5 , multiplied across all time . In accordance with the broad tuning of higher sensory areas , the “standard deviation , ” σtc , of the tuning curves , f i ( x ) = exp { − ( x − ξ i ) 2 2 σ tc 2 } , was chosen so that the full-width at half maximum is one-sixth of the space of feasible joint angles/torques , for all preferred stimuli ξi . However , joints and torques can in fact leave these “feasible spaces”: Although the system was designed to be stable ( eigenvalues of the state-transition matrix are within the unit circle ) , trajectories are nevertheless unbounded , since the input noise is unbounded ( normally distributed ) . We chose not to impose hard joint and torque limits , because this would make the dynamics nonlinear , vitiating the optimality calculations . Instead , stimuli beyond the feasible space simply “wrap” onto the opposite side of encoding space; that is , each population tiles its corresponding stimulus modulo the length of its feasible space . But for the dynamical systems on which model performance was tested , parameters were chosen to make wrapping unlikely ( but cf . the “no-spring” model described below ) . In particular , we used the discrete-time approximation to a damped harmonic oscillator , i . e . , m θ ¨ + c θ ˙ + k θ = u: A = [ 1 Δ − k m Δ 1 − c m Δ ] , b = [ 0 Δ m ] , with moment of inertia m = 5 J⋅ s2/rad2 , viscous damping c = 0 . 25 J⋅ s/rad2 , ideal-spring stiffness k = 3 J/rad2 , and sampling interval Δ = 0 . 05 s . This makes the system stable and underdamped ( oscillatory ) . The control decay , α , in Eq 2 was set to 0 . 9994 , making the dynamics close to a random walk , but mildly decaying towards zero . These parameters and the noise variances were chosen so that the system could not be well approximated by a lower-order one—i . e . , so that the uncontrolled and controlled systems were “truly” second- and third-order ( respectively ) . This was accomplished by ensuring that the Hankel singular values [14] for the system , with output matrix C = [1 0] and input matrix set by the noise variances , were within one order of magnitude of each other; that is , ensuring that the transfer function from noise to joint angle had roughly equal power in all modes . For the uncontrolled system , this was achieved with Σθ = diag ( [5e-7 , 5e-5] ) ; for the controlled system , Σθ = diag ( [5e-5 , 1e-6] ) and σ u 2 = 7 . 5 E − 4 . While this last choice of noise is large enough to ensure that the control’s contribution to the dynamics is significant , it is also small enough to keep wrapping rare . This facilitates the comparison between the benchmark models ( see below ) , which are acquired from non-wrapped trajectories , and the rEFH , which learns from sensory inputs with periodic tuning curves . That is , for fast enough trajectories on a circle , the dynamics would no longer be locally linear , and the learning and filtering tasks no longer comparable . The only other difference between the uncontrolled and controlled dynamical systems was that the former had , of course , no control signal ( or simply b = 0 ) and no control observations ( efference copy ) . For all models , the bias terms were set to zero: μθ = 0 and μu = 0 . The initial positions for all trajectories were drawn from a uniform distribution across joint space ( shoulder θ ∈ [−π/3 , π/3] radians; Fig 1C ) , up to a margin of 0 . 05 radians from the joint limits ( to discourage state transitions out of the feasible space ) ; for EM learning ( see below ) , this was treated as an infinite-covariance Gaussian centered in the middle of joint space . The initial velocity and initial control were normally distributed very tightly about zero , with a standard deviation of 5 E − 5 ( rad/s and J/rad , resp . ) . Hence ν0 = 0 , Υ0 = diag ( [∞ , 5e-10 , 5e-10] ) . The range ( modulus ) of feasible controls is u ∈ [−1 . 25 , 1 . 25] J/rad . For the receptive-field ( RF ) analyses , we used a third dynamical system . In the harmonic oscillator , whether driven or undriven , the non-zero stiffness ( k above ) couples velocity to position , making high speeds and far-from-zero positions unlikely to co-occur . This makes the RF analysis unreliable in the “corners” of position-velocity space , and the overall velocity-encoding harder to interpret . For the analyses presented in Figs 5 , 7 and 8 , therefore , we trained a ( third ) rEFH on a simplified version ( “no-spring” ) of the uncontrolled dynamics , setting the spring constant to zero ( eliminating oscillations ) . To encourage full exploration of the space , the variance of the state-transition noise was also increased by a factor of 50 . The more and less autocorrelated variants of Fig 5D were created by simply scaling up or down the damping coefficient: from left to right , c = 0 . 25/4 , 0 . 25/2 , 0 . 25 , 0 . 25 * 2 , 0 . 25 * 4 . For completeness , we nevertheless include , in the Supplement , the harder-to-interpret RF analyses for the rEFH trained on the ( undriven ) harmonic oscillator ( S3 Text ) . The network is very similar to that in [4] , but we repeat the description here briefly . The harmonium is a generalization of the restricted Boltzmann machine ( RBM ) beyond Bernoulli units to other random variables in the exponential family [3] . That is , it is a two-layer network with full interlayer connections and no intralayer connections , which can be thought of as a Markov random field ( undirected graphical model ) or as a neural network . In our implementation ( see Figs 1B and 3B ) , hidden units ( turquoise , Zt ) and recurrent units ( dark turqoise , Zt−1 ) are binary ( spike/no spike ) , and the “proprioceptive” ( orange , R t θ ) and “efference-copy” ( purple , R t u ) populations are non-negative integers ( spike counts ) . For all networks , the number of recurrent units is the same as the number of downstream or “hidden” units , because recurrent units at time t carry the activities of the hidden units at time t − 1—making the harmonium recurrent through time ( rEFH ) . We chose Nhid = Nrecurrent = 240 for the network trained on the uncontrolled system , and Nhid = Nrecurrent = 180 for the controlled system . We used fifteen proprioceptive units ( Nprop ) and , for the network trained on the controlled system , fifteen efference-copy units ( Nefcp ) , so the total number of “observed” ( or “input” ) variables was 255 = Nrecurrent + Nprop for the uncontrolled model and 210 = Nrecurrent + Nprop + Nefcp for the controlled model . During training and testing , the layers of the rEFH reciprocally drive each other , yielding samples from the following distributions: Z t ∼ q ( z t | z t − 1 , r t θ , r t u ) = ∏ i N hid Bern [ { z t } i | σ ( { W fb z t − 1 + W prop r t θ + W ctrl r t u + b hid } i ) ] ( 6a ) Z t − 1 ∼ q ( z t − 1 | z t ) = ∏ i N hid Bern [ { z t − 1 } i | σ ( { W fb T z t + b fb } i ) ] ( 6b ) R t θ ∼ q ( r t θ | z t ) = ∏ i N prop Pois [ { r t θ } i | exp ( { W prop T z t + b prop } i ) ] ( 6c ) R t u ∼ q ( r t u | z t ) = ∏ i N efcp Pois [ { r t u } i | exp ( { W efcp T z t + b efcp } i ) ] , ( 6d ) which corresponds to Gibbs sampling from the joint distribution represented by the harmonium , q ( zt , z t − 1 , r t θ , r t u ; W , b ) . The letter q is used for the probability density function assigned by the rEFH to distinguish it from the true distribution over the observed variables , p ( r t θ , r t u ) . Here the notation {x}i means the ith element of the vector x; the matrices W and vectors b are the synaptic connection strengths ( “weights” ) and biases , respectively; and the neural nonlinearities , the logistic ( σ ( x ) = 1/ ( 1 + e−x ) ) and exponential funtions , were chosen to produce means for each distribution that are in the appropriate interval ( [0 , 1] and ℝ + , resp . ) . The entire procedure is depicted graphically in Fig 9A . For the graphical models in Figs 1A and 3A , the solution to the filtering problem can be assimilated to a variant on the Kalman filter , and therefore computed in closed form . This is because , although the emission p ( r t θ | θ t ) is not a Gaussian distribution over r t θ , it is a Gaussian function of θt [4 , 7] ( i . e . , the likelihood is an unnormalized Gaussian over θt ) —or more precisely , of C θt , with C the observation matrix ( see Eq 4 ) —and this is the critical requirement for the derivation of Kalman’s recursive solution . The resulting modification is small: Where the emission variance and the ( Gaussian-distributed ) emission appear in the standard KF equations , we substitute , respectively , the scaled tuning-curve width , σ tc 2 / ∑ i r i θ , and the center of mass of the population response , ∑ i ξ i r i θ / ∑ i r i θ [16] . The same applies , mutatis mutandis , to the controls . In fact , the “controlled” case provides no additional complexity , since it corresponds to an uncontrolled third-order system ( since the control has its own dynamics ) whose state Xt is the concatenation of ϴt and Ut: p ( x t + 1 | x t ) = N ( Γ x t + μ x , Σ x ) , ( 10 ) with Γ : = [ 1 Δ 0 − k m Δ 1 − c m Δ Δ m 0 0 α ] , μ x : = [ μ θ μ u ] , Σ x : = [ Σ θ 0 0 σ u 2 ] . In both cases , then , the posterior ( filtering ) distribution over the state is always Gaussian , so at every time step , one computes the posterior mean and covariance , which can be expressed in terms of the filtering distribution at the previous time step , and of the current sensory information . A full derivation appears in S1 Text . Eq 10 ignores some independence statements asserted by the graph of Fig 3A . In fact , an EM algorithm that accounts for them can be derived; but in our experiments , this algorithm does not achieve superior results to the “agnostic” version that tries to learn unconstrained versions of Γ , μx , and Σx . Therefore , results for EM3 throughout use the unconstrained version of the algorithm . In the section Learned receptive fields and connectivity , in order to determine how the network has learned to solve the filtering problem , we sort hidden units by their “preferred” lags and “preferred” angles . These were computed as follows . First , we generated a new set of 40 trajectories of 1000 time steps apiece . Then we computed hidden-unit mean activities , i . e . , their probability of firing ( these are the same quantity because the hidden units are conditionally Bernoulli random variables ) . Angular positions for all 40000 time points were then discretized into 30 bins of uniform width spanning the feasible joint space . For each hidden unit , the following calculation was then performed . First , the empirical mutual information ( MI ) was computed , according to the standard formula [19] , between the two discrete random variables: the discretized position ( 30 bins ) and the binary ( spike/no spike ) hidden-unit response . Next , to reject spurious MI ( which will anyway be rare , given the number of data ) , for each of 20 reshuffles , the unit’s response was shuffled in time and the MIs recalculated . If the unit’s unshuffled MI fell below the 95th percentile of its shuffled MIs , the unshuffled MI was set to zero . The entire procedure was then repeated with the response time-shifted forward by one step , for each of 40 steps . Finally , the “preferred” lag was selected to be the time shift for which MI was maximized . These were used to sort the receptive fields in Figs 5A and 5B , 7 and 8B . For each unit , a “lagged” tuning curve can be constructed by considering its mean responses to past ( discretized ) stimuli; in particular , to stimuli at that unit’s preferred lag . These are the curves plotted as a heat map in Fig 5C , where they have been sorted by the locations of the tuning curves’ peaks . The same locations were used to sort the weight matrix in Fig 8A . Inverting the process , one can ask how well these tuning curves explain the receptive fields in the space of non-delayed position and velocity ( Fig 5A ) : apply each tuning curve to each of the 40000 stimuli , delay the responses by the units’ preferred lags , and then compute receptive fields with these responses . This is how Fig 5B was constructed . Finally , comparing the distribution of preferred lags ( Fig 5D ) to the autocorrelation of the stimulus required computing the autocorrelation of a circular variable ( angle ) . We used the angular-angular correlation measure given by Zar [20] . | A basic task for animals is to track objects—predators , prey , even their own limbs—as they move through the world . Because the position estimates provided by the senses are not error-free , higher levels of performance can be , and are , achieved when the velocity and acceleration , as well as the position , of the object are taken into account . Likewise , tracking of limbs under voluntary control can be improved by considering the motor command that is ( partially ) responsible for its trajectory . Engineers have built tools to solve precisely these problems , and even to learn dynamical features of the object to be tracked . How does the brain do it ? We show how artificial networks of neurons can learn to solve this task , simply by trying to become good predictive models of their incoming data—as long as some of those data are the activities of the neurons themselves at a fixed time delay , while the remainder ( imperfectly ) report the current position . The tracking scheme the network learns to use—keeping track of past positions; the corresponding receptive fields; and the manner in which they are learned , provide predictions for brain areas involved in tracking , like the posterior parietal cortex . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Learning to Estimate Dynamical State with Probabilistic Population Codes |
Gene expression during spore development in Bacillus subtilis is controlled by cell type-specific RNA polymerase sigma factors . σFand σE control early stages of development in the forespore and the mother cell , respectively . When , at an intermediate stage in development , the mother cell engulfs the forespore , σF is replaced by σG and σE is replaced by σK . The anti-sigma factor CsfB is produced under the control of σF and binds to and inhibits the auto-regulatory σG , but not σF . A position in region 2 . 1 , occupied by an asparagine in σG and by a glutamate in οF , is sufficient for CsfB discrimination of the two sigmas , and allows it to delay the early to late switch in forespore gene expression . We now show that following engulfment completion , csfB is switched on in the mother cell under the control of σK and that CsfB binds to and inhibits σE but not σK , possibly to facilitate the switch from early to late gene expression . We show that a position in region 2 . 3 occupied by a conserved asparagine in σE and by a conserved glutamate in σK suffices for discrimination by CsfB . We also show that CsfB prevents activation of σG in the mother cell and the premature σG-dependent activation of σK . Thus , CsfB establishes negative feedback loops that curtail the activity of σE and prevent the ectopic activation of σG in the mother cell . The capacity of CsfB to directly block σE activity may also explain how CsfB plays a role as one of the several mechanisms that prevent σE activation in the forespore . Thus the capacity of CsfB to differentiate between the highly similar σF/σG and σE/σK pairs allows it to rinforce the cell-type specificity of these sigma factors and the transition from early to late development in B . subtilis , and possibly in all sporeformers that encode a CsfB orthologue .
Developmental transcription networks underlie all cellular differentiation processes . These networks usually integrate a variety of environmental and cellular inputs to activate regulators such as transcription factors that control the expression of cell type-specific genes [1 , 2] . Superimposed onto these transcription networks , are protein-protein interaction , signal transduction and metabolic networks [3] . Overall , the combination of these different networks ensures the timely production of proteins and other components essential for the morphogenesis of newborn cells . Ultimately , a complete understanding of cellular differentiation requires detailed knowledge of how the circuitry of transcriptional regulators influences global gene expression in space and time to drive cell morphogenesis at sequential stages of development . Spore formation in the bacterium Bacillus subtilis is an example of a prokaryotic cell differentiation process . At the onset of sporulation , triggered by severe nutrient scarcity , the rod-shaped cell divides close to one of its poles producing a small forespore , the future spore , and a larger mother cell ( Fig . 1A ) . The mother cell nurtures development of the forespore , but undergoes autolysis to release the mature spore at the end of the process . Soon after asymmetric division , the mother cell engulfs the forespore , which becomes isolated from the external medium and separated from the mother cell cytoplasm by a double membrane and an intermembrane space . Following engulfment completion , gene expression in the mother cell drives the last stages of spore maturation by promoting the assembly of concentric protective structures . In parallel , gene expression in the forespore prepares the future spore for dormancy . The sporulation regulatory network includes four RNA polymerase sigma subunits that are activated in a cell type-specific manner and define a regulatory cascade that constitutes the core of the transcription network . σF and σE control early stages in development in the forespore and in the mother cell , respectively . At late stages of development , i . e . , post-engulfment , σF is replaced by σG , and σE is replaced by σK ( Fig . 1A ) . The genes for the cell type-specific sigma factors are part of a genomic signature for sporulation [4 , 5 , 6 , 7] . A second level of regulation results from the action of additional transcription factors , three in the mother cell ( SpoIIID , GerR , and GerE ) , and two in the forespore ( RsfA and SpoVT ) , which are less conserved among spore-formers . The combination of the primary regulators ( i . e . , the sporulation sigma factors ) and the secondary regulators ( ancillary transcription factors ) organizes the two cell-specific lines of gene expression into a series of interlocked type-1 coherent and incoherent feed-forward loops ( FFLs ) [8 , 9] ( Fig . 1B ) . Coherent type-1 FFLs are used as persistence detectors whereas incoherent type-1 FFL generate pulses of gene expression ( reviewed by [3] ) . The first mother cell-specific sigma factor , σE , has a central role in controlling engulfment . Together with the ancillary factor SpoIIID , σE also turns on the genes required for the synthesis of pro-σK and the machinery that triggers proteolytic activation of σK . In parallel , SpoIIID and GerR , both acting as repressors , switch off two classes of genes initially activated by σE . Thus , the early transcription network in the mother cell includes a type-1 coherent FFL ( with SpoIIID as an activator ) , leading to the production of σK , and two type-1 incoherent FFLs ( with SpoIIID and GerR as negative regulators ) . Both SpoIIID and GerR are therefore critical to the early to late developmental transition characterized by the activation of σK and the inhibition of a large fraction of the σE regulon in the mother cell [9] . Once active , σK triggers a negative feedback loop that lowers production of σE and decreases the levels of SpoIIID [10 , 11 , 12] . Superimposed onto the transcriptional network are several cell-cell signaling pathways that operate at critical stages of morphogenesis , across the forespore membranes , to allow activation of the next sigma factor in the cascade , in the adjacent cell ( Fig . 1B ) . The requirement for σG for σK activity is an example of such a pathway . σF drives the initial transcription of the gene for σG in the forespore [13] , but the main period of σG activity , dependent on activation of an auto-regulatory loop , only begins after engulfment completion [14 , 15 , 16 , 17] . σG then controls production of a signaling protein , SpoIVB , which is secreted into the intermembrane space and activates the machinery responsible for the proteolytical activation of σK [18] . Another level of control of sigma factor activity is through the inhibitory action of anti-sigma factors . CsfB ( also called Gin , for inhibitor of sigma G ) is a Zn2+-containing anti-sigma factor that inhibits σG [19 , 20 , 21 , 22] . CsfB is part of a composite negative feedback loop that together with another anti-sigma factor , SpoIIAB , and the LonA protease , prevents activation of the auto-regulatory σG in pre-divisional cells ( Fig . 1C ) [20] . After the onset of sporulation CsfB is also produced in the forespore under σF control [23] , and has a role in delaying the onset of σG activity until engulfment completion [14 , 24] . Importantly , while σF and σG are highly similar proteins , σF itself is refractory to the action of CsfB . The basis for this selectivity can be traced to a single asparagine residue in a conserved β´-interacting surface of σG [20] . Likewise , a conserved glutamate residue of σF is important for resistance ( Fig . 1D ) . The anti-σG activity of CsfB , and the delayed transcription of the gene encoding σG relative to other σF-dependent genes are redundant mechanisms ensuring that the onset of σG activity begins at the appropriate time , possibly to coincide with engulfment completion . Accurate temporal control of σG activity is important , because it leads to the activation of σK [18] and premature activity of σK interferes with spore morphogenesis [25] . Timely activation of σK is important , as it ensures that the final stages in the assembly of the spore surface structures only initiate after the forespore is engulfed [18] . The σG-dependent production of SpoIVB , together with delayed transcription of genes required for pro-σK synthesis and processing , which require both σE and SpoIIID ( Fig . 1B; above ) , effectively couples pro-σKactivation to engulfment completion . In total , timely activation of the cell type-specific sigma factors enforces the directionality and fidelity of the morphogenetic process [7 , 18] . This works focuses on an additional role of the anti-sigma factor CsfB in controlling cell type-specific gene expression during spore development . We show that in addition to its previously characterized functions in the forespore and pre-divisional cells , the anti-sigma factor CsfB also plays a role in the mother cell where its synthesis is activated under the control of σK . We show that CsfB binds to and inhibits σE in vitro and in vivo , while σK is resistant to CsfB . A single residue in conserved region 2 . 3 of the sigma subunit is sufficient for the discrimination by CsfB and swapping this specificity interferes with the temporal progression of the mother cell line of gene expression . Thus , CsfB is part of a negative feedback loop that acts to limit the activity of σE following engulfment completion . Furthermore , we show that CsfB is also involved in preventing ectopic activity of σG in the mother cell . Therefore , CsfB is intricately connected to both cell-specific lines of gene expression to enforce the cell-type specific action of σG and the early to late developmental transition in the two cell types undergoing sporulation .
During sporulation , expression of csfB in the forespore is controlled by σF and occurs prior to engulfment completion [20 , 23] . Accordingly , sequences centered at about 26 bp ( GTATA ) and 48 bp ( GGGGAGGCTA ) upstream of the csfB start codon match the consensus for σF-controlled promoters [8] ( Fig . 2A ) . Presumably , the same σF-type promoter can also be recognized by σG in pre-divisional cells [8 , 20] . In addition , sequences matching the -10 ( CATATACT ) and -35 ( AACACCGA ) elements of the σK consensus binding sequence are present in the csfB regulatory region , upstream of the putative σF promoter [26] ( Fig . 2A ) . This suggested to us that expression of csfB could also take place in the mother cell , at a later stage in development , after σK is activated [25] . To test this possibility , we first examined expression of a functional csfB-gfp fusion inserted at the non-essential amyE locus of a strain deleted for the native csfB gene [20] . Cells were induced to sporulate by resuspension in a nutrient poor medium and examined by phase contrast and fluorescence microscopy at different times after resuspension ( which marks the beginning of sporulation under these culturing conditions ) . In a wild type background , expression of csfB-gfp was first detected in the forespore of sporulating cells ( sporangia ) , two hours after the onset of sporulation , consistent with the time of activation of σF ( Fig . 2B ) [20] . However , by hour 3 of sporulation CsfB-GFP fluorescence was also observed in the mother cell of sporangia that had completed the process of forespore engulfment , as indicated by the loss of FM4–64 staining at the forespore membranes ( Fig . 2B ) . At hour 4 of sporulation , 55% of sporangia exhibited a fluorescence signal restricted to the forespore , 30% fluoresced only in the mother cell , and 15% in both cell types ( Fig . 2B ) . Moreover , the appearance of a CsfB-GFP signal in the mother cell seemed to occur concomitantly with a decrease in the forespore-specific fluorescence signal ( Fig . 2B ) . By contrast , no CsfB-GFP fluorescence was seen in the mother cell of a sigK deletion mutant , even though these sporangia remained capable of completing engulfment ( Fig . 2B ) . These results suggest that σK is responsible for the mother cell-specific accumulation of CsfB-GFP . An alternative explanation , however , is that CsfB-GFP may be exported from the forespore to the mother cell upon engulfment completion , in an unindentified process that would require σK activity . To test this idea we replaced the native promoter sequences ( PcsfB ) driving expression of the csfB-gfp fusion by PspoIIQ , a well-characterized forespore-specific , σF-dependent promoter [22 , 27 , 28] . In this strain , accumulation of CsfB-GFP was detected in the forespore from hour 2 of sporulation onwards as expected , but was never observed in the mother cell even after σK had been activated ( Fig . 2B ) . Thus , mother cell-specific transcription , and not transport from the forespore , is responsible for CsfB-GFP accumulation in the mother cell at late stages of development . To more precisely delineate the respective contributions of the csfB promoters , we introduced point mutations in each of the σF- and σK-type putative -10 elements ( highlighted in red in Fig . 2A ) and examined production of CsfB-GFP during sporulation . In a strain bearing mutations in the putative σK-10 element , in which csfB-gfp expression is presumably only driven by the σF-dependent promoter ( PsigF in Fig . 2C ) , CsfB-GFP is restricted to the forespore . Conversely , when point mutations were introduced in the -10 region of the putative σF-dependent promoter ( PsigK in Fig . 2C ) , CsfB-GFP only accumulates in the mother cell . Immunoblot analysis of whole cell extracts of sporulating cells producing CsfB-GFP under the control of the wild type promoters revealed a steady increase in the accumulation of the fusion protein , between hour 2 and 5 of sporulation ( S1A Fig ) . By contrast , when expression of the fusion was solely dependent on PsigF , the intensity of the CsfB-GFP band strongly decreased after hour 3 , whereas when expression was driven exclusively from PsigK , CsfB-GFP accumulated only from hour 3 onwards ( S1A Fig ) . These observations are consistent with the presumed windows of activity of each of the csfB promoters ( Fig . 2A ) , as well as the fluorescence microscopy experiments described above ( Fig . 2B and C ) . We also carried out experiments with a lacZ transcriptional fusion and similarly detected two periods of csfB expression , the first beginning around hour 2 of sporulation and the second around hour 3–4 ( S1B Fig ) . The first period of β-galactosidase activity was not observed when expression of the fusion relied only on PsigK , whereas the second period was absent when expression of csfB was controlled by PsigF ( S1B Fig ) . These results are in line with the view that expression of csfB takes place at two developmental stages during sporulation , first in the forespore , under the control of σF , and later in the mother cell , under the control of σK . Finally , we conducted in vitro transcription reactions in which core RNA polymerase , purified from B . subtilis , was reconstituted either with each of the four cell type-specific sporulation sigma factors or with the main sigma subunit , σA . All σ factors were overproduced and purified from E . coli cells ( S2A Fig; see also S1 Text ) . While the σA or σE forms of RNA polymerase did not initiate transcription from the csfB promoter region , a run-off product of 148 nucleotides was obtained with the σF- and σG-containing holoenzymes ( Fig . 2D ) . The size of this transcript is consistent with the location of PsigF and in line with the idea that σFand σG utilize the same csfB promoter ( Fig . 2A ) . In addition , a product of 220 nucleotides was obtained with the σK-reconstituted holoenzyme ( Fig . 2D ) , in agreement with the location of the predicted PsigK ( Fig . 2A ) . In all , our findings suggest that , following engulfment completion , expression of csfB is switched off in the forespore by an unknown mechanism and activated in the mother cell , under the direct control of σK . In pre-divisional sporangia , σG activity is inhibited by a composite negative-feedback loop involving CsfB [20] . We have argued that the σF-promoter of csfB is recognized and used by σG to achieve this mechanism of negative auto-regulation ( Fig . 1C ) . Thus , inactivation of PsigF should mimic the previously described activity of σG in pre-divisional cells of a csfB deletion mutant [20] . In support of this hypothesis , we found that PsigK cells ( carrying mutations in the σFpromoter ) show activity of σG in pre-divisional cells , as measured by the expression of transcriptional fusions of lacZ and cfp to the σG-dependent sspE promoter ( S3 Fig; see also S1 Text ) . These results confirm that the σF-type promoter of csfB is the one used by σG in pre-divisional cells . Next , we used DNA microarrays to examine sporulation-specific gene expression in the PsigF strain , which does not express CsfB in the mother cell , in comparison to the wild type . The total RNA used for these studies was extracted and purified from cultures in resuspension medium at hour 3 of sporulation , i . e . , shortly after csfB is turned on in the mother cell ( above ) . We observed important changes in expression of several of the sporulation sigma factor regulons . Nevertheless , the expression of nearly all of the genes in the σFregulon remained unchanged in the PsigF strain versus the wild type ( Fig . 3A; S3 Table ) ; however , this does not include the group of forespore-expressed genes that are under the dual control of σFand σG ( see below , σG regulon ) . By contrast , a large fraction of the σE-dependent genes was upregulated in the PsigF strain relative to the wild type ( Fig . 3A; S3 Table ) . Importantly , the σE regulon can be sub-divided into 4 groups: i ) genes that rely on σE alone for expression , ii ) those that require SpoIIID as a positive factor , iii ) those that are repressed by SpoIIID , and iv ) , those that are repressed by GerR [9] . A closer inspection of the transcriptional profiling data revealed that the σE-dependent genes that are upregulated in the PsigF strain are the first and second groups , i . e . , those that rely exclusively on σE for expression and those that are activated by SpoIIID ( Fig . 3A; S3 Table ) . Conversely , essentially no upregulated genes are seen among the σE-dependent genes that are repressed by GerR or SpoIIID ( Fig . 3A; S3 Table ) . These genes correspond to pulses X2 and X3 in Fig . 1B [9] . It is likely that transcription of these genes has already been turned off by the time CsfB starts to accumulate in the mother cell , explaining the lack of effect of the mutation . In contrast , the increased expression of the σE- and σE/SpoIIID-dependent genes suggests that in the absence of CsfB , the activity of σE in the mother cell is increased or prolonged . Several of the σE/SpoIIID-dependent genes are involved in production and activation of σK[9 , 29] . The σK regulon itself can be sub-divided into genes whose expression is activated by σK alone , those that require GerE for full expression ( pulse X7 in Fig . 1B ) and those repressed by GerE ( pulse X6 ) [9] . Transcriptional profiling of the PsigF strain shows that a significant fraction of the σK-dependent genes is upregulated , regardless of GerE dependency ( Fig . 3A; S3 Table ) . We presume that the increased expression of the σK regulon may result from augmented activity of σK ( but see also below , section on the ectopic activation of the σG regulon ) . In all , the genome-wide transcriptional profiling data highlight the importance of csfB for modulating the levels of σE- and σK-dependent gene expression in the mother cell , and are consistent with the interpretation that perturbations affecting the mother cell line of gene expression in a csfB mutant are mainly caused by an increased and protracted activity of σE . A somewhat unexpected result of the global transcriptional profiling analysis was that the mutation in the σK-dependent promoter of csfB also caused a generalized increase in the expression of the σG regulon ( Fig . 3A; S3 Table ) , including genes that are repressed ( pulse X4 in Fig . 1B ) or activated by SpoVT ( pulse X5 ) . We considered two possibilities . First that the increase in σE activity in the PsigF strain somehow led to an increase in the activity of σG in the forespore . Second , that eliminating expression of csfB in the mother cell alleviated some of the restrictions imposed on σG to become active in this cell type . To test these hypotheses , we first examined the localization of the fluorescence signal from a PsspE-CFP fusion in cells of the PsigF-csfB strain , in comparison with the wild type , at hour 3 of sporulation in resuspension medium . In the wild type strain , and as expected for a σG-controlled gene , the fluorescence signal was confined to engulfed forespores ( S4A Fig , yellow arrows ) . Fluorescence from PsspE-cfp was also detected in engulfed forespores for the PsigF strain ( S4A Fig , yellow arrows ) . However , for the PsigF mutant , a week fluorescence signal was also detected in the mother cell prior to engulfment completion ( S4A Fig , white arrows ) . The quantitative analysis of the signal shows that the mother cell-associated fluorescence in PsigF sporangia that have not completed the engulfment sequence is consistently higher than for the wild type ( S4B Fig; see also S1 Text ) . In an attempt to verify this observation , we turned to fluorescence in situ hybridization ( FISH ) using a specific antisense DNA probe labelled with Cy3 , in an attempt to localize the sspE mRNA [30] . Because at this time in sporulation the forespore is not yet recognizable by phase contrast microscopy , we relied mainly on the size of the fluorescence signal to distinguish mother cell or sporangia ( larger ) from forespore ( smaller ) localization . The FISH images suggest that the signal is confined to the forespore in the wild type strain , whereas the signal is mostly associated with whole cells ( sporangia ) in the PsigF strain ( Fig . 3B ) . In cells of the mutant strain , the size of the fluorescence signal is 1 . 5 to 2 times longer than that observed for the wild type strain ( Fig . 3B ) . Moreover , the size of the fluorescence signal for the PsigF strain coincides with the size of the sporangia as judged from the phase contrast images ( Fig . 3B ) . This suggests whole cell production of the sspE mRNA . Some cells of the PsigF mutant do not show a forespore-associated signal ( Fig . 3B , insert ) . This suggests that expression of sspE may initiate earlier in the mother cell than in the forespore , in line with the detection of PsspE-CFP in the mother cell , for the PsigF strain , prior to engulfment completion ( S4 Fig; see also above ) . In any event , the results are consistent with the hypothesis that the sspE mRNA is also produced in the mother cell . Therefore , these data suggest that eliminating transcription of csfB from the PsigK promoter increases transcription of σG-dependent genes in the mother cell and thus that CsfB contributes to the negative regulation of σG in this cell . Presumably , CsfB is part of a composite negative feedback loop limiting σG activity in the mother cell similar to the mechanism that prevents activation of σG in pre-divisional cells . In these cells , CsfB plays a key role in inhibiting σG activity , along with the SpoIIAB anti-sigma factor and the LonA protease [20 , 31 , 32] . Previous work has shown that when the sigG gene is placed under the control of a σE-controlled promoter , SpoIIAB and LonA are also important for the inhibition of σG activity in the mother cell [15 , 20 , 32] . In an extension of this work , we now show that SpoIIAB plays the leading role in σG inhibition in the mother cell , while LonA and CsfB appear to have largely redundant supportive contributions ( S5A Fig ) . Nevertheless , these results demonstrate the ability of CsfB to act as an inhibitor of σG activity in the mother cell , in line with the observed σG activity in the PsigF strain ( Fig . 3 and S4 Fig ) . Importantly , during the course of these experiments aiming at directly testing for a role of CsfB in the inhibition of σG in the mother cell ( above ) , we noticed that the forced production of σG in the mother cell resulted in premature activity of σK ( S5B Fig ) . Activation of σK requires pro-σK processing , which in turn depends on the activity of σG in the forespore , following engulfment completion [25 , 33] . Since σG shows some activity prior to engulfment completion in the PsigF strain ( above ) , we therefore considered the possibility that the augmented transcription of the σK-controlled genes , as detected in our microarray analysis ( above ) , could result in part from premature processing of pro-σK independently of the σG-dependent pathway that operates following engulfment completion ( S5C Fig ) . To test this idea , we surveyed the accumulation of pro-σK and σK by immunoblotting , throughout sporulation . Pro-σK was first detected in both the wild type strain and in the PsigF strain 140 min after resuspension in sporulation medium ( Fig . 3C ) . However , mature σK is first detected 180 min after resuspension for the PsigF strain , and only at minute 200 for the wild type ( Fig . 3C ) . Moreover , even at minute 200 , the fraction of mature σK is higher for the PsigF strain ( Fig . 3C ) . Therefore , the lack of csfB expression in the mother cell , can lead to premature activation of pro-σK , again underscoring the need for strict inhibition of σG activity in the mother cell . The analysis of the global transcriptional profiling data raised the possibility that CsfB could act as an inhibitor of σEand/or σK . This prompted us to determine whether the anti-sigma factor could form complexes with either sigma factor . To isolate CsfB-GFP and interacting proteins from sporulating cultures of a strain expressing csfB-gfp , we used a GFP-binding protein ( GBP ) coupled to a chromatographic matrix ( GFP-Trap beads ) . As controls , we examined a wild type strain carrying no gfp fusion and a strain producing GFP under the control of the xylose-inducible PxylA promoter . The extracts were prepared 4 hours after the onset of sporulation , when CsfB-GFP is known to accumulate in the mother cell ( Fig . 2B; above ) . Control experiments also confirmed the accumulation of σE , σG , σK or GFP in the whole-cell extracts prepared from the various strains ( Fig . 4A ) . The extracts from all strains were incubated with GFP-Trap beads , the bound proteins eluted and identified by immunoblot analysis with antibodies raised against σE , σG , σK or GFP . We found σE but not σK , to be pulled down efficiently from extracts of the strain producing CsfB-GFP . By contrast , σE was not recovered from extracts of the strain containing no GFP fusion or from the strain that produced unfused GFP ( Fig . 4A ) . Thus , retention of σE by the GFP-trap beads depended on formation of a complex with CsfB-GFP . As expected , CsfB-GFP was able to pull down σG [20] , a property used here as a positive control for the experiment ( Fig . 4A ) . Interestingly , a role for CsfB in inhibiting σE activity in the forespore soon after asymmetric division was previously suggested [34] . Under the conditions of our experiments , however , σE is only expected to accumulate in the mother cell and the expression of csfB has switched to the mother cell in most sporangia in the population when the cells were harvested for the pull down assays ( see above ) . Therefore , the result reported in Fig . 4A most likely reflects an interaction occurring between σE and CsfB in the mother cell and not in the forespore . In agreement with this interpretation , CsfB-GFP produced solely from the PsigK promoter , was still able to pull down σE ( Fig . 4A ) . In addition , this construct could also pull down σG and conversely CsfB-GFP produced from the PsigF was able to pull down σE ( Fig . 4A ) . Even though these observations could be explained by complex formation after the cells are disrupted in particular for the PsigF strain ( provided that there was enough free CsfB-GFP ) , we note that more σE and less σG seemed to be pulled-down in the strain where csfB-gfp expression is restricted to the mother cell than in the strain producing CsfB-GFP in both compartments ( Fig . 4A ) . Thus , it is possible that the amounts of σE and σG pulled-down from the PsigK-csfB-gfp carrying strains reflect the levels of both of these proteins in the mother cell . Considering that σG can accumulate and become active in the mother cell under certain conditions [15 , 20]; S4 Fig and S5 Fig; see also above ) , some σG might be present in the mother cells at hour 4 of sporulation . One of the consequences of inactivating the PsigK promoter of csfB was the protracted expression of a specific class of σE-dependent genes . A prediction that stems from these results and from the pull-down experiments described above is that CsfB binds directly to σE . To test this hypothesis , we purified His6-CsfB , σE and σK ( both lacking their pro-sequences ) overproduced in E . coli ( Fig . 4B , first three lanes and S2A Fig ) ( Note that in addition to σE , which has a molecular weigth of 29 kDa , a stable proteolytic fragment of about 20 kDa accumulated in the preparations; Fig . 4B ) . We then asked whether His6-CsfB immobilized onto a Ni2+-NTA column could capture untagged , purified σE or σK . While His6-CsfB bound efficiently to the Ni2+-NTA column , neither σE nor σK did ( Fig . 4B ) . However , in the presence of His6-CsfB , σE ( but not σK ) , was efficiently retained by the column ( Fig . 4B ) , a result fully consistent with the pull-down experiments described above ( note that both σE and the 20 kDa fragment were retained; Fig . 4B , left lanes ) . To confirm these results with a different technique , we also carried out a GAL4-based yeast two-hybrid assay , an approach that had been used before to dissect the interaction between σG and CsfB [20] . We constructed translational fusions of σF ( as a negative control ) , σG ( as a positive control ) , σE , σK , σA or CsfB to the C-terminus of the GAL4 DNA binding ( BD ) or activation domains ( AD ) . The various fusion proteins were expressed in different combinations in yeast cells and checked for their ability to interact in vivo , as assessed by the expression of a lacZ gene preceded by a GAL4-responsive element . As expected , we detected an interaction of CsfB with σG , but not with σF ( [20]; Fig . 4C ) . In addition , CsfB interacted with σE ( Fig . 4C ) , suggesting that the two proteins could indeed establish specific contacts . No interaction was detected between CsfB and σK or σA ( Fig . 4C ) . Thus , both the yeast two-hybrid and the affinity chromatography assays indicate that CsfB and σE directly interact . While CsfB binds to both σG [20] and σE ( [34]; see above ) , no study has shown direct inhibition of transcriptional activity by the anti-sigma factor . To test the ability of purified CsfB to inhibit σG- or σE-directed transcription , core RNA polymerase ( E ) was purified from B . subtilis , and reconstituted with σF , σE , σG , σK , or σA overproduced and purified from E . coli cells ( S2A Fig ) . As templates for in vitro transcription reactions , we used PCR-generated DNA fragments corresponding to promoters of genes known to be under the control of the sigma factors tested , and whose transcriptional start site has been mapped . As such , the gcaD gene was used as the template for σA-directed transcription [35 , 36] , spoIIQ as a template for EσF [27 , 37] , spoIID , spoIIM , spoIIIA p1 and spoIIIA p2 as templates for EσE [38 , 39 , 40] , sspB for EσG [30] , and gerE for EσK [33] ( Fig . 5A and S2B Fig ) . All forms of RNA polymerase tested directed the production of run-off transcripts of expected sizes ( Fig . 5B-D and S2C Fig ) . No specific transcription products were seen when templates were mixed with core RNA polymerase in the absence of a σ subunit ( S2C Fig ) . CsfB did not inhibit transcription by EσA , EσF ( Fig . 5B ) or EσK ( Fig . 5C ) , but inhibited the σE-directed utilization of the spoIIM ( Fig . 5B ) , spoIID ( Fig . 5C and D ) , spoIIIA p1 and spoIIIA p2 promoters ( S2C Fig ) . CsfB also inhibited the σG-directed transcription of sspB ( Fig . 5B and D ) . Inhibition of EσA , EσFor EσK by CsfB was not observed even at molar ratios higher than those that inhibited EσE ( Fig . 5C and D ) . Interestingly , inhibition of EσE , and to some extent of EσG , required molar ratios higher than 1 ( Fig . 5B and D; S2C Fig ) . One possibility is that active CsfB is a dimer ( or a higher order multimeric form ) , in agreement with the results of a previous study [21] . In total , the run-off transcription assays show that CsfB inhibits transcription by the RNA polymerases that contain the σ subunit to which it can bind in vitro , i . e . , σE or σG . To map the region ( s ) involved in the CsfB-σE interaction , we performed GST-pull down and yeast two-hybrid assays , two techniques that were used before to map the interaction between CsfB and region 2 . 1 of σG [20] . For that purpose , full-length mature σE or different fragments of the sigma factor were fused to either GST or to GAL4 ( Fig . 6A ) . For the GST pull-down assays , we used a CsfB-StrepII-tagged protein [20] and a GST-σE fusion protein , lacking its pro region ( sigma fragment a , Fig . 6A ) or GST fusions to different fragments of σE ( fragments b to h; Fig . 6A ) , overproduced and partially purified from E . coli . We incubated CsfB-StrepII-tag with GST or the GST-σE fusion proteins immobilized on glutathione beads . After washing , the presence of CsfB in the elution samples was assessed by immunoblotting with an anti-StrepII tag antibody . The CsfB protein was retained by GST-σE ( fragment a ) but not by GST alone ( Fig . 6B , left panel ) . GST fusions to σE fragments corresponding to regions 2 . 1 through 2 . 4 ( fragment c , GST-σE 30–141 ) , 2 . 1 through 3 . 1 ( fragment d , GST-σE 30–172 ) , and 2 . 2/2 . 3 ( fragment e , GST-σE 79–117 ) pulled down CsfB , whereas fragments encompassing regions 2 . 1/2 . 2 ( sigma fragment b; GST-σE 30–98 ) , 2 . 3/2 . 4 ( fragment f , GST-σE 98–141 ) , 3 . 1 ( fragment g , GST-σE 141–172 ) , and 4 . 1/4 . 2 ( fragment h , GST-σE 172–239 ) , did not ( Fig . 6B ) . With the exception of fragment e ( 2 . 2/2 . 3 , σE 79–117 ) , the same fragments also showed an interaction with CsfB in yeast two-hybrid assays ( Fig . 6C ) . The lack of interaction of fragment e in the yeast two-hybrid assay may be due to the topology and/or stability of the GAL4 fusion . Region 2 . 1 was previously implicated in the interaction of CsfB with σG [20] . However , this region does not appear to be sufficient to mediate the interaction of CsfB with σE . In contrast , because σE 79–117 ( regions 2 . 2/2 . 3 ) interacted with CsfB , but neither σE 30–98 ( regions 2 . 1/2 . 2 ) nor σE 98–141 ( regions 2 . 3/2 . 4 ) did , it is likely that residues at the end of region 2 . 2 and/or at the beginning of region 2 . 3 of σE are necessary for the interaction . In any event , our results show that CsfB and σE interact directly , and that the region required for the interaction differs from that found in σG [20 , 22] . The N45E substitution in σG results in loss of CsfB binding , whereas the E39N substitution in σF is sufficient for conferring CsfB binding ability [20] . However , the residue homologous to N45 in σE corresponds to a glutamate , E64 ( Fig . 7A ) . Together with the mapping experiments described in the preceding section , this observation suggests that the interaction of CsfB with σE differs from that with σG . To more precisely delineate the determinants for CsfB binding to σE , we sought to identify residues that are necessary for the interaction . An inspection of the amino acid sequence of σE in regions 2 . 2 and 2 . 3 revealed an asparagine residue located at position 100 ( i . e . , at the beginning of region 2 . 3 ) and conserved among all of Bacillus σE orthologues . Strikingly , the position homologous to N100 in σK is occupied by an aspartate ( E93 ) in B . subtilis and invariably ( with the exception of the σE proteins ) by an acidic residue in all other Bacillus sigma factors ( Fig . 7A ) . N100 in σE is located in the vicinity of several residues involved in promoter melting ( Fig . 7A ) . By contrast , E64 in σE and N45 in σG are located at the beginning of a helix within which several contacts are established with the β´ subunit of RNA polymerase ( S6 Fig ) . Thus , depending on which sigma factor is present in the complex , CsfB appears to bind to two distinct functional surfaces of the RNA polymerase holoenzyme . We wanted to test whether the CsfB anti-sigma factor discriminated between σE and σK by differentiating an asparagine from an acidic residue at the beginning of region 2 . 3 . We constructed GST and GAL4 fusions to forms of σE bearing the single amino acid substitution N100E and used them in pull-down and yeast two-hybrid experiments . We found that GST-σE N100E did not pull-down CsfB-strepII partially purified from E . coli cells ( Fig . 7B ) and showed a much-reduced interaction ( when compared to the wild type sigma factor ) with the anti-sigma factor in a yeast two-hybrid assay ( Fig . 7C ) . By contrast , a GAL4 fusion to σK bearing an E93N substitution interacted strongly with the anti-sigma factor in the yeast two-hybrid assay ( Fig . 7C ) . To determine whether the N100E and E93N substitutions in σE or σK affected their activity in vivo , the corresponding mutations were first transferred to the sigE and spoIVCB genes ( spoIVCB codes for the N-terminal half of σK; see the Material and Methods section for details ) . We then used PspoIVCA- ( σE-dependent promoter ) and PcotG-lacZ ( σK-dependent promoter ) transcriptional fusions to monitor the activity of σE and σK respectively , during sporulation . Consistent with loss of CsfB regulation , the N100E substitution increased expression of spoIVCA-lacZ , whereas the E93N replacement reduced expression of cotG-lacZ ( Fig . 7D ) . Expression of PspoIVCA-lacZ or PcotG-lacZ in a ΔcsfB mutant did not differ much from the wild type ( Fig . 7D ) , presumably because premature activation of pro-σK somehow compensates for increased activity of σE ( see also above ) . In any event , the N100E substitution is sufficient to render σE refractory to CsfB , whereas the E93N substitution suffices to make σK susceptible to CsfB . Assembly of a protective protein shell called the coat that forms the spore surface , involves the timely production of over 80 components under the control of σE or σK [41 , 42] . Since the N100E substitution increases the activity of σE and reduces that of σK we wanted to examine the coat protein composition in spores of the strain producing σE N100E . Proteins were extracted from purified spores by treatment with a buffer containing SDS and the reducing agent DTT , or by NaOH [43] . The first treatment releases about 80 proteins from wild type spores ( S7A Fig , top ) whereas NaoH extraction produces a much smaller collection of extractable proteins ( S7A Fig , bottom ) . We found that several proteins were more extractbale from N100E spores than from wild type spores , among which the species labelled a-e ( SDS/DTT extraction ) and f-k ( NaOH extraction ) in S7A Fig ( bottom panel ) . For example , the outer coat proteins CotA ( in band a ) , CotB ( in b ) , and Tgl ( in d ) , as well as the inner coat proteins CotN ( in bands h and i ) , YaaH ( in c ) and YybI ( in band d ) were more extractable from σE N100E spores ( S7A Fig , see also S1 Text ) . Synthesis of YaaH , CotN and YybI is under σE control , whereas production of CotA , CotB , and Tgl is mainly controlled by σK [9 , 29 , 41 , 44 , 45 , 46] . Thus , the N100E substitution has a strong and global impact on the assembly of the spore surface layers , affecting the assembly of proteins from both the inner and outer coat layers , which are produced at different periods ( controlled by σE or by σK ) during spore coat assembly .
Genome-wide transcriptional profiling analysis showed that inactivation of PsigK , which abolishes CsfB accumulation in the mother cell , caused increased transcription of σE-controlled genes that either rely solely on σE for expression or are dependent on the type I coherent FFL formed by σE and the ancillary transcriptional activator SpoIIID [9] . Increased σE activity was also observed using a σE-responsive lacZ reporter in a strain expressing the CsfB-resistant form of σE ( σE N100E ) . By contrast , σE-dependent genes repressed by GerR or SpoIIID ( i . e . , the output of the type I incoherent FFLs generating pulses X2 and X3 in Fig . 1B ) did not show increased expression in the PsigF-csfB strain , most likely because their transcription has already been switched off at the time of analysis due to the action of the two repressors ( Fig . 3A ) . In general , these observations are consistent with an increase in the activity of σE in the absence of CsfB and with the properties of the FFLs formed by σE [3] . They also support a model in which the appearance of CsfB in the mother cell promotes the transition from early to late cell-type specific gene expression ( i . e . , the σE to σK switch ) ( Fig . 8A ) . While a previously described negative feedback loop driven by σK reduces the levels of σE [11 , 12 , 48] , the feedback loop revealed in this work , and similarly initiated by σK , uses CsfB to limit the activity of σE in the mother cell . Thus two partially redundant feedback loops evidently function to promote proper switching from σE- to σK-dependent transcription following engulfment completion , and thus the developmental transition from early to late stages of morphogenesis ( Fig . 8A ) . Interference with csfB expression in the mother cell , as when PsigK is inactivated , also leads to increased expression of several genes under σK control ( Fig . 3 ) . Because CsfB does not bind to σK and does not inhibit σK-directed transcription in vitro , it is unlikely that the increased activity of σK is due to the loss of a direct interaction with CsfB . Instead , it is more likely to be a consequence of the ectopic activation of σG in the mother cell . Previous work has shown that two negative regulators of σG , the anti-sigma factor SpoIIAB and the LonA protease , effectively counteract σG in the mother cell , when its synthesis is artificially induced by driving expression of sigG from a σE-dependent promoter [15] . Under the same genetic conditions , we showed that CsfB contributes , along with LonA and SpoIIAB , to the inhibition of σG activity in the mother cell ( S5 Fig ) . In the PsigF-csfB mutant , our transcriptional profiling data indicate that most of the genes in the σG regulon show increased expression . This increased expression is due in part to increased activity of σG in the mother cell of PsigF sporangia . Not only transcription from the σG-dependent PsspE promoter fused to cfp is detected in the mother cell prior to engulfment completion ( S4 Fig ) , but FISH experiments also show accumulation of the sspE transcript in the mother cell ( Fig . 3 ) . Therefore , CsfB is one of several redundant mechanisms that act to silence σG in the mother cell . In wild type cells , it is the activity of σG in the forespore following engulfment completion that triggers σK activation in the mother cell , via the signalling protein SpoIVB . Mutations that bypass this signalling pathway result in premature activity of σK and defects in spore morphogenesis [25] . We have shown that the production of σG in the mother cell uncouples pro-σK processing from engulfment completion and leads to premature activation of σK because SpoIVB can activate processing of pro-σK directly in the cell where it is produced ( S5 Fig ) . Accordingly , processing of pro-σK in the PsigF mutant is detected earlier than in the wild type ( Fig . 3C ) . By contrast , in the σE N100E strain , expression of a σK-dependent lacZ fusion is reduced and delayed , presumably because pro-σK processing remains strictly dependent on the post-engulfment , σG-dependent SpoIVB signalling from the forespore . Therefore , the silencing of σG in the mother cell by CsfB , along with the contributions of SpoIIAB and LonA , prevents premature activity of σK . We have previously shown that N45 , an asparagine residue in region 2 . 2 of σG , is a critical determinant for CsfB binding [20] ( Fig . 1C ) . Not only is this residue conserved among Bacillus orthologues of σG but a glutamic acid residue , E39 , is found at the homologous position of the CsfB-resistant σF protein in B . subtilis , and an acidic residue is invariably found at the equivalent position in σF orthologues of other Bacillus species . Importantly , the N45E substitution renders σG refractory to CsfB binding and conversely , the E39N variant of σF is susceptible to the anti-sigma factor . Residue N45 of σG is most likely involved in a direct contact with the β´ subunit of RNA polymerase , suggesting that CsfB , like other anti-sigma factors interferes with the σ/β ´ interaction [20] . Our in vitro transcription assays support this model , as CsfB is sufficient to inhibit σG- and σF-directed transcription at bona fide promoters in vitro ( Fig . 5 ) . Our results now implicate a region encompassing regions 2 . 2 and 2 . 3 of σE in CsfB binding ( Fig . 6 ) , and an asparagine residue , N100 , at the beginning of region 2 . 3 , was found to be a key determinant ( Fig . 7A-C ) . In a striking parallel with the σF/σG pair , a homologous asparagine is found in all known orthologues of σE from related organisms ( Fig . 7A ) . Despite the high degree of similarity between the σE and σK proteins across sporeformers [4 , 5 , 6 , 7] , the homologous residue in the σK protein of B . subtilis is a glutamic acid , E93 , and an acidic residue is invariably found among Bacillus orthologues of σK ( Fig . 7A ) . Also reminiscent of the σF /σG pair , the N100E substitution makes σE refractory to CsfB , whereas the E73N variant of σK becomes susceptible to the anti-sigma factor ( Fig . 7B to D ) . Thus , binding of CsfB to any of the four sigma factors of sporulation is favored by the presence of a conserved asparagine in region 2 . 1 ( for σF/σG ) or 2 . 3 ( σE/σK ) and hindered by a glutamic acid at either position ( Fig . 8B ) . Resistance to CsfB binding requires both positions to be occupied by acidic residues . This discrimination is essential for the control of gene expression during sporulation: it allows σG activity to be inhibited in the forespore prior to engulfment completion , while allowing σF-dependent transcription; it also enables CsfB to antagonize σE and σG in the mother cell thus enforcing the cell type-specificity of σG and facilitating the timely switching from σE to the CsfB-immune σK . Making σF or σK susceptible to CsfB or making σE or σG resistant to the anti-sigma factor interferes with proper temporal control and compartmentalization of the forespore and mother cell lines of gene expression . The strict conservation of the discriminating residues in regions 2 . 1 or 2 . 3 of the sigma factors among Bacillus species ( Fig . 7A ) underscores the importance of CsfB for the activity of the cell type-specific sigma factors in sporeforming organisms closely related to B . subtilis . The crystal structure of the σ70-containing RNA polymerase holoenzyme from Thermus aquaticus shows that residue E189 , the homologue of N45 in σG , is involved in a direct contact with K159 in the β´ subunit [49 , 50 , 51 , 52] . We have argued that an asparagine residue could also contribute to the interaction with this site of β´ , consistent with the view that one mechanism by which anti-sigma factors function is by occluding sigma-core binding interfaces [53 , 54] . Occluding a β´-binding surface is consistent with a role for CsfB in preventing the ectopic activity of σG in pre-divisional cells [20] or the premature activation of σG in the forespore [14] . In contrast , N100 in σE is located in the beginning of region 2 . 3 , just upstream of a motif containing several conserved residues involved in promoter melting [55] ( Fig . 7A and S6 Fig ) . This suggests that CsfB may interfere with σE function at a step during transcription initiation subsequent to closed complex formation . In contrast to σG , whose ectopic or premature activation has to be prevented , σE is already engaged in transcription when CsfB appears in the mother cell . Therefore , the most effective mechanism for antagonizing σE activity may not be targeting a core-binding surface but rather a functional region that prevents the activity of the σE-containing RNA polymerase holoenzyme . We note , however , that N100 is also close to several residues in region 2 . 2 that have been implicated in core binding in E . coli σ70 , and are conserved in B . subtilis σA ( highlighted in grey in Fig . 7A ) ( reviewed by [55] ) . Thus , binding of CsfB in the vicinity of N100 could also possibly occlude core-binding sites in region 2 . 2 . The function of CsfB affects both the transcriptional and the cell-cell signalling networks that control spore differentiation ( Fig . 8A ) . In the forespore , CsfB contributes to the inhibition of σG during early stages of development [14] . The σK-dependent expression of csfB in the mother cell limits the activity of σE and prevents the ectopic activation of σG These activities of CsfB are each one part of redundant mechanisms that work to the same end . σK activity blocks expression of sigE by an unknown mechanism [11 , 12] , and ectopic activity of σG is limited by SpoIIAB and LonA in the mother cell . Our discovery that CsfB binds and inhibits the activity of σE also leads us to propose that in the forespore , CsfB binding to σE explains how CsfB functions as one of the several partially redundant mechanisms described by Piggot and co-workers [34] that prevent ectopic activity of σE in the forespore ( see also Fig . 1B ) . That several of the major roles for CsfB ( i . e . , preventing ectopic expression of σG and σE , and facilitating the transition between σE and σK ) involve partially redundant mechanisms speaks to the importance of controlling these processes , and provides an explanation for why CsfB is highly conserved among sporeformers [6 , 21] . Moreover , these redundancies may also explain why the phenotype of a csfB mutant is relatively mild . The original work on CsfB found no decernable phenotype [23] , whereas subsequent work from Stragier’s group [22] described a small but significant germination defect , but only when a ΔcsfB mutation was combined with an allele causing premature transcription of the gene for σG [22] . The PsigF and PsigK strains formed spores with the same efficiency ( 77% and 88% , respectively ) as the wild type strain ( 83% ) . However , we did observe a difference in the protein composition of the spore surface layers between σE N100E and wild type spores ( S7A Fig ) . It is unknown whether this phenotype provides some of the selective pressure for maintaining CsfB during the evolution of sporeformers , but we note the important role of the spore surface layers in mediating many of the environmental interactions of spores , including with cells of host organisms [42] . On the other hand , the evolution of redundant mechanisms to control key steps in development may have promoted the stabilization or canalization of the cell type-specific patterns of gene expression that led to the establishment of the endospore differentiation pathway [56] . The role of CsfB in favouring the switch from early to late stages in spore development is likely to be conserved among Bacillus species and other sporeformers . Importantly , at least in Bacillus species , the actions of the CsfB anti-sigma rely strongly on its ability to discriminate between the highly similar σF/σG and σE/σK pairs ( Fig . 8B ) . Interestingly , CsfB is not found in some Clostridia , a more distantly related class of sporeforming organisms , including the human intestinal pathogen C . difficile . The gene regulatory network for sporulation has been recently characterized in detail for C . difficile . It is interesting to note that in this organism , σG is active prior to engulfment completion and that σE remains active until the late stages of sporulation [57 , 58 , 59] . The looser temporal control of σ factor function in C . difficile and presumably also in other Clostridia may stem , at least in part , from the absence of a CsfB orthologue . Moreover , it may be interesting to investigate whether this looser regulation results in greater heterogeneity of morphology among spores of certain Clostridial species than among spores of Bacillus species .
Except for strain MBS3656 ( [60]; see below ) , all other B . subtilis strains used in this work are congenic derivatives of the Spo+ strains MB24 ( trpC2 metC3 ) or PY79 ( prototrophic ) . Their construction is detailed in S1 Text , and they are listed , with their genotypes , in S1 Table . All plasmids used in this work are described in S1 Text . Primers used for plasmid construction , mutagenesis or sequencing are listed , with their sequences , in S2 Table . LB medium was used for routine growth or maintenance of E . coli and B . subtilis , and sporulation was induced by growth and exhaustion in Difco sporulation medium ( DSM ) or by the re-suspension method [43] . The strains for sigma factor and CsfB expression were cultured in LB at 37°C with Ampicillin ( 100μg/ml ) . Arabinose was added to final concentration of 0 . 5% at an OD600 of 0 . 6 . The culture was allowed to incubate for another 1–2 hours before cells were harvested and stored at -80°C . Sigma factors and CsfB were purified using Qiagen Ni-NTA . The Ni-NTA-affinity purified protein was analyzed on a 10% SDS-PAGE . Fractions containing the sigma factor or CsfB were pooled and dialyzed against dialysis buffer ( 50mM Tris , 100mM NaCl , 3mM β-mercapatoenthanol ) . The dialyzed protein was chromatographed through a GE HiLoad 16/60 Superdex 75 column , fractions ( 1 ml ) collected , and those containing the desired protein pooled and dialyzed against 1X in vitro transcription reaction buffer ( 50mM Tris , 100mM KCl , 10% Glycerol , 10mM DTT ) . Protein concentration was measured using a Bradford assay ( Bio-Rad , Hercules , CA ) . Protein aliquots were stored at -20°C . RNA polymerase ( RNAP ) was purified from B . subtilis strain MH5636 [60] cultured in LB with chloramphenicol ( 5 μg/ml ) to an OD600 of 1 . 0 . Cells was harvested by centrifugation and lysed by treatment with lysozyme ( 5mg/ml ) for 30 minutes at 4°C followed by passage twice through a French pressure cell at a pressure of 20 , 000 psi . Core RNA polymerase was purified essentially as described by Burgess and colleagues [55] . Briefly the cell lysate supernatant was first purified using Qiagen Ni-NTA affinity column . Fractions containing RNAP were pooled and subjected to chromatography on a HiLoad 16/60 Superdex 200 gel filtration column . Fractions containing RNA polymerase were then purified by ion-exchange exchange chromatography on a GE Mono Q 5/50 GL column . The purity of RNAP was assayed on a 4–20% SDS-PAGE . The concentration of purified RNAP was determined using a Bradfors assay ( Bio-Rad , Hercules , CA ) . RNAP aliquots were at -20°C . Core RNAP ( 13 nM ) was incubated in transcription reaction buffer ( 40 mM Tris/HCl ( pH 8 ) , 50 mM KCl , 10 mM MgCl2 , 10mM DTT , 50 mg/ml acetylated BSA , 0 . 5 ml RNase-inhibitor [61] on ice for 30 minutes with 130 nM sigma factor and 1 . 0 μg of a DNA template purified through a CsCl/ethidium bromide density gradient and cleaved with a restriction enzyme ( BamHI or HindIII ) . After 10 min at 37°C , ATP , CTP , and GTP were added to a final concentration of 1 . 0 mM and 5 μCi UTP was added to 50 μl transcription reactions . The mixture was incubated for 10 minutes at 377°C after which 0 . 2 mM unlabeled UTP was added , followed by a further incubation for 10 minutes at 37°C . Finally , the reaction mixture was extracted with phenol-chloroform and the nucleic acids precipitated and electrophoretically resolved through 10% ( w/v ) polyacrylamide gels containing 7M urea . In the reactions containing CsfB , the anti-sigma factor was added at an amount equal to sigma ( 130 nM ) or 2 , 3 , 4 or 8-fold excess ( indicated as 1x to 8x in Fig . 6C and D ) , and incubated with sigma for 20 min at room temperature . RNAP was added and the reactions conducted as described above , and incubated with sigma for 20 min at room temperature . RNAP was added and the reactions conducted as described above . These analyses were carried out as previously described in Cozy et al . [62] and Winkelman et al . [63] . Briely , the arrays obtained form Agilent include 15 , 744 probes covering the annotated protein-coding genes of B . subtilis [64] and the small non-coding RNAs reported in Rasmussen et al . [65] and Irnov et al [66] . The arrays were designed using the Agilent eArray application . Strains AH6825 ( PcsfB-csfB ) and AH6827 ( PsigF-csfB ) were sporulated by resuspension in Sterlini-Mandelstam medium and samples collected at hour 3 of sporulation . Cells pellets were recovered by centrifugation after mixing with an equal volume of cold methanol . Total RNA was recovered using a hot acid-phenol protocol followed by clean-up using the Qiagen RNeasy kit . cDNA was synthesized from the purified total RNA and labelled using the Agilent Fairplay III kit . After hybridization and washes using standard protocols , the arrays were scanned in an Agilent Technologies DNA microarray scanner with Surescan high-resolution technology . Processed signal from the Agilent software was subjected to standard lowess normalization using Bioconductor run in R and the geometric mean of the probes was used to give the final value for each gene and nc-RNA . For RNA-FISH , we used the following protocol: cells growing in Re-suspension medium were fixed in Histochoice solution ( Ameresco ) for 15 min at room temperature and 30 min on ice . The samples were centrifuged three times at 3 . 000xg for 2 min and washed in 1xDEPC-treated PBS . The cell pellets were resuspended in 100 μl GTE buffer ( 50 mM glucose , 20 mM Tris-HCl pH 7 . 5 , 10 mM EDTA pH 8 ) . Sixteen microlitres of a 10 mg/ml lysozyme solution ( GTE , 4 mM vanadyl ribonucleoside complex ( VRC ) , 10 mg/ml lysozyme ) were added to 48 μl of cell suspension . The mixture was immediately placed onto poly-L-lysine-coated multi-well slides , and incubated for 10 min at room temperature . The excess liquid was aspirated and the slides were left 1 min to dry before putting them in -20°C methanol for 10 min . Next , the slides were dipped in -20°C acetone for 30 s . Once the slides were dry , they were incubated at 37°C for 30–60 min in a 40% formamide solution ( 40% formamide , 2x DEPC-treated saline-sodium citrate buffer ( SSC ) ) . LNA probe was added to the hybridization solution I ( 80% formamide , 1 mg/ml E . coli tRNA , 2x DEPC-treated SSC , 70 μg/ml calf-thymus DNA ) at a final concentration of 250 nM , and incubated at 80°C for 5 min before mixing with the hybridization solution II ( 20% dextran sulphate , 4mM VRC , 40U RNase inhibitor , 0 . 2% RNase-free BSA , 2x DEPC-treated SSC ) in a 1:1 ratio . The hybridization solution ( 25 μl ) was added to each well of the slide and hybridized for 2 h . The slides were then washed twice in 50% formamide and 2x DEPC-treated SSC solution for 30 min and briefly rinsed five times in DEPC-treated PBS . DAPI was added to a final concentration of 10 μg/ml in SlowFade solution ( Invitrogen ) were added to each well and the slide was covered and sealed using clear nail polish . The slides were either visualized immediately or stored in the dark at -20°C . The sequence of the probes , which were labeled with Cy3 , is given S2 Table . Samples ( 0 . 6 ml ) of cultures were collected , resuspended in 0 . 2 ml of phosphate-buffered saline ( PBS ) and the membrane dye FM4–64 ( Molecular Probes ) added to a final concentration of 10 μg ml−1 . Cells were observed on slides padded with 1 . 7% agarose . Images were acquired on Leica fluorescence microscopes DMR2A and DM6000B equipped with a Cool Snap HQ camera ( Roper Scientific , Arizona , USA ) and an iXonEM+ 885 camera ( Andor Technology , Connecticut , USA ) , respectively , or on a Nikon E1000 microscope equipped with an Orca-ER camera ( Hamamatsu Corporation , New Jersey , USA ) , using 63x or 100x lens objective plus an additional 1 . 6X optavar , phase-contrast optics and standard filters for visualization of GFP and FM4–64 . Images were acquired using Metamorph ( Molecular Devices , Berks , UK ) and processed for publication using Photoshop ( Adobe ) . The Genbank accession numbers for the sequences represented in Fig . 8A and S3C Fig , are as follows . For σG , CAB13407 . 1 ( B . subtilis ) and for σF , CAB14277 . 1 ( B . subtilis ) . For σA , NP_390399 . 2 ( B . subtilis ) , P00579 . 2 ( E . coli ) , AAG36964 . 1 ( T . aquaticus ) ; WP_011172619 . 1 ( T . thermophilus ) . For σE , NP_243422 . 1 ( B . halodurans ) , YP_175847 . 1 ( B . clausii ) , NP_389415 . 2 ( B . subtilis ) , ABS73878 . 1 ( B . amyloliquefaciens ) ; YP_006713129 . 1 ( B . licheniformis ) ; YP_085245 . 1 ( B . cereus ) , ABY76243 . 1 ( B . thuringiensis ) . For σK , NP_242151 . 1 ( B . halodurans ) , YP_175113 . 1 ( B . clausii ) , WP_003237137 . 1 ( B . subtilis ) , WP_015240253 . 1 ( B . amyloliquefaciens ) ; YP_079918 . 1 ( B . licheniformis ) ; YP_085663 . 1 ( B . cereus ) , ABY76244 . 1 ( B . thuringiensis ) . For the β´subunit of RNAP polymerase: WP_003225772 . 1 ( B . subtilis ) ; YP_491473 . 1 ( E . coli ) ; WP_003043700 . 1 ( T . aquaticus ) ; YP_005641350 . 1 ( T . thermophilus ) . | Precise temporal and cell-type specific regulation of gene expression is required for development of differentiated cells even in simple organisms . Endospore development by the bacterium Bacillus subtilis involves only two types of differentiated cells , a forespore that develops into the endospore , and a mother cell that nurtures the developing endospore . During development temporal and cell-type specific regulation of gene expression is controlled by transcription factors called sigma factors ( σ ) . An anti-sigma factor known as CsfB binds to σG to prevent its premature activity in the forespore . We found that CsfB is also expressed in the mother cell where it blocks ectopic activity of σG , and blocks the activity σE to allow σK to take over control of gene expression during the final stages of development . Our finding that CsfB directly blocks σE activity also explains how CsfB plays a role in preventing ectopic activity of σE in the forespore . Remarkably , each of the major roles of CsfB , ( i . e . , control of ectopic σG and σE activities , and the temporal limitation of σE activity ) is also accomplished by redundant regulatory processes . This redundancy reinforces control of key regulatory steps to insure reliability and stability of the developmental process . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"Methods"
] | [] | 2015 | Dual-Specificity Anti-sigma Factor Reinforces Control of Cell-Type Specific Gene Expression in Bacillus subtilis |
Cystic echinococcosis ( CE ) is one of the most widespread helminthic zoonoses and is caused by the tapeworm Echinococcus granulosus complex . CE diagnosis and monitoring primarily rely on imaging techniques , complemented by serology . This is usually approached by the detection of IgG antibodies against hydatid fluid ( HF ) , but the use of this heterogeneous antigenic mixture results in a variable percentage of false positive and negative results , and has shown to be useless for follow-up due to the long persistence of anti-HF antibodies in cured patients . To improve test performances and standardization , a number of recombinant antigens mainly derived from HF have been described , among them the B2t and 2B2t antigens . The performance of these antigens in the diagnosis and follow up of patients with CE has been so far evaluated on a limited number of samples . Here , we evaluated the performances of tests based on B2t and 2B2t recombinant antigens compared to HF in IgG-ELISA and immunochromatography ( IC ) for the diagnosis and follow-up of patients with CE in a retrospective cohort study . A total of 721 serum samples were collected: 587 from 253 patients with CE diagnosed by ultrasonography ( US ) , 42 from patients with alveolar echinococcosis and 92 from healthy donors from Salamanca ( Spain ) . The highest overall sensitivity was obtained with HF in ELISA ( 85 . 5% ) , followed by IC containing HF and 2B2t-HF ( 83 . 0% and 78 . 2% , respectively ) . The lowest sensitivity was obtained with B2t and 2B2t in ELISA ( 51 . 8% ) . The highest specificity was obtained with IC containing 2B2t-HF ( 100% ) , and the lowest with HF-ELISA ( 78 . 0% ) . The lowest cross-reactivity with sera from patients with alveolar echinococcosis was detected with the recombinant antigens in ELISA ( 9 . 5% - 16 . 7% ) and the highest with the HF-IC ( 64 . 3% ) . The results of B2t and 2B2t-ELISA were influenced by cyst stage , as classified by US according to the WHO-Informal Working Group on Echinococcosis ( WHO-IWGE ) , with low sensitivity for inactive ( CE4 and CE5 ) cysts , and by the drug treatment , with higher sensitivity in patients after drug treatment compared with patients not subjected to drug treatment . The two recombinant antigens in ELISA provided promising results for monitoring patients in follow-up , although their use is limited to patients with positive serology against them at the beginning of the follow-up . Potential biological reasons behind the low sensitivity of the recombinant antigens and possible strategies to enhance the performance of CE serology are discussed .
Cystic echinococcosis ( CE ) is a widespread zoonosis caused by Echinococcus granulosus sensu lato . Human CE is clinically complex and difficult to manage , due to the growth of the parasitic cysts in different organs ( although liver involvement represents around 70% of cases; [1] ) with variable clinical outcomes . Currently the management of CE encompasses surgery , percutaneous drainage , pharmacological treatment with benzimidazoles , and the “watch and wait” approach [2] . Clinical management is mainly based on cyst stage according to the WHO Informal Working Group on Echinococcosis ( WHO-IWGE ) applicable to liver cysts , ultrasound cyst classification and other clinical factors [2 , 3] . The diagnosis of CE is primarily based on imaging , complemented by serology . However , the performance of current serodiagnostic tools based on the detection of antibodies against hydatid fluid ( HF ) is far from optimal , due to poor specificity , sensitivity , and tests standardization . Factors influencing the sensitivity of serodiagnostic tests include , among others , cyst number , size and stage , and timing of serum collection in correlation to treatment ( rev . in [4]; [5 , 6] ) . A second point of concern in using HF-based tests is their low specificity , as sera from patients with a wide range of other infectious and non-infectious diseases , some of which commonly found in CE endemic areas ( e . g . alveolar echinococcosis–AE- , cysticercosis , clonorchiasis , fasciolosis , schistosomosis , ascariosis , amoebosis and malignancy ) have been described as cross-reacting with this crude antigen ( rev . in [4]; [5 , 7 , 8] ) . Especially high ( >50% ) is the cross-reactivity of HF with sera from patients with AE [9] . False-positive reactions against HF have been also described in healthy donors , with especially low specificity in areas where CE is endemic ( e . g . specificity of 53 . 8% in Iranian donors; [10] ) . Although this could be partially attributed to contact with parasite eggs , at the moment this occurrence could not be verified , its frequency is not quantified , and , in any case , apparently it does not anticipate the future development of a CE cyst [11] and is therefore useless for clinical purposes . Additionally , antigens used for the serological diagnosis of CE are not standardized , partly accounting for the large variability of results when applied in different settings , and the difficulty in comparing results from different groups . Finally , HF-based serology tests are of limited use for follow-up due to the long persistence of anti-HF antibodies even in cured patients [12] . Although monitoring of titres over time can indicate the outcome of therapy , this is not always clear-cut and years long follow-up of patients with imaging is required to monitor the evolution of cysts . Relapse rates vary greatly between reports , depending on cyst localization , stage , and type of therapy . In case of treatment of active hepatic cysts , mean relapse rates range from 2% to 24% after surgery [13 , 14] , to ≤15% after percutaneous treatment [15 , 16 , 17] , to an average of 25% at 2 years after albendazole therapy [18 , 19] . Clearly , the possibility to differentiate with a serological marker between patients with good progression ( to inactive cysts ) and cured patients from patients with recurrences is key , to provide a more precise prognosis and shorten patient follow-up [20 , 21] . Some recombinant antigens have shown better potential than HF to differentiate CE from other pathologies by serology and for the follow-up of treated patients , including several isoforms of the antigen B ( AgB1 , B2 , B3 and B4; e . g . , [22 , 23] ) and the antigen 5 ( Ag5 ) , among others ( rev . in [4]; [5] ) . AgB and Ag5 are the most abundant immune reactive molecules identified in HF [24] . Both antigens are expressed in all parasite’s life cycle stages , and AgB isoforms , encoded by a multigene family having at least five gene loci subunits ( B1–B5 ) appear to be variably expressed among E . granulosus strains , life cycle stages and cysts stages [25 , 26] . In previous publications , our group described the recombinant antigens B2t and 2B2t , derivatives of antigen B2 ( AgB2 ) and representing one and two tandem–head to tail- repeats of the E . granulosus G1 genotype AgB2 lacking the signal peptide , respectively , which showed preliminary good sensitivity and specificity in IgG-ELISA and good potential for the follow-up of patients with CE after surgical treatment [9 , 12] . Later , we showed that after surgical treatment , antibodies to the recombinant antigens 2B2t and EgP29 became negative in the majority of CE-confirmed , surgically cured patients included in the study . Nevertheless , the relatively low percentage of patients positive to these recombinant antigens before treatment raised doubts about their use for documenting primary cure [27] . However , these studies were limited by the low number of patients included and the deficiency of some patient’s clinical data . The achievement of a reliable serological diagnosis of CE could serve as an adjunctive to image techniques , especially when a challenge in distinguishing between an echinococcal cyst and other lesions , ranging from simple cysts to neoplasms , is found , and in follow-up cases when the interpretation of the treatment outcome is difficult ( e . g . in distinguishing between post-surgical cavity and relapse ) . The development of easy-to-use diagnostic devices based on the use of recombinant antigens is also a much-needed development in CE , to provide accessible testing in underserved areas where equipped laboratories are lacking , and to support field studies [28] . Five recent studies evaluated the performances of some of the currently commercially available rapid diagnostic tests for the diagnosis of CE [28–32] ( rev . in [4 , 33] ) . The performances of these rapid diagnostic tests seem overall comparable to those of traditional lab-based assays , however , as for other serodiagnostic tests , rapid diagnostic tests still lack standardization and show unsatisfactory sensitivity and specificity . To further assess the performance of B2t and 2B2t recombinant antigens in comparison to HF , we evaluated the three antigen preparations in IgG-ELISA on an extended panel of sera from patients with CE , both for the primary diagnosis of CE and for the follow-up of patients after different clinical management approaches . Additionally , the 2B2t antigen was used to develop a rapid immunochromatographic test containing the recombinant antigen in the test line and compared with the commercial kit VIRapid HYDATIDOSIS based on a semi-purified fraction of HF enriched in Ag5 and AgB ( Vircell S . L . , Granada , Spain ) for its diagnostic and monitoring performances .
A retrospective cohort study was conducted among patients with CE identified and followed-up by ultrasonography ( US; reference standard ) at the Division of Infectious and Tropical Diseases , San Matteo Hospital Foundation , Pavia , Italy , from 1998 to 2010 . Number of patients in this cohort were 253 . CE cysts detected in these patients were classified according to the recommendations of the WHO-IWGE on liver cysts [3] , with minor modifications regarding cyst activity [34] , as follows: active cysts CE1 , CE2 and CE3b , transitional cysts CE3a , and inactive cysts CE4 and CE5 . Patients with more than one cyst were classified according to the stage of the most active cyst . For the follow-up analysis , all patients with at least two consecutive sera were considered . Changes in cyst stage during the follow-up period were assessed by US . Patients who had undergone surgery or an aspiration technique were divided into “cured patients” , if showing no US images suggesting relapses during the follow-up period , and “non-cured patients” , if showing relapses detected by US during the monitoring period . Patients treated with albendazole were divided into patients with good response when the image evolved from active or transitional stages to inactive stages ( CE4 and CE5 ) and with poor response when the image did not change from active or transitional stages to inactive stages . Patients without treatment , included in the “watch and wait” approach ( showing CE4 or CE5 cysts ) were also checked for US stage changes during the follow-up period . To check the specificity and cross-reactivity of the serological tests , 42 samples from patients with AE confirmed by Em2plus serology were donated by Prof . Gottstein ( Institute of Parasitology , University of Berne , Switzerland ) , and 92 sera from healthy donors were kindly donated by Dr . Muñoz ( University Hospital , Salamanca , Spain ) . Human samples used in this study are included in the EchinoBiobank . The use and transfer of stocked human serum leftover from routine analyses carried out in San Matteo Hospital Foundation , Pavia , Italy , was approved by the Ethics Committee of IRCCS San Matteo Hospital Foundation , Pavia , Italy ( Acceptance Report 2015041 of 06/07/2015 ) . All samples used in this study were anonymized . HF was aseptically obtained from fertile sheep hydatid cysts at the Coreses slaughterhouse ( Zamora , Spain ) from sheep processed as part of the normal work of the abattoir and after receiving consent , using sterile syringes and gauges . The HF was centrifuged at 1 , 000 g for 5 min , and the protein concentration in the supernatant measured with the Micro BCA Protein Assay Kit ( Pierce ) . The supernatant was stored at -80°C until used . The expression vectors pGEX-4T2 and pGEX-4T1 ( GE Healthcare ) containing the relevant nucleotide sequences of the antigens B2t and 2B2t were used to transform Escherichia coli BL21 CodonPlus RIL competent cells ( Stratagene ) . Induction , expression , purification and thrombin cleavage of both proteins were performed as previously described [9 , 12] . Antigens used in this study are available in the EchinoBiobank ( https://biobancos . isciii . es/ListadoColecciones . aspx; collection nb . C0003432 ) , a collection of human and animal samples constituted under the umbrella of the HERACLES FP7 project [35] ( http://www . heracles-fp7 . eu/ ) . Ninety-six-well polystyrene plates ( Corning , Spain ) were incubated at 4°C overnight with 100 μl/well of HF ( 5 μg/ml ) , B2t or 2B2t ( 0 . 5 μg/ml ) in carbonate buffer ( pH 9 . 6 ) . Plates were then washed six times with phosphate-buffered saline ( PBS ) ( pH 7 . 4 ) with 0 . 05% Tween 20 ( washing buffer ) and blocked for 1 . 5 h at 37°C with 200 μl 1% bovine serum albumin ( BSA; Sigma Aldrich , Spain ) in washing buffer . Sera were then added in duplicate ( 100 μl/well ) at 1:200 dilution in blocking buffer , and plates incubated for 1 h at 37°C . After washing as described above , the secondary antibody ( peroxidase-labeled rabbit anti-human IgG; Sigma Aldrich , Spain ) was added ( 100 μl/well ) at a 1:2 , 000 dilution in blocking buffer , and plates incubated for 1 h at 37°C . After one further washing as described above , the reaction was developed with 100 μl/well of citrate buffer ( pH 5 ) , plus orthophenylene diamine ( 0 . 28 mg/ml; Sigma Aldrich , Spain ) and hydrogen peroxidase ( 0 . 4 μl/ml; Sigma Aldrich , Spain ) . The reaction was stopped with 50 μl/well of 3N sulfuric acid ( Panreac , Spain ) , and plates were read at 492 nm in an ELISA reader ( EAR 400; SLT Lab Instruments , Germany ) . The serological index ( SI ) was calculated for each optical density in each plate using the following formula: [ ( NC-S ) / ( NC-PC ) ]x100 , where NC and PC represent the negative and positive controls , respectively , and S stands for each serum . Negative control consisted of a pool of 10 sera from healthy donors with an OD in ELISA of 0 . 1 to 0 . 2 and positive control was a pool of 10 sera from patients with CE with an OD in ELISA of 0 . 4 to 0 . 5 . The use of SI was elected to avoid biases in the OD due to assay variability . An immunochromatographic strip ( IC ) carrying the recombinant 2B2t antigen in the test line and a semi-purified fraction of HF enriched in Ag5 and AgB ( 5/B ) in the conjugate was developed by Vircell S . L . ( Granada , Spain ) ( Fig 1 ) . Briefly , the 2B2t recombinant antigen was dispensed in the test line ( Biodot ZX1000 , UK ) on a nitrocellulose membrane ( Millipore HF135 , UK ) at concentrations ranging from 0 . 5 to 5 mg/ml on different polystyrene cards ( Lohmann , USA ) diluted in PBS pH 7 . 2 ( Sigma Aldrich , Spain ) using 5% methanol ( Panreac , Spain ) as co-precipitator agent . Cards were dried for 10 minutes at 50°C ( 0–250 °C Indelab 6741 A , Spain ) . Chicken IgY ( Jackson Immunoresearch , UK ) at 0 . 3 mg/ml ( PBS pH 7 . 2 ) was used as the control line . The 5/B antigen conjugate was prepared with 80 nm colloidal gold particles , manufactured by the seeding growth method defined by [36] . Protein conjugation was performed at 5 μg/ml in 5 mM phosphate buffer pH 7 for 15 minutes , followed by two blocking steps , with PEG 20 , 000 1% ( Sigma Aldrich , Spain ) and BSA 0 . 5% ( ID Bio , France ) respectively . After four centrifugation cycles at 17 , 500 rpm for 20 minutes ( Sigma 3K30 , Germany ) , the coated particles were resuspended in 5 mM sodium tetraborate buffer pH 8 ( Sigma Aldrich , Spain ) , containing 0 . 5% BSA and 0 . 025% PEG 20 , 000 . The conjugate for the control line of the test was prepared with the same particles and method aforementioned , and the adsorbed protein was donkey anti chicken IgY ( Jackson Immunoresearch , UK ) . Both conjugates were diluted in drying buffer pH 8 ( sodium tetraborate 5 mM , 1% BSA , 1% Tween 20 and 6% sucrose -Sigma Aldrich , Spain- ) and then dried at 50°C on a 6 mm wide polyester pad–sample pad- ( 7403 grade; Hollingsworth & Vose , UK ) for 20 minutes . The conjugate OD was determined in a UV-VIS spectrophotometer Cary 50 ( Varian , Australia ) . The adsorbent pad was cotton-litter paper ( 470 grade; Whatman ) . All these elements , along with the 2B2t striped membrane , were assembled on a polystyrene card and then cut with an automatic guillotine ( CM400; Biodot , UK ) . The different 2B2t concentrations in the test line were evaluated with serum samples already tested as positive or negative with the commercial Virapid Hydatidosis IC test [37] , resulting 2 . 5 mg/ml the optimal concentration for the development of the reaction in the test line ( Fig 1A ) . The performance of this IC test was compared with that of the Virapid Hydatidosis commercial IC test from Vircell containing a semi-purified fraction of HF enriched in Ag5 and AgB in both the conjugate and the test line . After adding the sample ( 50 μl of serum ) a colored band will appear in 10 min if the specific antibodies reacting with the antigen contained in the device are present in the sample . Each IC strip contains a control line for the validation of the assay . IC tests were used according to the manufacturer’s instructions . Briefly , 30 μl of serum were applied in the sample well and 2 drops of developer solution were added to the well after 5 minutes . Results were read after 30 minutes . In order to perform the reading of the test and to determine the positivity of the samples , the intensity card included in the kit was used , indicating 4 levels of color intensity ranging from 0 . 5 to 3 . When a visible test line has a color intensity lower than 0 . 5 , the result is considered negative . For ELISA tests , the best cut-off value for each antigen ( SI = 50 ) was previously determined by receiver operator characteristic analysis ( ROC ) using sera from patients with CE , considered true positive , and sera from donors plus sera from patients with AE , considered true negative [9] . Sera from patients with AE were considered true negative , due to the particular difficulties in differentiating some AE lesions from CE by imaging techniques , thus trying to define the less cross-reactive antigen in our settings . Sensitivity , specificity and cross-reactivity of the three antigens in ELISA and IC were estimated together with 95% confidence interval ( 95% CI ) , and compared between paired samples through the McNemar test . Test sensitivity was assessed using the first time point serum available from each patient with CE , with the exclusion of patients visited for the first time after surgical or percutaneous treatment . Specificity was calculated as the percentage of negative reactions in serum samples from healthy donors , while cross-reactivity was calculated as the percentage of positive reactions in sera from patients with AE . Sensitivity was also estimated separately for samples collected before and after albendazole treatment , and according to cysts characteristics ( i . e . , number , stage , size , and location ) . Differences in sensitivity by presence/absence of albendazole treatment and cyst characteristics were evaluated using multivariable logistic regression to account for potential confounding . The follow-up analysis was conducted considering as baseline the first test available after surgical or percutaneous treatment , or after the time-point before the end of the last treatment cycle with albendazole . Time at first test available was considered for untreated patients . The percentage of positive results of ELISA-based tests over time since the start of treatment ( ≤ 24 months; 25–48 months; > 48 months ) was calculated and graphically compared between samples from cured and non-cured patients who underwent surgery/percutaneous treatment , and between samples from patients with good and poor response after drug treatment . Moreover , the same percentage over time since first testing was presented for samples from untreated patients . Among patients who were positive at baseline , we compared rates of negativization between cured and not cured patients in those who underwent surgical or percutaneous treatment , and between patients with good and poor response in those treated with albendazole . Moreover , we compared rates of negativization among patients who underwent surgical or percutaneous treatment , those with a good response to albendazole treatment , and untreated patients . The incidence rates of negativization per 100 person-months ( PM ) of observation were estimated together with their 95% CI and differences among groups evaluated through the log-rank test . For each patient , person-time of observation was calculated as the number of months elapsed from baseline to last test available ( patients who remained positive during follow-up ) , or from baseline to negativization ( patients who became negative during follow-up ) . Time of negativization of ELISA was estimated by linear interpolation of SI values at the time of last positive test ( SI ≥ 50 ) and time at first negative test ( SI < 50 ) , while time of negativization of IC was estimated as the mid-time between the last positive and first negative test . Finally , we used the Wilcoxon rank-sum test to compare the median SI with interquartile range ( IQR ) in ELISA between patients treated with albendazole at the time they reached inactive cysts stage ( CE4 or CE5 ) and untreated patients with inactive cysts in watch and wait at baseline . P-values < 0 . 05 were considered statistically significant . All statistical analyses were performed using Stata 13 . 1 ( StataCorp LP , College Station , Texas , USA ) .
Demographics and relevant clinical data of the 253 patients with CE whose sera were included in the study are detailed in Table 1 . All patients were older than 18 years . Presence of the most active cyst in locations other than liver were found in 18 patients , including peritoneum ( 6 ) , spleen ( 4 ) , lung ( 2 ) , kidney ( 2 ) , retrovesical ( 1 ) , diaphragm ( 1 ) , pelvic ( 1 ) and thigh ( 1 ) . To evaluate the sensitivity of the serological tests , the first available serum from patients with CE was used . This excluded 33 patients who had undergone surgery or percutaneous treatment before the time of collection of the first sample available for this study , and included 220 patients with detectable CE cysts , from which 220 serum samples were used for ELISA ( 123 collected from patients who were not treated with drugs when the first serum was collected , and 97 collected after drug treatment ) . From these , only 165 serum samples had enough volume to be also tested in the two IC devices ( 92 collected from patients without previous drug treatment and 73 collected after drug treatment ) . Patients in follow-up with at least two consecutive serum samples ( n = 105 , from which only 74 had enough serum volume to be also tested in IC ) were divided in three groups according to their clinical management: ( i ) surgical or percutaneous treatment , ( ii ) drug treatment only and ( iii ) non-treated watch and wait patients with CE4 and CE5 cysts . All patients included in the follow-up study had liver CE . Patients who had undergone percutaneous treatment or surgery were divided into cured patients with stages that had progressed to inactive residual lesions after percutaneous treatment or after surgery ( n = 17 , 55 samples tested with ELISA and n = 12 , 28 samples tested with IC ) and non-cured patients showing relapses ( n = 4 , 13 samples tested with ELISA and n = 3 , 9 samples tested with IC ) . Patients treated with albendazole were divided into two groups . The first group comprised patients whose US image changed from active or transitional stages to inactive stages ( CE4 and CE5 ) in response to treatment ( n = 18 , 70 samples tested with ELISA and n = 13 , 42 samples , tested with IC ) over the follow-up period , with no sign of reactivation after having reached the CE4 stage . The second group comprised patients with poor response to treatment , in whom there was no permanent change from active or transitional stages to inactive stages ( treatment failure ) during the follow-up period ( n = 46 , 186 samples tested with ELISA and n = 31 , 106 samples tested with IC ) . Twenty patients without treatment having inactive cysts managed with the watch and wait approach were checked for US changes during the follow-up period . No patient showed US changes . This group was assessed for SI in ELISA ( n = 20 , 63 samples ) or band intensity in IC changes ( n = 15 , 38 samples ) during the follow-up period . Additionally , 25 patients who underwent drug treatment reached inactive cyst stage ( CE4 ) during follow-up , while 24 untreated patients did so spontaneously , showing inactive cysts at baseline . A flow chart of participants , including the mean follow-up time with standard deviation ( SD ) and the number of samples collected for each of the above-mentioned groups is shown in Fig 2 . Sensitivity , specificity and cross-reactivity for each test are shown in Table 2 . From the highest to the lowest sensitivity , tests results were as follows: HF-ELISA ( 85 . 5%; 95% CI: 80 . 1–89 . 8% ) , VIRapid Hydatidosis ( 83 . 0%; 95% CI: 76 . 4–88 . 4% ) , IC containing 2B2t-HF ( 78 . 2%; 95% CI: 71 . 1–84 . 2% ) , B2t-ELISA and 2B2t-ELISA ( 51 . 8% both; 95% CI: 45 . 0–58 . 6% ) . When sensitivity was compared between samples collected before and after treatment , all tests performed better with serum samples collected after treatment , especially the two recombinant antigens in ELISA . Both recombinant antigens in ELISA showed a significantly lower sensitivity compared with the ELISA-HF and both IC tests . Specificity as assessed with sera from healthy donors , from the highest to the lowest , was as follows: IC containing 2B2t-HF ( 100%; 95% CI: 88 . 4–100% ) , 2B2t-ELISA ( 98 . 8%; 95% CI: 93 . 4–100% ) , VIRapid Hydatidosis ( 93 . 3%; 95% CI: 77 . 9–99 . 2% ) , B2t-ELISA ( 88 . 3%; 95% CI: 79 . 0–94 . 5% ) and HF-ELISA ( 78%; 95% CI: 67 . 6–86 . 4% ) ( Table 2 ) . Cross-reactivity of the different tests checked against samples from patients with AE was very high for the HF both in the VIRapid Hydatidosis IC ( 64 . 3%; 95% CI: 48 . 0–78 . 4% ) and ELISA ( 52 . 4%; 95% CI: 36 . 4–68 . 0 ) , and moderate to low for the recombinant antigens in IC ( 31 . 0%; 95% CI: 17 . 6–47 . 1% ) and ELISA ( 16 . 7%; 95% CI: 7 . 0–31 . 4% for 2B2t and 9 . 5%; 95% CI: 2 . 7–22 . 6% for B2t ) . Statistically significant differences in specificity were found between the ELISA-HF and both the 2B2t-ELISA and 2B2t-HF IC test , and between the latter and the B2t-ELISA test . Differences in cross-reactivity were statistically significant between both recombinant antigens in ELISA and the HF-based ELISA and VIRapid Hydatidosis IC test , between B2t-ELISA and 2B2t-HF IC , and between IC tests . The analysis of the association between clinical variables and sensitivity of the different ELISA and IC tests is presented in Table 3 . Independently of the other clinical characteristics , cyst stage was found to influence sensitivity of all serological tests . Among patients harboring active or transitional cysts , the highest sensitivity was obtained in HF-ELISA , with the exception of patients with CE3b cysts , among whom the 2B2t-HF IC showed the highest sensitivity . Sensitivity of the two recombinant antigens in ELISA was the lowest for all cyst stages , ranging from 84 . 6% ( 2B2t for CE1 cysts ) to 64 . 8% ( B2t for CE3b cysts ) among patients harboring active cysts , and drastically dropping among those with inactive cysts , especially CE5 cysts ( 5 . 1% and 2 . 6% for B2t and 2B2t , respectively ) . Sensitivity of the HF-based ELISA was significantly influenced by the location of cysts , with a higher sensitivity observed among samples of patients with cysts located in the liver compared with those of patients with cysts located in other organs . The time point of serum collection showed to influence the sensitivity of ELISA based on the recombinant antigens , which was found to be significantly higher among samples collected after treatment . Finally , sensitivity of the IC containing the 2B2t recombinant antigen was found to significantly increase with cyst size . The percentages of positive samples in ELISA tests over time after surgery or percutaneous treatment among patients cured and not cured are shown in ( Fig 3A and 3B ) . Overall , the percentage of positive samples in cured patients decreased over time when using recombinant antigens but not HF ( Fig 3A ) , while this percentage increased over time and remained stable at 100% in non-cured patients using all antigens ( Fig 3B ) . Evolution with time of the percentage of positive samples in HF-ELISA showed a constant trend , with percentage at about 100% , among all patients who underwent drug treatment , regardless of their response to therapy ( Fig 3C and 3D ) . Among patients with both good or poor response to drug treatment , a drop in the percentage of samples tested positive using recombinant antigens was observed in the second time period assessed ( 2 to 4 years since the start of treatment ) , with an increase in the last time period ( > 4 years since the start of treatment ) . Among untreated patients , the percentage of positive samples in HF-ELISA was over 60% for the first evaluated period ( 0 to 2 years from baseline ) , increasing to 100% after 4 years of follow-up . On the contrary , positivity rate in ELISA based on recombinant antigens was very low ( < 20% ) at the beginning of the studied period and decreased over time to reach 0% after 4 years ( Fig 3E ) . Progression to negativity by treatment outcome ( cured vs . non-cured patients ) among patients who underwent surgical intervention or percutaneous treatment and classified as positive at baseline is shown in Table 4 separately for each test . The B2t-ELISA is the only test showing a statistically significant difference in the incidence rate of negativization between cured and non-cured patients . This is also the antigen showing the best incidence of negativization in cured patients ( 13 . 04 per 100 PM ) , followed by the 2B2t in ELISA ( 9 . 52 per 100 PM ) , the HF-based IC strip ( 1 . 16 per 100 PM ) and the HF-ELISA ( 1 . 01 per 100 PM ) . None of the patients with a negative outcome ( non-cured ) was tested negative with any test during the follow-up . Incidence of negativization in patients who underwent drug treatment and classified as positive at baseline did not show any statistically significant difference between patients with good and poor response to treatment ( Table 5 ) . Moreover , progression to negativity in patients who were positive at baseline did not significantly differ between patients under drug treatment with good response and patients followed with the watch and wait approach ( Table 6 ) . However , the median SI in drug-treated patients at the time inactive cysts ( CE4 ) were detected was significantly higher compared to median SI in watch-wait patients in all ELISA tests ( Table 7 ) .
The recombinant antigens B2t and 2B2t [9 , 12] were tested in ELISA to compare their sensitivity with that of HF using an extended number of samples . Additionally , the 2B2t antigen was included in an IC test and compared with the HF ( VIRapid Hydatidosis ) in the same test . Similar to our previous results , recombinant antigens showed lower sensitivity compared to HF in ELISA , and although the recombinant 2B2t antigen showed enhanced sensitivity in the IC test , this was still lower than the sensitivity of the HF . A deeper analysis of the different clinical variables that could account for the low sensitivity of the recombinant antigens in ELISA showed that these were influenced by the cyst stage and the timing of serum collection ( before and after drug treatment ) . Patients with inactive cysts were clearly testing more frequently negative with the two recombinant antigens than patients with active or transitional cysts . These differences were also statistically significant for the HF-ELISA and for the two IC tests . Similar results were obtained by Pagnozzi et al . [38] using a purified Ag5 from HF in ELISA and by Yang et al . [39] using AgB1 . Tamarozzi et al . [22] also detected a statistically significant difference between OD values of active and inactive cyst groups against the recombinant antigen B1 in ELISA . In a recent study , variation of the immune-proteome profile of HF along the cyst progression has been shown [25] , with patients having CE2 and CE3 stages exhibiting strong antibody responses against diverse AgB and Ag5 proteoforms , while sera from patients with CE1 , CE4 , and CE5 stages mainly reacting to Ag5 and cathepsin B but not against AgB proteins . This has been attributed to changes in the antigenic composition of cysts depending on cyst stages , e . g . , the AgB2 antigen expression is reduced in degenerating and inactive cysts [40] . This supports empirically the finding that the different antigenic composition of the various cyst stages could largely influence the antibody reactivity found in patients with active , transitional or inactive cysts against specific antigens , including AgB2 antigen . In this respect , sensitivity of serological tests could be enhanced combining several recombinant antigens , as suggested by Jiang et al . [23] , who found a differential response of patients with CE against different recombinant antigen B isoforms . Additionally , it should be mentioned that AgB genes present high polymorphism and variable transcription profiles in different E . granulosus genotypes ( rev . in [4] ) . This could also influence the immune response against the corresponding native and recombinant antigens . The second variable statistically influencing the positivity in the ELISA containing any of the two recombinant antigens tested here was the drug treatment . Albendazole treatment results in a significant increased number of patients testing positive against B2t and 2B2t . Drug treatment similarly affects the serological results obtained with the recombinant antigen B1 [6 , 22] . These authors have suggested that drug treatment could result in the spillage of antigens from damaged cysts and in the increase of antibody responses , which seems to be especially true for defined antigens [22] , at least during one year post-treatment [6] . Sensitivity of the ELISA test based on the recombinant antigen B2t reported in previous works contrasts with what found here . Hernandez-Gonzalez et al . [12] reported a sensitivity of 91 . 2% for B2t antigen in ELISA , similar to that reported for HF in the same publication . Nevertheless , the majority of clinical data of the 102 CE patients whose sera were included in that publication were lacking , including cyst stage , thus comparison of those results with the present work is not feasible . Later , an extended panel of sera from patients with CE whose clinical data were available was tested against both recombinant antigens B2t and 2B2t in ELISA [9] . Sensitivity reported in that study was 79% for B2t and 87 . 6% for 2B2t . Although closer to results of the present study , it is noteworthy that of the 186 sera tested , only 2 were from patients with CE4 cysts , and none from patients with CE5 cysts . To further stress the importance of the distribution of cyst stages in the panel of samples used for accuracy studies , the logistic regression analysis showed that only cyst number per patient influenced the outcome of tests based on either recombinant antigen , thus the reliability of the influence of the cyst stage in the performance of these antigens was biased in that study . Indeed , the mean value of sensitivity for active and transitional cysts in our study was 73 . 8% for B2t and 76 . 9% for 2B2t , close to that found by Hernandez-Gonzalez et al . [9] for the same antigens mainly challenged with sera from patients with active and transitional cysts . Additionally , other forms of the recombinant AgB1 and AgB2 have been tested by different authors , showing in general higher sensitivity than the detected here ( rev . in [4] ) . Majority of these previous studies cannot be compared with the present study , due to the lack of clinical data in most of the studies , although differences could be also attributed , at least in part , to the antigenic differences among different versions of the same recombinant antigen . When the recombinant antigen 2B2t was used in the IC test combined with HF in the conjugate , the sensitivity was higher than that found for the same recombinant antigen in ELISA . In particular , increased sensitivity was found for the detection of inactive cysts , similar to what observed using the IC test containing only HF ( VIRapid Hydatidosis ) , but with higher specificity and less cross-reactivity with sera from patients with AE . These results suggest the use of recombinant antigens in combination with HF in easy-to-use diagnostic tests to improve the performance of tests based only on HF . One of the many contentious points in the clinical management of patients with CE is the usefulness of serology for the follow-up of treated patients . Antibodies against HF persist after cure , and although specific antibody isotypes and sub-isotypes against HF , including IgE , IgM , IgG2 and IgG4 , have been suggested to be of use for follow-up by several authors , the demonstration of usefulness of serology has been hampered by the low number of tested samples in published work and the lack of these antibodies in a percentage of patients with CE ( rev . in [4 , 5] ) . Similarly , the use of a number of recombinant or purified native antigens has been investigated for the follow-up of treated patients . For example , a drop in specific antibody titers after successful surgical treatment for CE has been reported against the recombinant AgB , AgB2t , AgP29 , and HSP20 [12 , 39 , 41 , 42 , 43] . Published work shows that cytokine detection seems not to be more useful than antibody detection for primary diagnosis of CE patients ( rev . in Siles-Lucas et al . , 2017 ) . Detection of specific cytokines and their fluctuation ( e . g . , IL-4; [44 , 45] ) could have potential in the definition of cyst vitality ( viability ) and follow-up of treated patients , with limitations regarding test sensitivity and available facilities . Additionally , extended cohorts of patients should be tested before drawing conclusions about the usefulness of cytokine detection in the follow-up of CE patients . Here , our results also show a decrease in the number of positive patients against the recombinant antigen-based ELISA in cured patients after surgery or percutaneous treatment , in comparison with non-cured patients , especially against the B2t antigen . Taken together , these results suggest that serology performed using defined recombinant antigens could be useful for the follow-up of surgically and percutaneously treated patients . However , this approach is only possible when patients show positive serology against the target antigens , a condition unfortunately found only in a variable percentage of subjects ( e . g . for P29 and 2B2t; [42 , 43] ) . This was also true for the recombinant antigens B2t and 2B2t in the present study , and although a decline in specific antibodies can be detected in cured patients after intervention , the number of patients testing positive at baseline was very low . This is related with one of the limitations of the present study , because although the number of patients in this clinical management modality was preliminary acceptable for statistical analysis , those showing positive serology at the time of first test available were limited . Nevertheless , sample size was enough to suggest the uselessness of HF and the apparent uselessness of the IC test for the follow-up of patients after surgery or aspiration treatment , due to persistence of positivity against them for prolonged time regardless of treatment outcome . Very similar conclusions can be extracted from our results regarding the usefulness of serology based in HF for the follow-up of drug-treated patients . Contrasting , recombinant antigens in ELISA showed here a higher number of positive patients at baseline for drug treated patients and results suggest that progression to negativity is faster in patients with good response to treatment compared with patients with poor response , although differences were not statistically significant . Thus , the low reactivity of B2t and 2B2t recombinant antigens against sera from patients with inactive cysts , similar to what reported for other recombinant antigens ( e . g . HSP60 and B1; [39 , 46] ) could be also of use for the follow-up of patients treated medically . In these patients , US evolution from active or transitional to inactive stage may be achieved , but a proportion of cysts , although inactive on imaging , remain biologically viable , as reactivation is observed over time . This is best shown by the behavior of CE3b cysts that almost invariably reactivate , after an initial solidification , after the end of albendazole treatment [18 , 47] . Patients analyzed here with spontaneously inactivated CE4/CE5 cysts without treatment ( in watch and wait ) showed no changes in cyst stage along the follow-up period , in accordance with what observed in previous studies [27 , 47 , 48] . However , a decline of specific antibodies along the follow-up could be detected . This shows , as expected , that a period of time has to elapse between the inactivation of the cyst and the loss of circulating antibodies against defined antigens . Further follow-up studies using B2t and 2B2t-based serology on a cohort of patients undergoing inactivation and reactivation , not available in this cohort , would therefore be of interest to assess if tests based on these antigens can indicate evolution of cysts before morphological features change on imaging , thus reducing patient follow-up . This would be absolutely useful especially in resource poor settings where expertise and availability of portable US are scant , where the reliability of such result , especially if obtainable using a rapid IC test , could save transportation time and costs for patient . In the present study , antibody levels were significantly higher in patients with CE4 cysts that have evolved from active or transitional stages after drug treatment than in patients reaching the CE4 stage spontaneously . Whether the higher number of positive , drug-treated patients with CE4 cysts compared with patients with “spontaneous” CE4 cysts is due to the incomplete loss of viability of CE4 cysts after drug treatment , as is highly likely , should be further investigated . In summary , sensitivity of serodiagnostic tests based in ELISA containing only derivatives of AgB2 in their recombinant form is substantially lower compared to tests based on HF , especially for the detection of patients with CE harboring inactive cysts . However , this lower sensitivity for inactive lesions could be exploitable for the follow-up of patients whose serology is positive against the recombinant B antigens at the beginning of observation . Nonetheless , specificity and cross-reactivity performance of the recombinant antigens was much better than the performance of the HF , which indicates the use of these antigens especially in patients showing images of lesions similar to CE . A further encouraging result is that the combination of recombinant antigens derived from AgB2 with HF in easy-to-use devices ( IC ) has improved sensitivity compared to tests based only on recombinant antigens in ELISA , while offering higher specificity and less cross-reactivity than similar IC devices containing only HF . The IC test containing the 2B2t recombinant antigen was also tested for its usefulness in the follow-up of patients with CE for the above-mentioned groups . No statistically significant differences were found when comparing any of the patients’ groups , thus showing no applicability in the follow-up of patients with CE after treatment . | Cystic echinococcosis ( CE ) is a helminthic zoonosis caused by Echinococcus granulosus sensu lato . CE diagnosis and monitoring is of paramount importance for the clinical management of patients and primarily rely on imaging techniques , complemented by serology . CE serology is usually based on the detection of antibodies against hydatid fluid ( HF ) , but the use of this heterogeneous antigenic mixture shows several drawbacks , including false positive and negative results , unsatisfied predictive values , and long persistence of detectable antibody levels in cured patients . As an alternative , to improve test performances and standardization , several recombinant antigens have been described , but these have been so far evaluated only on a limited number of samples . Here , two recombinant antigens derived from one of the immunodominant HF antigens ( antigen B2 ) have been tested in enzyme-linked immunosorbent assay ( ELISA ) and in immunochromatographic strips ( IC ) against 721 serum samples . Although more specific than the HF , the recombinant antigens in ELISA showed low sensitivity for patients with inactive ( CE4 and CE5 ) cysts and for patients not subjected to drug treatment . This limited their use for follow-up , although promising , to those patients with positive serology at the beginning of the follow-up period . These results will aid in the future development of a serological test with enhanced performance in the diagnosis and follow-up of patients with CE . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"cystic",
"echinococcosis",
"medicine",
"and",
"health",
"sciences",
"enzyme-linked",
"immunoassays",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"immunology",
"tropical",
"diseases",
"parasitic",
"diseases",
"surgical",
"and",
"invasive",
"medic... | 2018 | Evaluation of the recombinant antigens B2t and 2B2t, compared with hydatid fluid, in IgG-ELISA and immunostrips for the diagnosis and follow up of CE patients |
Research on mate choice has primarily focused on preferences for quality indicators , assuming that all individuals show consensus about who is the most attractive . However , in some species , mating preferences seem largely individual-specific , suggesting that they might target genetic or behavioral compatibility . Few studies have quantified the fitness consequences of allowing versus preventing such idiosyncratic mate choice . Here , we report on an experiment that controls for variation in overall partner quality and show that zebra finch ( Taeniopygia guttata ) pairs that resulted from free mate choice achieved a 37% higher reproductive success than pairs that were forced to mate . Cross-fostering of freshly laid eggs showed that embryo mortality ( before hatching ) primarily depended on the identity of the genetic parents , whereas offspring mortality during the rearing period depended on foster-parent identity . Therefore , preventing mate choice should lead to an increase in embryo mortality if mate choice targets genetic compatibility ( for embryo viability ) , and to an increase in offspring mortality if mate choice targets behavioral compatibility ( for better rearing ) . We found that pairs from both treatments showed equal rates of embryo mortality , but chosen pairs were better at raising offspring . These results thus support the behavioral , but not the genetic , compatibility hypothesis . Further exploratory analyses reveal several differences in behavior and fitness components between “free-choice” and “forced” pairs .
The evolution of mate choice has been the focus of much research , and many studies have attempted , with a variety of experimental approaches , to measure the fitness benefits gained by choosy individuals ( e . g . , [1 , 2–7] ) . Those benefits can be either direct , if offspring quality or quantity is increased due to the partner’s behavior ( including reproductive investment ) , or indirect , if offspring quality is improved by the genetic contribution of the partner . To date , the central debate has been about ( i ) the relative importance of direct versus indirect fitness benefits arising from the overall quality of the chosen partner ( i . e . , good parent versus good genes; Fig 1 , vertical black arrow ) [8 , 9] , or about ( ii ) the relative importance of the two types of indirect benefits ( i . e . , good genes versus compatible genes; Fig 1 , horizontal black arrow ) [10–13] . Several studies on mate choice have shown that , in some species , mating preferences can be largely specific to the individual [14–19] . Such mate preferences may function to maximize offspring viability by bringing together compatible combinations of genes ( top right in Fig 1 ) . However , the alternative hypothesis that mate choice could lead to direct benefits arising from the phenotypic ( e . g . , behavioral ) compatibility of the two partners ( bottom right in Fig 1 ) has received only little attention [20–26] , despite suggestive evidence that the combination of both parents’ behaviors or other phenotypes can affect breeding success . Compatible partners could , for instance , be better at coordinating tasks , at sharing them or at complementing each other’s performance on various tasks [23–25 , 27–31] , or they might simply be more effective at stimulating one another’s reproductive investment [32–34] . Mate choice for such behavioral compatibility might be especially important in species with intense bi-parental brood care and with long-lasting , monogamous pair bonds , like humans or many bird species . Previous experiments that aimed to quantify the fitness benefits of mate choice arising from partner compatibility typically compared two categories of individuals: those paired up with their preferred partner versus those that were given a non-preferred partner [35–43] , or a random partner from the population [44 , 45] . The problem with this approach is that the effects of individual quality and pair compatibility are confounded , because only the force-paired group includes individuals that might never have been chosen ( i . e . , low-quality individuals ) . Some studies addressed this issue by presenting evidence that the rejection of a particular mate depended on the choosing individual’s identity [46] , or that non-preferred and preferred individuals did not differ in morphological traits [43] . Other studies have compared the reproductive success of a choosing individual ( paired with its preferred partner ) with the reproductive success of a naïve individual paired with that same ( or another ) preferred individual [2 , 47–50] . In the latter design , chosen and assigned partners are on average of equal quality . However , choosing individuals were often discarded if they did not meet a certain criterion regarding their strength of preference ( [2 , 48 , 49] , but see [47 , 50] ) . If choosiness is associated with an individual’s quality [17 , 51] , the selected subset of choosing individuals might differ in quality from the random pool of naïve individuals to which they are compared . However , in two experimental studies on invertebrates , none of the above issues apply; these studies found no fitness benefit of mate choice for compatibility [47 , 50] . Here , we employ an experimental design somewhat similar to [50] , to eliminate the effect of mate quality: we compare the fitness of individuals that bred with their preferred partner with those that obtained , after having expressed their preference , the preferred partner of another individual . The main aim of this study is thus to quantify the benefits of mate choice that arise from partner compatibility , while circumventing confounding effects of variation in partner quality . The second aim of our study is to tease apart indirect compatibility advantages ( compatibility of parental genes expressed in the offspring ) from direct ones ( parental phenotypic compatibility ) , using a model species in which these benefits of mate choice can be disentangled . The zebra finch ( Taeniopygia guttata ) is a socially monogamous species with biparental care , in which partners mate for life [52] . In this species , female mate preferences are predominantly individual-specific ( i . e . , females show little consensus regarding which male is the most attractive ) [15 , 53–57] , suggesting that they may target genetic or behavioral compatibility . In captive and wild populations , high rates of embryo and offspring mortality are found , even in the absence of inbreeding [25 , 52 , 58–60] . Cross-fostering of freshly laid eggs ( see [61 , 62] and S1 Text ) showed that most of the variance in embryo mortality ( before hatching ) is explained by the identity of the genetic parents rather than the foster‐parents ( based on n = 1 , 529 fertilized eggs , S1 Table ) , whereas most of the variance in offspring mortality ( after hatching ) is explained by foster-parent rather than genetic parent identity ( n = 1106 offspring , S1 Table ) . Based on these results , we assume that , in zebra finches , embryo mortality primarily reflects genetic incompatibility ( as in other species [63 , 64] ) , while offspring mortality primarily results from the behavior of the caring parents ( here broadly referred to as “behavioral incompatibility” ) . Experimentally preventing mate choice should thus lead to an increase in embryo mortality if mate choice is targeting genetic compatibility , and to an increase in offspring mortality if mate choice is targeting behavioral compatibility . Alternatively , if individual-specific mate preferences only reflect indecision by the animal or measurement error [15] , preventing mate choice would have no fitness consequences . We studied 160 bachelor birds from a recently wild-derived population of zebra finches . Each individual could freely choose a partner from a group of 20 individuals of the opposite sex during a long , nonbreeding season . This setup reflects the natural situation in the sense that zebra finches are opportunistic breeders and do not reproduce if the environment is not suitable , but they still form life-long pair bonds irrespective of breeding opportunities . Furthermore , the species is gregarious , such that individuals have many potential partners to choose from . Pairs were identified by the occurrence of allopreening because we found that this best reflects mutual preferences rather than being the outcome of intra-sexual competition ( see S2 Text ) . We hereafter focus on female preferences; however , because observed allopreening preferences were mainly reciprocal between females and males ( see also S2 Text ) , any observed effects of the experimental treatment described below could be due to females , males , or both sexes not being able to breed with their ( most ) preferred partner . Females from these pairs were alternately assigned to one of two treatments: half of them were allowed to stay with their chosen partner , while the other half were force-paired with the chosen partner of another female from the same aviary . This ensured that , on average , individuals of both treatments were of the same quality , even if assortative pairing for quality had happened due to intra-sexual competition . All pairs were then placed in individual cages for a few months to enforce pair-bonding in the non-chosen pairs ( force-pairing is effective in this species if assigned mates are co-housed in a cage for long enough , see “Methods” ) . After this period in separate cages , pairs were given the opportunity to breed for about five consecutive months ( allowing about three successful broods ) in communal aviaries , each containing three pairs from each treatment group . This entire procedure was repeated a second time with the same birds ( i . e . , free choice during a nonbreeding period , force pairing in cages , and breeding in communal aviaries ) . This was planned a priori to obtain repeated measurements on individuals under different pairing conditions with a large enough sample size to allow the detection of weak effects . For the second breeding period , two-thirds of the pairs from the first breeding period were broken up; individuals chose a new partner and were either assigned to the same or the other treatment . The other third of the pairs were allowed to keep their partner ( chosen or non-chosen ) from the first breeding period . This allowed us to better control for any effects of pair-bond duration in statistical models comparing chosen and non-chosen pairs , given that pair-bond formation in chosen pairs systematically started earlier ( during the free choice period ) than in non-chosen pairs ( in cage ) . In total , we monitored behavior and reproductive success of 46 chosen pairs ( C ) and 38 non-chosen pairs ( NC ) . Measures of reproductive success were based on paternity analyses that included dead embryos , dead chicks , and surviving offspring . Behavior was scored based on direct observations ( 285 h ) and video recordings ( 1 , 424 h ) .
We calculated relative fitness of individuals as the total number of genetic offspring produced in a given breeding period that reached independence ( 35 d old ) , relative to the number produced in the same period by the other individuals in the same aviary . Males of chosen pairs had a 45% higher relative fitness than males of non-chosen pairs ( C = 1 . 16 , NC = 0 . 80 , p = 0 . 03 , n = 84 male breeding periods , see T1-1 for model details , Fig 2 ) . Females of chosen pairs had a 30% higher relative fitness than those of non-chosen pairs , but the difference was not significant ( C = 1 . 09 , NC = 0 . 84 , p = 0 . 12 , n = 84 , T1-2 , Fig 2 ) . The difference between the sexes was not significant ( interaction between treatment and sex: p = 0 . 36 ) and resulted from extra-pair paternity ( see below ) . Thus , on average , individuals from the chosen pairs had a 37% higher fitness . This difference in fitness was not due to differences in pair bond duration between the treatments groups , as this covariate did not correlate with fitness ( non-significant trends against the expectation , Table 2: T2-1 and T2-2 ) and was therefore removed from the models T1-1 and T1-2 . The overall fitness difference observed was not due to differential investment in egg production by the females of the two treatment groups ( total number of eggs laid: C = 13 . 5 , NC = 14 . 4 , p = 0 . 56 , n = 84 , T1-3 ) . However , non-chosen pairs tended to have a higher proportion of disappeared or buried eggs ( C = 12% , NC = 19% , p = 0 . 07 , n = 1172 eggs laid , T1-4 ) , and had significantly more clutches that contained infertile eggs ( C = 8% , NC = 23% , p = 0 . 01 , n = 216 clutches , T1-8 ) . To test the genetic incompatibility hypothesis , we compared the proportion of dead embryos between treatment groups , considering all fertilized and incubated eggs . We only included the genetic eggs of each pair , that is , we excluded all extra-pair young ( 9% of the eggs ) , but included eggs that were dumped into the nest of other pairs ( 13% of the genotyped eggs ) . Note that removing dumped eggs ( potentially suffering higher rate of embryo mortality [65] ) from the analysis did not change the conclusions . Furthermore , we only included eggs that were incubated without interruption , excluding those that were buried in the nest material before incubation was completed ( based on daily nest checks ) . The rate of embryo mortality did not differ between chosen and non-chosen pairs ( C = 20% , NC = 22% , p = 0 . 68 , n = 707 fertilized eggs , T1-5 , Fig 3A ) . To test the behavioral compatibility hypothesis , we compared the proportion of dead offspring between treatment groups , considering all hatched eggs in a pair’s nest ( including extra-pair offspring and hatchlings from dumped eggs ) . Offspring mortality was significantly higher when chicks were reared by non-chosen pairs ( C = 32% , NC = 52% , p = 0 . 03 , n = 594 hatched eggs , T1-6 , Fig 3B ) . Pair bond duration did not influence this result ( T2-6 ) . The probability of survival may also decrease if the offspring is unrelated to one or both of the parents . To check this , we added the status of the offspring ( within-pair versus extra-pair young , offspring from dumped versus not dumped egg ) into model T1-6 . We found that the treatment effect was still significant ( p = 0 . 045 ) , but offspring status was not ( mortality of within-pair young = 38% , extra-pair young = 55% , p = 0 . 15; dumped = 40% , non-dumped = 39% , p = 0 . 91; underlying data can be found in S1 Data ) .
Many studies have attempted to quantify the benefits of mate choice [2–7 , 35–46 , 48–50 , 66 , 67] , but only a few have quantified the fitness benefits of mate choice for compatibility while excluding quality benefits ( see e . g . [46] and [50] ) ( Fig 1 ) . Our experimental design allowed us to circumvent the potentially confounding effect of mate quality by comparing pairs of individuals that chose each other with pairs that were composed of random individuals who did not choose each other , but had both been chosen by another individual . Pairs that formed through free mate choice had a 37% higher fitness than pairs that were “forced” experimentally ( Fig 2 ) . This suggests that it is unlikely that the between-individual disagreement about mate attractiveness simply reflects indecision or measurement error . Our results suggest instead that individual-specific mate preferences lead to significant fitness consequences . Our study system , furthermore , allowed us to disentangle direct ( behavioral ) benefits of mate choice from indirect ( genetic ) benefits ( Fig 1 ) . Chosen pairs , compared to arranged ones , had a 38% lower rate of offspring mortality ( Fig 3B ) . Under the assumption that offspring mortality systematically depends on parental behavior , this result supports the hypothesis of mate choice for behavioral compatibility . Ideally , our experiment should be repeated while cross-fostering eggs to exclude confounding factors . Indeed , our conclusions depend on the generalizability of the results from our previous study ( S1 Text ) . The finding that offspring mortality after hatching primarily depends on the rearing parents and not on the genetic parents ( S1 Table ) can likely be generalized from our previous cross-fostering experiment to this study; in both studies , many offspring apparently died from starvation , and an offspring that is not fed will die irrespective of its genetic quality . Finally , chosen and arranged pairs had an equal rate of embryo mortality ( Fig 3A ) . Given that embryo mortality primarily depends on the genetic parents and less on the incubating parents ( S1 Table ) , this result argues against the hypothesis of mate choice for genetic compatibility . At least , our results suggest that individuals did not select a partner with whom they would have minimized the rate of embryo mortality . Several earlier experimental studies favored the genetic compatibility hypothesis based on the observation that offspring from “free-choice” pairs had a higher viability than those from “forced” pairs [35–37 , 40 , 43 , 46 , 66] . However , in these experiments females were forced to mate with random males from the population or with non-preferred males , some of which may have been of lower absolute quality ( but see [46] ) . Hence , the previously observed effects on offspring viability may be explained by differences in both genetic quality and compatibility . In general , mate choice for genetic compatibility may not easily evolve , because it requires that the incompatibility-causing loci are tightly linked ( e . g . , via pleiotropy ) to a detectable phenotype and to a mechanism ensuring the appropriate assortative or disassortative preference [68] . At least in zebra finches , such a complex adaptation that would allow them to minimize embryo mortality by choosing a genetically compatible partner , does not seem to exist ( this study ) . Similarly , inbreeding avoidance is absent in this species when birds can only judge genetic similarity per se [61] ( although it does take place when siblings are familiar with each other [69] ) . Our results are consistent with the hypothesis that behavioral compatibility between the pair members leads to benefits of mate choice . This could come about through different mechanisms: the emerging behaviors of a pair in terms of coordination or complementarity [23 , 24 , 27–29] , and/or the individual-specific stimulation of a partner’s sensory system leading to a greater investment in reproduction [32–34] . Currently it is unclear which of these factors leads to the observed variation in parental care compatibility , and it is also unclear to what extent there is a genetic basis for this variation in compatibility . In the following , we discuss our exploratory analyses on fitness components and behaviors of “free-choice” and “forced” pairs , to provide testable ideas about how such behavioral compatibility benefits could arise . We found that non-chosen pairs ( 1 ) more often had clutches with infertile eggs , ( 2 ) had more offspring dying at an early stage ( presumably from starvation ) , and ( 3 ) tended to have more eggs that disappeared ( presumably due to poorer care and nest defense ) . These effects on components of fitness may be due to differences in the behavior of chosen and non-chosen pairs . The most prominent behavioral differences were that ( a ) females with assigned partners responded less positively to within-pair courtship and they tended to copulate less frequently with their partner , and ( b ) males with assigned partners showed poorer nest attendance during the egg hatching period . The females’ reduced tendency to participate in within-pair courtship and copulation when in a “forced” pair may explain the higher incidence of infertile eggs . Indeed , in a previous experiment in which continuous video recording allowed us to witness about 80% of all copulations over a 4-mo period ( partly reported in [53] ) we found that the probability of laying an infertile egg declined significantly with the number of copulations witnessed during the 10 d prior to egg laying ( p = 0 . 04 , n = 376 eggs laid by 31 females , estimates: 27% infertile at 0 copulations versus 15% infertile at the median of 5 copulations; underlying data can be found in S1 Data ) . Alternatively , apparently infertile eggs may in fact represent cases of very early embryo mortality . This seems unlikely because egg fertility scores in zebra finches were tightly linked to the number of sperm that reached the egg [70] . Likewise , the lower nest attendance during hatching by males in non-chosen pairs could indicate a reduced motivation to care for the young or defend the nest when in a forced partnership , leading to greater offspring mortality and egg loss . Consequently , the results of these exploratory analyses further support the behavioral compatibility hypothesis . If males and females in “forced” pairs indeed invest less in reproduction ( copulation or care ) , as our results suggest , the question remains why . Reduced investment by members of “forced” pairs could be a long-term effect of a single stressful event ( trauma ) , namely the loss of the chosen partner ( an event that could also happen in the wild due to predation [71] ) . This explanation seems unlikely , however , because fitness was affected by the treatment per se and not by the number of partner losses experienced by an individual ( see scheme in “Methods” ) when both factors were fitted within one model ( males: treatment p = 0 . 02 , number of mate losses p = 0 . 63; females: treatment p = 0 . 06 , number of mate losses p = 0 . 35 ) . Alternatively , being forced to breed with a non-preferred partner ( unlikely to occur in the wild ) might cause chronic stress . Being chronically stressed when paired to a specific partner A but not when paired to partner B would be part of the “phenotypic incompatibility” phenomenon . Our score of “pair harmony , ” which was based on affiliative and sexual behaviors , as well as behavioral synchrony and the tendency to reunite , did not significantly correlate with pair fitness . A study on zebra finches in the wild reported that behavioral synchrony was associated with brood size [25] , but further experimental work suggested that variation in synchrony might have been the consequence and not the cause of variation in reproductive success [72] . Evidence supporting the idea that pair coordination is important mainly comes from studies showing an increase in breeding success with pair bond duration ( [27–29 , 73 , 74] but see [75] ) . We specifically designed our experiment to create variation in pair-bond duration ( pairs stayed together for one or two breeding periods ) . However , this covariate did not have an effect on any of the fitness components ( mostly showing non-significant trends opposite to expectation , Table 2 ) and was therefore removed from most final models . This suggests that behavioral compatibility ( with synergistic effects on fitness ) did not increase with pair bond duration . The only traits that were affected by pair-bond duration were those related to extra-pair behavior ( Table 2 ) : females responded less positively to extra-pair courtships and received fewer extra-pair courtships with increasing pair-bond duration . In contrast , male courtship rate towards extra-pair females increased with pair-bond duration . In other words , it seems that females decreased and males increased their promiscuous behavior . It has been suggested that individuals choose each other based on their respective personality , which would determine their behavioral compatibility [22] . Individuals that show similar behavioral types , or similar plasticity ( and therefore predictability ) , could be better at negotiating or coordinating their actions , and could therefore have reduced conflicts over parental care and higher reproductive success ( [26 , 76 , 77] but see [78] ) . So far , besides observational studies [76 , 79 , 80] , only two experiments ( both conducted on zebra finches ) aimed at testing this hypothesis , and none of them found consistent evidence for pair combination effects on rearing success , based on any of the personality traits measured [26 , 78] . We did not measure any personality traits of individuals prior to the experiment , because we did not have clear a priori predictions about the advantages of being behaviorally similar . Instead , we scored the synchrony of activities during breeding , but this did not differ between treatment groups ( see S4 Text ) . Although an effect of lack of coordination between pair members cannot be excluded , our exploratory analyses suggest a reduced investment or commitment in individuals of “forced” pairs ( lower female within-pair responsiveness , higher male extra-pair courtship rate , lower male nest attendance ) . Previous experimental work on zebra finches shows that the amount of male singing activity can affect egg quality [33] . More generally , courtship and other affiliative behaviors , which may occur more frequently in chosen pairs , may affect the level of reciprocal stimulation [32 , 81 , 82] . Earlier studies that favored the genetic compatibility hypothesis cannot rule out that the treatment ( chosen versus non-chosen pairs ) affected maternal investment ( e . g . , egg quality ) with potential effects on offspring viability [35–37 , 43 , 46] . Artificial insemination would be needed to experimentally demonstrate that higher offspring viability arises from genetic compatibility and not from maternal ( e . g . , egg nutrients ) or paternal effects ( e . g . , sperm allocation ) following greater stimulation by a preferred partner ( see e . g . , [5] ) . If forced pairs reduced their investment in breeding together , as our analyses suggest , the question remains whether this behavior is adaptive . Reduced investment in current reproduction could be adaptive , if it saves resources for future reproduction with a better ( preferred ) partner . However , this explanation seems unlikely for a species such as the zebra finch , because life-long monogamy largely precludes breeding with a different partner in the future [52] . Moreover , in a follow-up experiment consisting of a third breeding season where all individuals could freely choose their mate , individuals could not compensate for the lower fitness previously obtained with a non-chosen partner ( see S6 Text ) . Therefore , the reduced investment in breeding by members of non-chosen pairs could be maladaptive , either because this never occurs under natural conditions ( because individuals are never forced to mate or breed with a particular partner ) , or because some constraints limit the adaptive behavioral flexibility of the animals . To conclude , chosen pairs had significantly higher fitness than forced pairs , apparently due to behavioral rather than genetic compatibility effects . The mechanisms behind such behavioral compatibility , in terms of willingness or ability to cooperate with certain individuals and in terms of coordination between partners need further study , in particular in the context of offspring provisioning . In humans , some studies suggest that individuals are more satisfied , more committed , and less likely to engage in domestic violence , when involved in a love-based rather than an arranged marriage ( [83 , 84] , but see [85] ) . The challenge there is also to find out whether stable and happy marriages result from motivation to cooperate ( and to identify what stimulates such feelings , see [86–89] ) , or from congruence in terms of partners’ intrinsic behavioral types [90] .
A scheme of the design with its timeline is depicted in Fig 4 . All experimental birds hatched in the summer of 2011 in large semi-outdoor aviaries . The origin of the birds ( population #4 in [91] ) , and rearing and housing conditions have been described in detail elsewhere [60] . This population has been derived from the wild only about ten generations ago [91] . Shortly after independence ( age 45 d ) , individuals were put into eight mixed-sex peer-groups of ten males and ten females . When birds reached sexual maturity ( 100 d old ) they were color-banded , and peer-groups were joined two by two ( yielding four groups , each allowing 20 possible pairs to form ) . Sixty-six pairs were identified during ad libitum observations in the winter of 2011–2012 . Mid-April 2012 , half of the identified pairs were randomly assigned to the treatment group NC ( in which all birds are assigned the partner of someone else: “non-chosen” ) , while the remaining pairs went to the treatment group C ( in which birds are allowed to stay with their chosen partner ) . In order to induce pair formation in the randomly created pairs of the NC group , these pairs were put into individual cages for a period of two months and were allowed to lay one clutch . Pairs from the C group also went to such cages and were allowed to lay one clutch in order to standardize all experiences apart from the re-pairing . On the 21st of May , three pairs of each treatment group ( chosen randomly but excluding the initial chosen partners of individuals of non-chosen pairs ) were put into a breeding aviary ( 10 replicates , 60 pairs in total ) . Both members of each pair had been previously color-banded on both legs with one random color out of six ( dark blue , light blue , black , yellow , orange , white ) , so that a pair would be unmistakably identifiable in its aviary . Forty-five pairs ( 26 C , 19 NC ) did not divorce and were considered for the analyses . After one week of intensive focal pair observations , we introduced nest material , and checked nests daily until 21 August , when the experiment was stopped and newly laid eggs were replaced by dummy eggs , but pairs were still allowed to raise all offspring from eggs laid before that . In October 2012 , once all offspring had reached independence , we assigned treatment groups for the second breeding season . First , we randomly selected eight pairs from each treatment group ( among the 26 C and 19 NC that were available ) that were allowed to stay with their partner throughout the second season . In this way , we could better separate the effects of choice treatment from the potential effects of pair-bond duration . These 16 pairs and all other adults ( remaining C , NC , and previously divorced birds ) were put into one of two big flocks , to allow a second round of choosing a partner . Each group contained 20 widowed males ( i . e . , their former breeding partner was in the other group ) , 20 widowed females , and eight established pairs . As a result , each widowed female could choose a new partner among a set of 20 new males , which never included her previous breeding partner ( but for half of the females from non-chosen pairs it did include again the initially chosen mate because this could not be avoided ) . In December 2012 , after pair identification and random assignment to treatment group ( without regard to previous treatment ) , pairs were put into cages for six months and allowed to lay two clutches . The longer period of force-pairing in cages resulted in a lower divorce rate compared to the previous season ( only one pair of each treatment divorced ) . On 21 May 2013 , pairs were put again into breeding aviaries and allowed to breed as in the previous year . Of 52 pairs identified in the winter groups , 42 ( 21 of each treatment group , across seven aviaries ) contributed to the second breeding period ( 12 birds died accidentally because food dispensers were blocked for 2 d in early March 2013 ) . The design itself and the accidental food shortage may have led to selection of the highest quality individuals . Although it did not induce bias ( selection was independent of treatment group ) , it can result in an underestimation of the real fitness benefits of mate choice . Furthermore , one member of a pair died for unknown reasons ( and its partner was removed ) within the first week of each breeding period ( 1 C in 2012 and 1 NC in 2013 ) , and these two pairs were excluded from the analyses . Each aviary contained 7 nest boxes . Every morning , all nests were checked , the individual ( s ) attending the nest identified , and the fate of each egg and each offspring noted . Unhatched eggs were opened when neglected by the parents ( for instance , after offspring had fledged ) and embryos were collected for parentage analysis ( using 11 microsatellite markers , following [92] and [93] ) . For the same purpose , small ( ~10 μl ) blood samples were taken from 8–10 d old offspring , or tissue samples if they died earlier . Of 1 , 434 eggs laid by all birds including divorcees , 28% ( n = 402 ) could not be assigned through parentage analysis , and were assigned to the social pair that attended the nest . These eggs included apparently infertile eggs ( 5 . 6% , n = 80 ) , and eggs that were buried in the nest and did not develop ( typically after a nest take-over by another pair ) or disappeared ( presumably they broke and were eaten by the birds ) ( 21 . 6% , n = 310 ) , as well as eight dead embryos and four dead hatchlings that yielded bad DNA samples . Relative fitness of an individual was calculated as the total number of genetic offspring produced in a given breeding period that reached independence ( age 35 d ) divided by the average number of offspring produced by all same-sex individuals of the same aviary that did not divorce . Each aviary was equipped with a dome camera set to record different aviary positions during each day of the week . During 3 d , we filmed an artificial tree , on which 69% of all courtships took place ( calculated from direct observations , described below ) . For one day , we recorded each of the two sets of nest boxes , and for 2 d , a set of perches on which individuals often allopreen . We analyzed the first hour of each day , when copulations are most frequent [53] . In all pairs considered for the analyses ( those that did not divorce ) , we recorded 1 , 942 within-pair ( WP ) and 2 , 999 extra-pair ( EP ) courtships ( in the latter , a divorced female or male may have been the extra-pair partner ) . For each courtship , we scored female responsiveness as follows: threat or aggression toward the male ( −1 ) , flying away ( −0 . 5 ) , mixed or ambiguous signs ( 0 ) , courtship hopping and beak wiping ( +0 . 5 ) , and copulation solicitation ( +1 ) , and noted whether it led to a successful copulation . We also conducted direct observations , following a protocol inspired by studies on cockatiels , Nymphicus hollandicus [21 , 23 , 94] . Observations were carried out both in the pre-breeding period ( first week after release into aviaries before nesting material was added ) and during the entire breeding period , to test whether pairs with greater behavioral compatibility before breeding ( as in [23] ) , or during breeding activities , would have greater reproductive success . The observer stood behind a one-way glass window ( built into each aviary door ) and carried out focal-pair watches by monitoring a pair for 3 min . During these watches we observed 613 WP and 800 EP courtships . We noted their location and whether they led to a successful copulation . For a subset of 561 WP and 782 EP courtships , we also scored female responsiveness , as described above . During focal-pair watches , we also recorded whether within-pair allopreening or aggression occurred during the 3 min period ( “yes” or “no” ) . Every 30 s , we recorded the distance between the partners and their activity . Distance was averaged for each 3 min watch . Activities were split into nine categories: feeding , cleaning , nesting or parental behavior ( nest building or attendance , and feeding of fledglings ) , sleeping , sitting , involved in aggression , involved in courtship , flying , and “other . ” We defined pair synchrony as the sum of the observations in which both partners engaged in the same activity ( range 0–6 ) . For each pair member , we also recorded all occurrences of an individual flying away from or back towards ( <50 cm ) its partner ( e . g . , female flying away: Faway , male flying back: Mback ) . From those counts , we calculated the tendency of the pair to reunite: ( Σ Fback + Σ Mback ) / ( Σ Faway + Σ Maway ) , and a mate guarding index: ( Σ Faway − Σ Fback ) − ( Σ Maway − Σ Mback ) . The latter is positive in case of male mate guarding , and negative for female mate guarding . The six pairs in an aviary were watched successively in a randomized predetermined order , and the time of observation of each aviary was randomized over the course of each day ( i . e . , from sunrise to sunset ) . In 2012 , pairs were watched 9–13 times ( median = 11 ) in the pre-breeding period , and 37–39 times ( median = 38 ) during the breeding period . In 2013 , 16–21 focal watches ( median = 21 ) per pair were performed during pre-breeding , and 68–70 ( median = 69 ) during breeding . For each pair , all measures were averaged for all focal watches separately for the pre-breeding and breeding period , because these periods were analyzed separately as planned a priori ( see Results and S4 Text ) . Male courtship rates ( WP and EP courtships per hour ) and best linear unbiased predictors ( BLUPs , i . e . , random effect estimates ) of female responsiveness ( to WP and EP courtships ) were also calculated ( see S5 Text ) for both periods and included in a principal component analysis ( PCA ) . All observations were done blind to the treatment of the birds . All statistical tests were conducted in R [95] . General and generalized mixed-effect models were performed with the “lmer” and “glmer” function of the lme4 package [96] and the PCAs with the “principal” function of the psych package [97] . All fixed effects were chosen a priori by considering ( a ) their biological relevance ( e . g . , hatching order when looking at offspring mortality ) , ( b ) their mathematical relevance ( e . g . , clutch size when looking at the presence of infertile eggs in a clutch ) , ( c ) the experimental design ( e . g . , year ) , and ( d ) consistency with previously published models ( e . g . , how to model the fertile period when looking at female responsiveness ) . P-values for general mixed effect models ( lmer ) were obtained from model comparison ( with and without the explanatory variable ) with the function anova in R; p-values for generalized mixed effect models ( glmer ) were taken from the model output ( calculated from z-values ) . | The last half century has seen a tremendous interest in the study of mate choice and the evolution of traits that make individuals attractive to others . In some species , however , individuals can differ substantially in who they find attractive , and this variation has typically been interpreted as “mate choice for compatibility . ” Here , we quantify the benefits of such mate choice in a socially monogamous passerine bird , the zebra finch . We found that pairs that resulted from free mate choice achieved a 37% higher reproductive success than pairs that were forced to mate with a randomly assigned individual . Forced pairs suffered from increased failure to fertilize eggs and from increased mortality of hatched offspring . In females , we observed a reduced readiness to copulate with the assigned partner , while males that were force‐paired showed reduced parental care and increased activity in courting extra‐pair females . These findings support the hypothesis that zebra finches choose mates on the basis of behavioral compatibility . In contrast , it appears that zebra finches have not evolved a mechanism that would allow them to select a partner with whom they could minimize the rate of embryo mortality . This argues against mate choice for genetic compatibility . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Fitness Benefits of Mate Choice for Compatibility in a Socially Monogamous Species |
The 1000 Genomes Project data provides a natural background dataset for amino acid germline mutations in humans . Since the direction of mutation is known , the amino acid exchange matrix generated from the observed nucleotide variants is asymmetric and the mutabilities of the different amino acids are very different . These differences predominantly reflect preferences for nucleotide mutations in the DNA ( especially the high mutation rate of the CpG dinucleotide , which makes arginine mutability very much higher than other amino acids ) rather than selection imposed by protein structure constraints , although there is evidence for the latter as well . The variants occur predominantly on the surface of proteins ( 82% ) , with a slight preference for sites which are more exposed and less well conserved than random . Mutations to functional residues occur about half as often as expected by chance . The disease-associated amino acid variant distributions in OMIM are radically different from those expected on the basis of the 1000 Genomes dataset . The disease-associated variants preferentially occur in more conserved sites , compared to 1000 Genomes mutations . Many of the amino acid exchange profiles appear to exhibit an anti-correlation , with common exchanges in one dataset being rare in the other . Disease-associated variants exhibit more extreme differences in amino acid size and hydrophobicity . More modelling of the mutational processes at the nucleotide level is needed , but these observations should contribute to an improved prediction of the effects of specific variants in humans .
With the release of the 1000 Genomes Project ( 1 kG ) data [1] , it has become feasible to study human protein variation on a large scale . The main aim of the 1 kG project was to discover and characterize at least 95% of human DNA variants ( with a frequency of occurrence of >1% ) found in multiple human populations across the world . Five main populations were sampled with ancestry in Europe , West Africa , the Americas , East Asia and South Asia . The project has provided a rich set of synonymous ( sSNPs ) and non-synonymous ( nsSNPs ) variants for 1092 individuals from diverse populations . It is estimated from the 1 kG data that each individual will , on average , differ from the reference human genome sequence at 10 , 000–12 , 000 synonymous sites in addition to 10 , 000–11 , 000 non-synonymous sites [1] . As these nsSNPs change the amino acid sequence of the protein , the changes have the potential to affect the structure and function of the corresponding proteins . The 1000 Genomes Project data set is valuable in that it is large and not derived from a disease cohort but rather seeks to capture variants found in a disparate set of healthy individuals . This can be used to characterise differences on average between disease-associated and benign mutations ( or at least mutations not known to be associated with disease ) as well as exploring their structural characteristics and preferences . The reports from the 1000 Genomes Consortium [1] , [2] have focused on genome and nucleotide variation , and other papers consider mutations in association with a specific disease ( e . g . cancer ) [3] . Various databases such as the Online database of Mendelian Inheritance in Man ( OMIM , [4] ) , the UniProtKB human polymorphism set ( Humsavar , [5] ) and the Human Gene Mutation Database ( HGMD , [6] ) collect information on inherited diseases associated with variants . The Humsavar database contains disease-associated variants from the literature and OMIM . OMIM currently contains information on approximately 10 , 200 nsSNPs associated with diseases ( December 2011 ) and Humsavar about 23 , 500 disease-associated nsSNPs . Most of the phenotypical effects and their molecular origins are not well established , so predicting the functional effect of a single amino acid variant is of great medical interest . The main methods assume that mutations in highly conserved residues cause disease and thus , by using alignments to homologous sequences and residue similarity , the severity of the variant can be gauged . More advanced methods include information derived from protein structures ( such as solvent accessibility , free energy changes , environment specific substitution tables and functional annotations ) to improve the accuracy ( see review by [7] ) . The advantage of using a 3D approach for prediction is that the consequence and characteristics of the variant can be studied in its specific environment in the protein . This provides a level of information beyond a sequence or a sequence alignment [8] . If there are ligands present , the interaction between the mutated amino acid and the ligand can be studied . This has been successfully applied to various individual proteins on a case-by-case basis [9] , [10] . In total over 30 different programs to predict the effects of these variants have been published , including Condel [11] , SNAP [12] , SDM [13] , PolyPhen [14] , VEP [15] , SIFT [16] , [17] and SNP&GO [18] . Most of these algorithms can only predict whether a specific variant will be neutral or deleterious for the protein with various degrees of accuracy , although measuring accuracy is challenging in the absence of a good benchmark . To allow the accurate prediction of functional effects of SNPs , we need a thorough understanding of why amino acids mutate in humans . Various groups have worked on the effect of the mutations and numerous studies have been done on small specific sets of proteins [8] , [19]–[22] . Blundell and co-workers have found that the local environment around an amino acid plays a large role in the effect that selection has on a mutation in a specific position [21] . This has led to the development of environment specific substitution matrices [23] , [24] that incorporate structural constraints . Subramanian and Kumar [25] did a detailed analysis on a set of 8 , 627 disease-associated mutations and found that disease-associated mutations tend to occur on inter-species conserved residues . The common factor between these studies is that they try to understand the effect that selection and structural constraints have on disease vs non-disease states in selected sets of proteins . Very few studies have tried to unravel the underlying cause for mutation patterns seen in human proteins . With this work we aim to elucidate why certain amino acids mutate more and try to understand the underlying mechanisms present in the mutation process . We gather the data for all the amino acid mutations found in the 1000 Genomes Project to characterise their sequence and structural properties , providing a benchmark background against which to compare the disease-associated nsSNPs in OMIM and Humsavar .
Figure 1 shows the amino acid exchange matrix generated from the ∼106 , 000 nsSNPs found in the 1 kG data . Amino acid mutations requiring two or three base changes are not defined in this dataset due to technical reasons . The 1 kG matrix exhibits several interesting features , most of which reflect the genetic code and the differential mutability of various codons . All possible single base changes are observed . The matrix is not symmetrical as a result of the differences in frequency of occurrence of amino acids as well as differences in their mutabilities [26] , [27] . As expected there is a strong correlation ( r = 0 . 786 ) between the frequency of occurrence of amino acids in the human proteome and the number of associated codons . Figure 2 shows that , excluding Arg and Leu which are extreme outliers , there is a strong trend for amino acids with a higher frequency of occurrence to have more mutations ( r = 0 . 836 ) . Taken together this leads to a relatively strong correlation ( r = 0 . 741 ) between the number of codons and the number of mutations . In contrast , the frequency of the gained amino acids , resulting from the mutation , shows little correlation between frequency of occurrence and number of mutations ( r = 0 . 349 ) . The mutabilities of the amino acids ( see methods ) in the 1 kG dataset are shown in the last column of Figure 1 . Arg ( 0 . 031 ) is the most mutable , whilst the more chemically complex amino acids , Trp ( 0 . 004 ) and Phe ( 0 . 005 ) have the lowest mutabilities . There is no correlation in the 1000 Genomes data between mutability and frequency of occurrence ( r = −0 . 003 excluding Arg ) nor between mutability and the number of codons ( Figure 3 ) . It is well known that CpG dinucleotides in DNA tend to mutate at rates 10–50 times higher than other dinucleotides [28] , [29] and thus amino acids with a CpG present in their codons will mutate with a higher probability ( see Figure 4 ) . Four out of the six codons for Arg include CpG sequences , and Arg mutates more frequently than any other residue , with a mutability ( 0 . 031 ) which is over twice as high as its nearest rival . This high mutability also reflects the fact that the CpG in the Arg codons occur in the non-wobble positions so nucleotide mutations give rise to non-synonymous SNPs . In contrast Leu which also has six codons , none of which contain CpG , has a low mutability ( 0 . 005 ) and mutates six times less frequently than Arg . However the correlation with CpG is far from perfect and other factors must have an effect . For example , Met , which has only one codon with no CpG dinucleotide , is the second most mutable amino acid ( 0 . 014 ) . Figure 4 shows the clear pattern of amino acid gain and loss in the human proteome . Jordan [26] and Zuckerkandl [30] long since identified that Cys , Met , His , Ser and Phe are being accrued significantly in the human proteome . Our data confirm a net gain of these five amino acids , and Val , Asn , Ile and Thr were also confirmed as weak gainers . Jordan and co-workers also identified strong losers and our data again confirm that Pro , Ala , Gly and Glu are strong losers . Lys was identified as a weak loser but our larger dataset suggests that lysine should be considered a weak gainer in humans . Arg is the strongest loser in the human genome ( similar to the human set in [26] but not other considered species ) . We calculated the mutability for every amino acid on a population specific basis . None of the populations showed a different pattern of amino acid mutabilities , compared to the overall trend with correlation coefficients equal to 1 . 0 ( Figure S1 ) . Using the individual amino acid mutabilites , we looked at aggregate protein mutability differences by adding up the individual mutabilities for every amino acid in each protein in the data set and normalising by protein length . This was compared to the aggregate mutabilities of proteins involved in disease as classified by OMIM and Humsavar . The average score for disease-associated proteins was 0 . 0103 and for non-disease proteins 0 . 0102 with a median of 0 . 01022 ( σ = 0 . 0006 ) and 0 . 01018 ( σ = 0 . 0005 ) , respectively , indicating that protein aggregate mutability has no bearing on disease-association ( Figure S2 ) . As well as constraints on the mutational process at the DNA level , the consequence of a variant on the protein structure and function will also have an impact on the number of observed mutations . If a variant interferes with the structure and function of a protein and that protein is essential , then this variant is less likely to be seen . However comparison of mutability with the size and hydrophobicity of the amino acid shows very little correlation in the 1 kG dataset . There is a moderate anti-correlation between higher mutability and size ( r = −0 . 474 ) , with the smaller amino acids mutating more frequently , but no correlation at all between mutability and hydrophobicity ( r = −0 . 082 ) although the large hydrophobic amino acids ( Leu , Phe and Trp ) have the lowest mutability scores . Trp has the fewest mutations ( 544 , even though all SNPs in Trp codons result in a change of amino acid ) and also the lowest mutability score ( 0 . 004 ) together with Phe . In addition to their complexity and low abundance , Phe and Trp often occur in specialized roles such as the interior of proteins , π-π stacking or ring interactions and this might add to their low mutability . The mutability of Cys is also low , perhaps reflecting its role in disulphide bridges , which help to stabilise extracellular proteins . To investigate the structural characteristics of these variants , three sets of protein structures were compiled , namely the 3D set , the monomer set and the model set ( Table 1 ) . The 3D and monomer set were constructed from data in the PDB ( see methods ) while the model set and the subsequent variant modelling was created and performed using Modbase [31] and Modeller [32] , built into an in-house homology modelling pipeline . The 3D set contains 2 , 139 protein chains . A total of 10 , 628 1 kG nsSNPs were found in these chains , of which protein models , based on the known structures of human proteins could be built for 5 , 524 . The monomer set contains 325 protein chains identified as monomers and a total of 1 , 461 1 kG nsSNPs were found , of which 897 could be modelled . The model set , including models based on homologues from the PDB , contained 2 , 630 protein chains and 12 , 432 out of 13 , 037 nsSNPs could be modelled . For the Humsavar set we found 5 , 592 nsSNPs of which 3 , 942 could be modelled . Figure 5A shows a comparison of the solvent accessibility distribution for all residues compared to that for the variants . On average the variants in the 1 kG are slightly more exposed . An analysis of the solvent exposed residues found that , for the most accurate monomer set , 79% of nsSNPs are solvent exposed compared to 73% of all residues ( p = 0 . 001 ) . For the structures in the model set , 81 . 9% of nsSNPs were solvent exposed . For all three datasets , the 1 kG variants have a slight preference to occur on the surface of proteins compared to all residues . Figure 5B shows that there were no appreciable differences in secondary structure preferences between variants and other residues . Functional annotation for each human protein was derived using SAS ( Sequence Annotated by Structure , [33] ) . Table 2 shows the different functional annotations for each set . The vast majority of functional annotations identified , make contacts to ligands ( using PDBsum data , [34] ) or site interactions in the proteins ( as defined in the PDB ) . Only 15 . 5% of the mutations ( 1 , 648 of 10 , 628 ) in the 3D set were annotated with a function compared to 29 . 1% of all residues in the set of human structures ( Figure 5C ) . These data show that the observed mutations in the 1000 Genomes occur less frequently in the functionally annotated residues compared to all residues . Residue conservation scores , defined as the variation of the residues at a given site in the protein across multiple species , were obtained for all sites in the human proteome ( where sufficient data are available ) from the Evolutionary Trace server [35] . These scores are distributed across the whole range of conservation ( Figure 6 ) with a mean score of 0 . 48 . The scores for all the sites with mutations in the 1000 Genomes data show a slightly different distribution from all residues , with a small but significant shift ( p<2 . 2×10−16 ) towards the less conserved sites and a reduced mean conservation score of 0 . 43 . Clearly natural variation occurs across all conservation levels and is not limited to non-conserved residues . For each amino acid the mutation profile can be calculated showing the preference for specific X = >Y mutations in the 1000 Genomes data . These profiles , given for all the amino acids in Figure 7 , show that there are striking differences in frequency of occurrence for the different exchanges . For example , in the 1 kG set Arg shows a strong preference to mutate to Gln and His , whilst mutations to Ser , Gly and Pro are much less frequent . All the amino acids show these differential exchange rates . Figure 8A shows the distribution of changes in energy of the whole protein caused by each mutation , evaluated as the statistical potential energy DOPE score ( Discrete Optimised Protein Energy ) in Modeller . 68 . 1% of the 1 kG variants increase the DOPE score ( i . e . make the protein less stable ) . This implies that most natural variants decrease the stability of the protein , albeit by a very small amount . The distribution of changes in size and hydrophobicity for all observed mutations ( Figure 8B and 8C ) show that 59 . 4% of mutations increase the hydrophobicity of the amino acid and 52 . 4% of mutations increase the size . Over 84% of variants change their size by less than 50 Da . 72% of variants change their hydrophobicity by less than 1 unit . Extreme changes are rare . At this stage these observations provide empirical expectation rates for amino acid exchanges in humans and result from the genetic code , the nucleotide exchange rates and also some selection at the protein level . However without a good random model it is difficult to be confident about the importance of the different contributions to such variation . The 1 kG counts matrix is a snapshot of mutations that have occurred in humans in a short period of time . To understand this process the count matrix can be converted into an instantaneous rate matrix describing the rates of change of each amino acid in humans in a time-independent manner [36] . Instantaneous rate matrices have previously been built from a wide selection of protein alignments across many species including nuclear proteins , mitochondrial proteins , chloroplast proteins , buried protein domains and exposed protein domains . PCA can be used to compare these inter-species matrices with the 1 kG intra-species matrix ( Figure 9A–C ) . The 1 kG matrix was built using data where the direction of the mutations is known whereas all other matrices were calculated assuming direction is unknown . This was compared to the WAG [37] and PAM matrix [38] . To check that any differences between the 1 kG matrix and the other matrices are not caused by using direction , a directionless matrix has also been included in the plot ( Figure 9D ) . In this plot , principal component one clearly separates the 1 kG matrices , which are placed very close together , from all of the previously calculated matrices . Principal component two then spreads matrices out based on whether the alignments used to build them are made up mainly of exposed or buried domains , with the mitochondrial matrices at the one extreme built from nearly all membrane proteins , and matrices built from only exposed regions of proteins at the other . A difference between the intra-species data and the inter-species matrices is the amount of selection which has occurred . Due to the time-scale for the 1 kG data and the relatively weak selection in human populations [39] , [40] the only mutations which are not observed are lethal mutations . This means that there should be a limited effect of selection on the 1 kG matrix . By using no allele frequency cutoff for the minor alleles when building the count matrix , we gather the maximum amount of information about the mutation process . The counts are necessarily shaped by mutation and selection but will mostly reflect the mutation process . The inter-species matrices ( e . g . PAM and WAG in Figure 9B , C ) on the other hand are subject to selection pressures . This could explain why the 1 kG matrix is so different from the other matrices . One clear factor is CpG hypermutability: for example , changes from Arg , an amino acid with four of six codons containing a CpG , have a very high rate in the 1 kG data , and not in WAG ( Figure 9A , B ) . In fact only codons containing a CpG have high rates overall ( Figure 10 ) . The most plausible explanation is that these CpG mutations are occurring at a very high rate and then are selected out so that the effect is not seen as strongly when looking across multiple species . For comparison , we have constructed the amino acid exchange counts matrix for data from the OMIM database and the associated plots for these mutations ( Figures 1–8 ) . Disease variants from the UniProtKB/Swiss-Prot Human polymorphisms and disease mutations index ( Humsavar ) were also included with plots available in the supplement ( Figures S3 , S4 , S5 ) . Our focus however is on the OMIM set . In contrast to the 1 kG data , various double and triple base mutations are observed in the OMIM set , however the three triple base changes ( Phe-Lys , Met-Tyr and Trp-Ile ) were checked back to the publications and all were found to be errors either in the paper or in OMIM and were removed . 82 two base changes were found in OMIM and a few ( 10% ) randomly selected changes were manually checked with no errors found . Clearly the OMIM data are radically different from the 1000 Genome data , in that they are all independent observations of variable confidence and manually determined by individual scientists . They only represent a small fraction of disease-associated nsSNPs and the number of mutations ( ∼10 , 000 ) , is approximately ten times smaller than the number of 1000 Genomes mutations . The normalised OMIM counts that differ from the 1 kG dataset are coloured in Figure 1 . Considering just the residue type , if we exclude Arg , the overall correlation between the normalised frequencies of occurrence of the mutated residues in the two datasets is only 0 . 14 and between 1 kG and Humsavar it is 0 . 48 . If we compare all 148 observed X = >Y frequencies , the correlation between 1 kG and OMIM is 0 . 51 and 1 kG and Humsavar is 0 . 79 . Previous studies have found that mutations from Arg and Gly are the major contributors to human genetic disease and have been shown to make up about 30% of the mutations involved in disease [41] . In this updated and much expanded set , variants from Arg and Gly only make up 15% of the disease causing mutations . However mutations to Arg are still the biggest contributor to genetic disease with ∼19 . 4% of all mutations . Figure 11 shows a rank order comparison between the frequency of occurrence of the 1 kG and OMIM variants ( r = 0 . 09 ) as well as between 1 kG and Humsavar ( r = 0 . 31 ) and Humsavar and OMIM ( r = 0 . 51 ) , normalised for amino acid occurrence . Unlike for the 1 kG data , the disease-associated variants show moderate inverse correlations between their frequency and the frequency of occurrence of the residue type ( r = −0 . 67 ) implying that , at least for OMIM , the mutations to the rarer amino acids ( with fewer codons ) are more likely to be associated with disease . As with the 1 kG data there is no strong correlation between a residue type being associated with a disease in the OMIM data and the number of codons . For hydrophobicity and size , the disease associated variants show the opposite trend to the 1 kG dataset with a moderate correlation between lower frequency and smaller size ( r = 0 . 528 , excluding Cys and Trp ) but no correlation between frequency and hydrophobicity ( r = 0 . 289 ) . It is interesting to note that the least mutable amino acid in the 1 kG data ( Trp ) turns out to be the residue whose mutation is most likely to result in disease in the OMIM variants and is highly ranked in the Humsavar set . Trp , the largest amino acid , often occurs in specialized roles in proteins as does Cys , the second most frequent variant residue type in OMIM . Amino acids with a lower frequency of occurrence tend to be the more complex amino acids and are frequently found in specialized roles . Mutating them will result in the possible loss or alteration of protein function , hence the over-representation in OMIM and Humsavar . In a number of cases the OMIM and 1 kG variant preferences appear to behave in an opposite way from one another e . g . in Figure 7 Arg most frequently mutates to Gln in the 1000 Genomes and a variantion to Gly is much less common , whilst Arg to Gly is the most common variant in the OMIM dataset and a variation to Gln is rare . We observe a reasonable correlation between the OMIM and Humsavar mutabilities ( r = 0 . 51 ) , but some amino acids appear to behave completely differently in the two datasets . Gly and Ala are much more frequently mutated in the Humsavar set than in OMIM , whilst Gln , Lys and His have mutabilities in the Humsavar set similar to those observed in the 1 kG dataset and much smaller than those in OMIM . This may reflect the larger Humsavar dataset ( but this seems unlikely since Gly and Ala are quite common amind acids ) , so these specific discrepancies may rather reflect the origins of mutations in the two separate datasets . The disease-associated OMIM variants show a slight preference for buried sites ( 33% ) compared to all residues ( 27% ) in the human proteome ( Figure 5A ) is even stronger in the Humsavar data ( 41% ) . This contrasts with the ‘natural’ variants of the 1 kG data , which show a decreased preference ( 18% ) for the interior . Our work broadly agrees with a smaller study done by Gong and Blundell [21] that showed 60–65% of disease associated nsSNPs are solvent exposed . We found an almost identical distribution of OMIM and Humsavar variants compared to all residues and the 1 kG variants between the different secondary structures ( Figure 5B ) . Figure 8A shows the differences in the DOPE scores [42] calculated for each variant during the structural modelling process for the 1 kG , OMIM and Humsavar datasets . The distribution for the disease-associated variants is shifted towards larger positive energies in both datasets , indicating that the variants destabilize the protein slightly more than the non-disease variants . In contrast to the 1 kG data , OMIM mutations are more likely to increase polarity ( 54% ) and more likely to decrease size ( 51 . 6% , Figure 8B , C ) . The two datasets show some detailed differences in size and hydrophobicity changes . The Humsavar variants less frequently reduce size or decrease hydrophobicity compared to OMIM mutations . In the OMIM set , 11 . 2% ( 209 of 1 , 864 ) of the modelled mutations were annotated with a function ( Figure 5C and methods ) . This is less than the distribution for all residues ( 29 . 1% ) and that seen for the 1 kG variants ( 15 . 5% ) . For the Humsavar data this drops to only 6 . 5% . This is a surprising finding , which needs further validation . It implies that most disease-associated mutations do not have a direct effect on the proteins' catalytic or binding sites but instead act through other , unannotated residues such as those which affect overall structure and stability or are involved in as yet unidentified protein-protein interfaces . There is a clear difference in the conservation score distribution between natural variants and the OMIM and Humsavar variants ( Figure 6 ) . The natural variants occur across the entire range of conservation but the OMIM and Humsavar variants show a peak in the more conserved residues . This is consistent with the idea that mutations in conserved residues often lead to disease .
The results presented herein are subject to a few caveats , the most serious being related to the limited and possibly biased disease-associated data in OMIM . There are only ∼10 , 000 variants in our OMIM set and these have variable experimental validation , and may indeed be biased according to scientists' preconceptions that such mutations should correspond to the residues that are most conserved and the amino acid exchanges that generate the largest changes in physicochemical characteristics . The Humsavar set has over 23 , 000 disease variants , however the requirements for inclusion are based on an annotation of ‘involvement in disease’ . This annotation is derived from either OMIM annotations or associations found in literature during curation of the SwissProt data . Notwithstanding , the OMIM dataset is one of the best available at the present time , although the coming years will see major expansion and hopefully improvements in such data . The results highlight the complex interplay of features from the level of the DNA up to protein sequence and structure . The codon CpG dinucleotide content plays a large role in determining which amino acids mutate . This in turn affects the mutability of amino acids and a clear difference was seen between non-disease and disease variants where amino acids that are naturally very mutable , show the opposite trend in the disease-associated data . The data for the 1000 Genomes provides a new experimental baseline against which amino acid profiles may be compared . Although there might be sequencing biases due to the DNA sequencing techologies used [43] , every effort has been made by the 1000 Genomes consortium to correct for this . They estimate that using consensus calling on data produced by multiple platforms results in an error rate of 1–4% , thus having a small but negligible impact on our results . The current results show evidence for some protein selection , mainly in that the variants occur slightly more often on the surface of the protein and are much less likely to be annotated as functional than expected by chance . However , we should note that even the best definition of functional , taken from structural data , is limited . At one level , the definition is rather broad . For example , all residues in contact with a ligand are described as functional , but this is a major underestimate since many cognate ligands are not present in the crystal structures and similarly protein-protein interactions are rarely captured . In addition there are still relatively few complete structures for human proteins , which makes analysis of the effects of variants more difficult . Even with these caveats , it is clear that the 1 kG variants eschew functional residues as defined here , a trend which is surprisingly even stronger in the OMIM and Humsavar data . The preference for OMIM mutations to be more buried and less functional supports the suggestion that these variants predominantly affect the structure and stability of the protein [4] . This is a similar result to that found by Sunyaev and co-workers [44] on a much smaller set . They found that 35% of disease variants were buried and a more detailed analysis found that ∼70% of the variants are located in structurally and functionally important regions . Therefore these disease-associated mutations may well target residues that are remote from the active site , which modulate rather than obliterate the function of the protein . For example , for an enzyme , the primary catalytic residues are rarely targeted , but the ‘secondary’ residues in the interior ( which affect stability ) or on the surface , which may affect protein-protein interactions , could modulate function . However , the higher than average conservation scores for OMIM and Humsavar sites suggest that these disease-associated residues , although not defined as ‘functional’ , are still important for the organism . This needs further investigation , with particular attention to how ‘functional’ residues are defined and whether we can improve on this definition . Bringing together all the above observations for disease-associated and natural variants in ∼1000 humans , we observe that the mutability of amino acids is largely driven by the properties of the DNA and mutational mechanisms , which favour mutations at codons containing a CpG dinucleotide . Therefore mutations to Arg residues are more than twice as common as any other mutation . However there are clearly other factors at play , which determine the frequency of variants , even at the DNA level . Although the disease-associated variants ( both OMIM and Humsavar ) follow the same pattern as the 1 kG variants ( i . e . the same mutations are present in both sets , as dictated by the genetic code ) , the rank order of amino acids , according to their probability of being disease-associated , is radically different from that expected on the basis of the 1 kG data , with some of the rarer amino acids being shifted to the top of the list . There is a small but significant impact of the protein structure on amino acid mutability , so that natural variants occur slightly more often in non-conserved regions . 59 . 4% of variations increase the hydrophobicity of the amino acid and 52 . 4% increase its size in the natural set , while OMIM variants often result in larger changes in the size and hydrophobicity of the amino acid and are more destabilising on average than 1 kG variants . The Humsavar data supports this idea that disease variants result in more extreme changes . The selection pressures captured in the WAG and PAM matrices ‘purify’ out the ‘natural’ variants , removing variants with large changes in size and hydrophobicity . The amino acids all show distinctive exchange profiles , whereby some exchanges are very common and some very rare , which provides an empirical expectation for any specific exchange in humans . As the cost of sequencing drops rapidly , many more genomes will be sequenced and experimental validation of disease-causing mutations will improve as a result of more data . Much better codon-based models of evolution will be attainable , allowing in turn a better dissection of the impact of selection at the protein level . The data herein will be used to develop an improved method to predict the effects of individual mutations , to explore cancer-related amino acid mutations , to investigate and compare mutational profiles in different organisms as well as improving codon mutation models for human DNA .
UniProt [5] was queried for all reviewed protein sequences belonging to Homo sapiens . 19 , 058 entries were retrieved . The Ensembl transcript ID [45] was obtained for each protein sequence using the mapping provided by UniProt ( 17 , 708 UniProt entries were mapped to 40 , 351 Ensembl transcript IDs ) . Immunoglobulins and major histocompatibility complex proteins were excluded as they are inherently variable . For every protein , the Ensembl v67 Perl API was used to query the transcript ID in Ensembl for nsSNPs found in the 1 kG data set ( as available on 1 August 2012 ) . To reduce the inherent uncertainty involved in determining the ancestral allele , only mutations that occurred in one of the 1000 Genomes described populations were used , with the allele present in all populations considered the ancestral , hence defining the direction of the mutation . This increases the chances that the variant found in the 1 kG data is a mutation away from the ancestral genome . 106 , 311 mutations were found and this data set , containing the ‘natural’ variants found in the 1 kG project , will be referred to as the 1 kG set . Residue conservation scores for each residue in every protein sequence were calculated using the Evolutionary Trace server [35] . Conservation scores for 2 , 274 sequences could not be calculated due to the methodology used by the Evolutionary Trace server that disregards residues in columns of the multiple alignment containing more than 60% gaps and ranked as being non-conserved , as well as residues judged by the algorithm not to have enough information . This process almost certainly preferentially excludes surface residues ( where insertions and deletions are most common ) but since we are using the conservation distribution for comparisons , this bias is not significant . The UniProt sequences were used to calculate the relative abundance of amino acids in human proteins . A total of about 10 . 5 million amino acids were counted . For each protein sequence , the OMIM Mutations search tool ( http://www . bioinf . org . uk/omim ) was queried with the UniProt entry ID to retrieve variants found in OMIM . Only variants for which the correct amino acid position in the protein has been verified , were used for the OMIM data set and will be referred to as the OMIM set . 556 of the OMIM mutations were found in the 1 kG set ( 0 . 5% ) . Although these represent a very small fraction we removed them so that they did not bias the results . The instantaneous rate change matrices were derived using the DCFreq method [36] and the human proteome frequencies . A mutability score for every amino acid was calculated by taking the total number of mutations for a specific amino acid in the data and dividing by the frequency of occurrence for the specific amino acid in the human genome . The proportional representation of each amino acid in the human proteome is given in supplemental Table S1 . We compared the amino acid variant counts in the 1 kG and OMIM data using Fischer's exact test in the R package ( R Development Core Team , 2011 ) . Multiple comparison correction was done on the p-values for each amino acid using p . adjust in R with the Benjamini-Hochberg-Yekutieli method [46] , [47] . P-values lower than 0 . 01 were considered statistically significant . For correlation values , r>0 . 7 and r<−0 . 7 were considered strong , 0 . 4<r<0 . 7 and −0 . 4>r>−0 . 7 were considered moderate and 0 . 3>r>−0 . 3 weak or no correlation . The protein structure data set was constructed by first taking all the above mentioned protein sequences and annotating each with their respective Pfam [48] domains . Only proteins for which there were matching entries in the Protein Data Bank ( PDB , [49] ) were kept . This resulted in a list containing the UniProt identifiers for all known human proteins that have at least one structure in the PDB . For accuracy , the corresponding PDB structures were then filtered to include only X-ray structures . Using the Pfam mapping , only protein structures containing all the protein's Pfam domains were kept . The final list contained 2 , 139 protein chains and will be referred to as the 3D set . A set consisting only of human monomeric proteins was also constructed . An algorithm was implemented whereby a protein was classified as being either a multimer or a monomer based on a majority vote . The predictions used were from PISA [50] , UniProt , 3DComplex [51] , PIQSI [52] , PQS-PITA [53]–[55] , relevant PubMed abstracts and REMARK 350 records from the PDB structure file . The oligomeric predictions from each of the servers were collected for every protein in the 3D set . Only when the majority of the servers agreed on the most probable oligomeric state of the protein , was it designated as either a multimer or a monomer . The monomeric protein list contained 325 proteins and will be referred to as the monomer set . Another homology-based set was constructed using the human models in ModBase [31] . Models with 90–100% sequence identity and coverage were used as templates . This set contained 2 , 630 models and will be referred to as the model set . Each protein chain in the 3D , monomer and model sets was annotated with information from various databases and online resources . Information about protein properties such as catalytic residues , metal-binding residues , ligand-binding residues and PROSITE patterns [56] were extracted from PDBsum [34] and additional functional residue annotations were retrieved using SAS ( Sequence Annotated by Structure , [33] ) . The 3D coordinates for each of the proteins in the structure data sets were retrieved from the PDB . To maintain consistency between the PDB and UniProt residue numbering , the SIFTS mapping [57] for each protein chain was used . NACCESS was used to calculate the relative solvent accessibilities for the individual residues in a chain . A cut-off of 5% solvent exposure was used to distinguish between buried and exposed residues . To investigate the effect a nsSNP might have , each individual nsSNP was mapped to its correct amino acid in the protein structure . For every such nsSNP that could be mapped , a homology model of the protein containing the nsSNP was built using Modeller 9v3 [32] with the original protein structure serving as the template . A maximum of 200 steps of conjugate gradient minimization followed by 200 rounds of molecular dynamics at 300 K ( using Modeller ) was applied to each variant and its structural context analysed . NACCESS was run on all the variant models to identify changes in solvent accessibility . Comparisons of the Modeller DOPE score ( Discrete Optimized Protein Energy , [42] ) were made between the nsSNP model and the reference structure to estimate the magnitude of change that a variant might cause . The 1 kG models are available in PDBsum ( http://www . ebi . ac . uk/pdbsum/ ) by looking at the specific PDB code of interest . | In this paper we compare the differences between ‘natural’ and disease-associated amino acid variants at both sequence as well as structural levels . We used data from the 1000 Genomes Project ( 1 kG ) , the OMIM database and UniProtKB Humsavar . The results highlight the complex interplay of features from the level of the DNA up to protein sequence and structure . The codon CpG dinucleotide content plays a large role in determining which amino acids mutate . This in turn affects the mutability of amino acids and a clear difference was seen between non-disease and disease variants where amino acids that are naturally very mutable show the opposite trend in the disease-associated data . The current results show evidence for some selection , mainly in that the variants occur slightly more often on the surface of the protein and are much less likely to be annotated as functional than expected by chance . However we should note that even the best definition of functional , taken from structural data , is limited . Even with these caveats , it is clear that the 1 kG variants eschew functional residues as defined here , a trend which is surprisingly even stronger in the OMIM data . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2013 | Amino Acid Changes in Disease-Associated Variants Differ Radically from Variants Observed in the 1000 Genomes Project Dataset |
Omics data integration is becoming necessary to investigate the genomic mechanisms involved in complex diseases . During the integration process , many challenges arise such as data heterogeneity , the smaller number of individuals in comparison to the number of parameters , multicollinearity , and interpretation and validation of results due to their complexity and lack of knowledge about biological processes . To overcome some of these issues , innovative statistical approaches are being developed . In this work , we propose a permutation-based method to concomitantly assess significance and correct by multiple testing with the MaxT algorithm . This was applied with penalized regression methods ( LASSO and ENET ) when exploring relationships between common genetic variants , DNA methylation and gene expression measured in bladder tumor samples . The overall analysis flow consisted of three steps: ( 1 ) SNPs/CpGs were selected per each gene probe within 1Mb window upstream and downstream the gene; ( 2 ) LASSO and ENET were applied to assess the association between each expression probe and the selected SNPs/CpGs in three multivariable models ( SNP , CPG , and Global models , the latter integrating SNPs and CPGs ) ; and ( 3 ) the significance of each model was assessed using the permutation-based MaxT method . We identified 48 genes whose expression levels were significantly associated with both SNPs and CPGs . Importantly , 36 ( 75% ) of them were replicated in an independent data set ( TCGA ) and the performance of the proposed method was checked with a simulation study . We further support our results with a biological interpretation based on an enrichment analysis . The approach we propose allows reducing computational time and is flexible and easy to implement when analyzing several types of omics data . Our results highlight the importance of integrating omics data by applying appropriate statistical strategies to discover new insights into the complex genetic mechanisms involved in disease conditions .
Integrating different omics data types , such as genomics , epigenomics and transcriptomics , may provide a new strategy to discover unknown genomic mechanisms involved in complex diseases [1–3] . In cancer , tumor initiation and progression are the consequence of alterations in multiple pathways and biological processes including gene mutations , epigenetic changes , modifications in gene regulation , and environmental influences . In the process to integrate all of this information many challenges arise , among them the high dimensionality of data—since >2 omics data sets with millions of measurements are available from the same set of individuals—and the huge heterogeneity of omics data due to the different measurement scales [4] . Besides that , the data might be highly correlated , i . e . Single Nucleotide Polymorphisms ( SNPs ) that are in high linkage disequilibrium ( LD ) block or DNA CpG sites that belong to the same CpG island , contributing to multicollinearity in the analysis . Another challenge in omics data integration regards to the very small number of individuals in comparison to the number of parameters ( “n << p” ) . In addition , interpretation and validation of omics derived results require of resources that are still lacking at present . In this rapidly evolving scenario , advanced methodological techniques are continuously emerging , demanding the development of improved data analysis tools [5–7] . Integrative omics analysis refers to the combination of at least two different types of omics data . Relationships between two sets of omics parameters such as the expression quantitative trait loci ( eQTL ) [2 , 8 , 9] or the methylation-QTL ( methQTL ) [3 , 10 , 11] , have been recently reported . The approach most commonly used for this type of pairwise analysis has been univariate models ( i . e . , Spearman/Pearson correlation or linear regression models ) , assuming that the changes in gene expression levels are only affected by one parameter . Until present , the combination of >2 omics data has been less explored . Towards this end , the previously mentioned challenges are magnified and there is a lack of advanced methodologies to deal with them . Recently , we published an integrative framework as a first approach to integrate genomics , epigenomics , and transcriptomics in individuals with urothelial bladder cancer ( UBC ) [12] . In that work , we found that some gene expressions were co-regulated by both DNA methylation and genetic variants , both acting together in trans relationships . Therefore , the integration of multiple types of omics data by applying multivariable approaches becomes essential to understand the intricacy of the genomic mechanisms behind complex diseases and to overcome the abovementioned challenges . In this regard , previous developments are Principal Component Analysis ( PCA ) , to reduce data dimensionality , or Canonical Correlation Analysis ( CCA ) to investigate the overall correlation between two sets of variables . However , these methods are descriptive or exploratory techniques rather than hypothesis-testing tools . While some statistical applications have been developed in an omics integrative framework ( sparse canonical correlation analysis [13] , multiple factor analysis [14] , or multivariate partial least square regression [15] ) , none of them offers the possibility to combine >2 omics data together in the same model . The Least Absolute Shrinkage and Selection Operator ( LASSO ) proposed by Tibshirani in 1996 [16] and the Elastic Net ( ENET ) proposed by Hui Zou and Trevor Hastie in 2005 [17] are penalized regression methods that , after appropriate standardization , can model more than one type of omics data , face multicollinearity issues , and mitigate the “n << p” problem . More importantly , both methods simultaneously execute variable selection and parameter estimation , thus reducing the computation time , while the traditional methods work on the two problems separately , first selecting the relevant parameters and then computing the estimates . LASSO and ENET have already been applied to GWAS studies [18–20] as well as in the context of integrative studies [21] . One limitation of penalized regression techniques is that the penalty produces biased estimators; consequently , standard errors are not meaningful and cannot provide p-values to assess significance . Here , we propose a permutation-based approach to assess significance and we combine it with a correction for Multiple Testing ( MT ) using the MaxT algorithm [22] . We apply this permutation-based MaxT method with LASSO and ENET to identify relationships between common genetic variation , DNA methylation , and gene expression , all determined in UBC tumor samples . Specifically , we first built a two omics integrative model associating SNPs or CpGs with gene expression levels and , then , we integrated the three omics data to assess whether changes in gene expression levels could be confounded/modified by genetic variants and/or DNA methylation .
LASSO and ENET penalized regression methods are applied to high-dimensional problems with a large number of parameters . The penalization produces a shrinkage of the regression coefficients towards zero given a sparse model reducing the irrelevant parameters . Both methods deal with highly correlated variables though in a different way . LASSO tends to select one variable from a group of correlated features whereas ENET selects the whole group of variables , when evidence for their relevance exists . The shrunk estimators introduce a bias while reducing the variance resulting in a better precision and accuracy model and , therefore , increasing its statistical power . MaxT algorithm of Westfall & Young [22] is a step-down FWER-controlling MT procedure . The method uses the raw p-values or directly the statistics as explained in [26] . Using this approach , the permutation needed to obtain the p-values was combined with the one needed to apply the MaxT algorithm saving computational time . In this work , we used the deviance obtained per each of the permuted LASSO/ENET model to compute the MaxT algorithm and individuals within gene expression measure were permuted , that is the dependent variable in the models . The algorithm is explained in Box 1 . 70 patients with a histologically confirmed UBC were recruited in 2 hospitals during 1997–1998 as part of the pilot phase of the Spanish Bladder Cancer/EPICURO Study . According to established criteria based on tumor stage and grade for UBC , the tumors were classified as low-grade non-muscle invasive , high-grade non-muscle invasive , and muscle invasive . Three sets of omics data were obtained using fresh tumor tissue , including common genetic variation ( GSE51641 ) , DNA methylation ( GSE71666 ) , and gene expression ( GSE71576 ) . The three omics data overlapped in 27 individuals that are included in this study and comprise 44% low-grade non-muscle invasive tumors , 30% high-grade non-muscle invasive tumors and 26% muscle invasive tumors . S1 Table shows the IDs of the 27 samples used in the following analysis . The local ethics committee of the participating centers approved the study and written informed consent was obtained from all participants at the time of recruitment . Genotyping of tumor samples was performed using Illumina HumanHap 1M array . A total of 1 , 047 , 101 SNPs were determined in 46 individuals and , after the standard quality control and filter the SNPs that were in perfect LD ( r2 = 1 ) , they resulted in 567 , 513 SNPs . The application of multivariable models required no missing values , so genotypes were imputed with BEAGLE 3 . 0 method [27] . CpG methylation data was generated using the Infinium Human Methylation 27 BeadChip Kit . At each CpG site , the methylation levels were measured with M-values using the log2 transformation of the β-values since they are more statistically valid due to a better approximation of the homoscedasticity . The initial number of CpGs in the studied array was 27 , 578 and after background normalization and QC , a total number of 23 , 034 CpGs were left for analysis . Gene expression data were obtained from 44 tumor samples using the Affymetrix DNA Microarray Human Gene 1 . 0 ST Array with 32 , 321 probes . After the application of QC , it resulted in 20 , 899 probes determined in 37 individuals . Further details about the preprocessing of the data and the quality control applied can be found elsewhere [12] . The three measures were annotated using the UCSC hg19 , NCBI build 37 to make them comparable and homogenize their position in the genome . To generate a simulation sample , the association between SNPs and/or CpGs with gene expression was broken and therefore no significant results should be observed . To do that , 10-gene expression probes were randomly selected from our discovery sample showing no correlation structure between the probes and following a multivariate normal distribution . Then , the mean ( μ = 8 . 4 ) and variance ( σ2 = 0 . 4 ) of all the probes together were obtained . Finally , a simulated set of gene expression probes was generated using the normal distribution obtained and considering the same sample size of the discovery phase ( p = 20 , 899 probes and N = 27 individuals ) . UBC tumor data were obtained from The Cancer Genome Atlas ( TCGA ) consortium ( https://tcga-data . nci . nih . gov/tcga/ ) to replicate our findings . Data was downloaded and processed with the TCGA-Assembler [28] . The study included only individuals with muscle invasive UBC and the tumors were profiled with genome wide 6 . 0 Affymetrix , RNASeqV2 , and HumanMethylation450K Illumina arrays yielding data for 20 , 502 gene expression probes , 905 , 422 SNPs , and 350 , 271 CpGs . The total number of individuals with overlapping data from the three platforms was 238 and they were used in the replication phase of this contribution . S2 Table shows the IDs corresponding to these 238 samples . Penalized regression methods LASSO and ENET were applied to the discovery data in combination with the proposed permutation-based MaxT method to select the SNPs and/or CpGs associated with gene expression levels in the following multivariable models: SNP model: GeneExpressionlevelsi=α1SNP1+α2SNP2+⋯+αpSNPp;i=1…m CPG model: GeneExpressionlevelsi=γ1CPG1+γ2CPG2+⋯+γpCPGp;i=1…m Global model = SNP + CPG model: GeneExpressionlevelsi=α1SNP1+⋯+αpSNPp+γ1CPG1+⋯+γpCPGp;i=1…m To apply this integrative idea to our set of data the following steps were performed: ( 1 ) SNPs and CpGs that were in a 1MB window upstream and downstream were selected from each probe in the gene expression array; ( 2 ) LASSO and ENET were applied to each probe and model ( SNP , CpG , and Global models ) obtaining the deviance per model; and ( 3 ) , the permutation-based MaxT method was applied to obtain the adjusted p-values ( B = 100 permutations and significant adjusted p-value < 0 . 1 ) . The scenario and workflow is represented in Fig 1 . Subsequently , this analysis flow was applied to the simulated data set using the same criteria . In the replication scenario , we aimed at determining whether the genes that were significant in the discovery phase were also significant in the replication dataset . Therefore , the analysis was restricted to the genes found to be significant in the discovery phase considering all models ( SNP , CPG and/or Global ) and methods ( LASSO and/or ENET ) . Following the pipeline shown in Fig 1 , we focused on the significant genes found in the discovery phase and SNPs and CpGs were selected in 1MB window from the TCGA database , even if the SNPs and CpGs were not the same as those analyzed in the discovery phase . Second , LASSO and/or ENET were conducted to SNP , CPG , and/or Global models . Finally , the permutation-based MaxT method was applied to obtain significance and correct for multiple testing . The replication analysis was performed with the same software and criteria as in the discovery analysis . To provide a biological interpretation to the results , the entire list of the significant genes identified in the discovery phase by both LASSO and ENET , and by the three models , was used to perform a gene enrichment analysis with the bioinformatics tool DAVID [29 , 30] . The functional annotation clustering analysis module offered by DAVID was used . The gene term annotation is based on 14 annotation categories ( Gene Ontology ( GO ) , Biological process , GO Molecular Function , GO Cellular Component , KEGG Pathways , BioCarta Pathways , Swiss-Prot Keywords , BBID Pathways , SMART Domains , NIH Genetics Association DB , UniProt Sequence Features , COG/KOG Ontology , NCBI OMIM , InterPro Domains , and PIR Super-Family Names ) collected in the DAVID tool knowledgebase ( https://david . ncifcrf . gov/knowledgebase/DAVID_knowledgebase . html ) . The method identifies related genes by measuring the similarity of their global annotation profiles . So , the “grouping term” is based on the idea that two genes that have similar annotation profiles are functionally related . Each group term provides an enrichment score ( ES ) that indicates biological significance when ≥1 . 3 ( equivalent to non-log scale 0 . 05 ) . DAVID also provides a p-value to examine the significance of gene-term enrichment , which is corrected by Benjamini MT [31] .
LASSO and ENET were applied to 20 , 899 gene expression probes in each of the three models . Under the conditions mentioned above , LASSO yielded 9 genes with a significant signal in the SNP model , 19 in the CpG model , and 23 in the Global model . Table 1 shows the significant genes , mapped to each probe , with its deviance and p-value . Fig 2A–2C display all the probes analyzed with their deviances represented across the genome . Detailed information about the SNPs and/or CpGs mapped to these genes is provided as Supplementary Material ( S1–S6 Excel ) . ENET identified a lower number of significant genes: 11 in the SNP model , 6 in the CpG model , and 4 in the Global model . These results are shown in Table 2 and Fig 2D–2F . When the MT correction threshold was relaxed , ENET provided additional significant genes . Some genes overlapped among methods and models: CLIC6 was identified by the three LASSO models; AIM2 and SCNN1A came out in the SNP and CpG models; PTN , CRTAC1 , SERPINB3 and SERPINB4 were identified in the SNP and Global models; and S100A9 , IGJ , FREM2 , C15orf48 and KRT20 emerged in the CpG and Global models . Interestingly , 15 genes showed significance in the Global model when combining 3 omics data while they were not detected when analyzing only 2 types of omics data . The overlap of genes identified by the ENET model was lower: MSMB and IGF2 were identified by the SNP and CpG models , and PTN and SERPINB3 were selected by the SNP and the Global models . When comparing the methods , an overlap between LASSO and ENET was found for four ( PTN , SERPINB3 , SERPINB4 and CEACAM6 ) , one ( MSMB ) , and three ( SERPINB3 , PTN and IGHD ) significant genes in the SNP , CpG , and Global models , respectively . These results are displayed in Fig 3 using Venn diagrams . In the simulation study , as expected , no gene was significantly associated with any of the two methods and the three models . An example of the deviances of each gene for the SNP 303 model and LASSO method is shown in S1 Fig . The replication study was restricted to those genes ( n = 48 ) that showed significant results in the discovery phase and we applied the same models , methods , and criteria of analysis to the TCGA data . Overall , we were able to replicate 75% of the results: 36 out of the 48 genes yielded a significant association at least in one of the models considered . Regarding the LASSO models , we replicated 3/9 genes from the SNP models , 17/19 genes from the CPG models , and 19/23 genes from the Global models ( Table 3 ) . Regarding ENET , we replicated 3/10 genes from the SNP model , 3/6 genes from the CPG model , and 3/3 genes from the Global model ( Table 4 ) . Using DAVID , 46 out of 48 genes showing significant signals in the discovery phase were annotated from 14 public categories . After enrichment analysis , 7 clusters with an ES ≥1 . 3 were found ( S3 Table ) . The cluster with the highest ES ( 3 . 5 ) regarded to the terms “extracellular region , secreted , and signal peptide” grouping the genes OLFM4 , CRTAC1 , MSMB , IGJ , MMP7 , IGF2 , PIGR , TCN1 , CXCL17 , S100A9 , SAA1 , IGHD , CRH , CTSE , FREM2 , PLA2G2A , CEACAM7 , CEACAM6 , CEACAM5 , REN , PTN , CP . The rest of the clusters with an ES ≥1 . 3 were not significant after MT correction . Cluster 5 ( ES = 1 . 4 ) contains 3 genes coding for keratins ( KRT5 , KRT13 , KRT20 ) , cytoskeletal components that are regulated during urothelial differentiation , whose expression is altered in UBC , that have been proposed as markers for the molecular taxonomy of UBC [32] . In addition , cluster 7 “EF hand and calcium ion binding” ( ES = 1 . 3 ) contains multiple genes shown to play an important role in cancer ( S100A9 , S100A2 , CAPNS2 , ANXA10 , CRTAC1 , FREM2 , MMP7 , PLA2G2A ) , including two members of the S100A family of proteins .
Common genetic variation ( GSE51641 ) , DNA methylation ( GSE71666 ) , and gene expression ( GSE71576 ) data for the discovery phase are available in GEO . | At present , it is already possible to generate different type of omics–high throughput–data in the same individuals . However , we lack methodology to adequately combine them . Many challenges arise while the amount of data increases and we need to find the way to identify and understand the complex relationships when integrating data . In this regard , new statistical approaches are needed , such as the ones we propose and apply here to integrate three types of omics data ( genomics , epigenomics , and transcriptomics ) generated using bladder cancer tumor samples . These innovative approaches ( LASSO and ENET combined with a permutation-based MaxT method ) allowed us to find 48 genes whose expression levels were significantly associated with genomics and epigenomics markers . The adequacy of this approach was confirmed by the use of an independent data set from The Cancer Genome Atlas Consortium: 75% of the genes were replicated . Previous sound biological evidences further support the results obtained . | [
"Abstract",
"Introduction",
"Material",
"and",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | Integration Analysis of Three Omics Data Using Penalized Regression Methods: An Application to Bladder Cancer |
The P2X4 receptor ( P2X4R ) is a member of a family of purinergic channels activated by extracellular ATP through three orthosteric binding sites and allosterically regulated by ivermectin ( IVM ) , a broad-spectrum antiparasitic agent . Treatment with IVM increases the efficacy of ATP to activate P2X4R , slows both receptor desensitization during sustained ATP application and receptor deactivation after ATP washout , and makes the receptor pore permeable to NMDG+ , a large organic cation . Previously , we developed a Markov model based on the presence of one IVM binding site , which described some effects of IVM on rat P2X4R . Here we present two novel models , both with three IVM binding sites . The simpler one-layer model can reproduce many of the observed time series of evoked currents , but does not capture well the short time scales of activation , desensitization , and deactivation . A more complex two-layer model can reproduce the transient changes in desensitization observed upon IVM application , the significant increase in ATP-induced current amplitudes at low IVM concentrations , and the modest increase in the unitary conductance . In addition , the two-layer model suggests that this receptor can exist in a deeply inactivated state , not responsive to ATP , and that its desensitization rate can be altered by each of the three IVM binding sites . In summary , this study provides a detailed analysis of P2X4R kinetics and elucidates the orthosteric and allosteric mechanisms regulating its channel gating .
Purinergic P2X receptors ( P2XRs ) are a family of ligand-gated non-selective cation channels that are activated by extracellular adenosine 5'-triphosphate ( ATP ) . In mammals , there are seven distinct subunits of this family of proteins , labeled P2X1-7 . Each subunit contains intracellular N- and C-termini connected to the first and second transmembrane ( TM ) domains , respectively , followed by a large extracellular loop commonly referred to as the ectodomain . It is also well established that P2X subunits aggregate to form functional trimers [1–5]; receptors may be composed of either one type of subunit ( homotrimer ) or a mixture of more than one type of subunit ( heterotrimer ) [6] . When coordinated in a trimer , the interfaces between adjacent ectodomains form three binding pockets for ATP [7] . These ectodomains also form fenestrations which are lined by negatively charged amino acids that attract cations , and the cation selectivity of these channels is determined by the selectivity filter localized in the TM2 domain [8] . The binding of two or three ATP molecules to the extracellular binding sites induces conformational changes in the ectodomain and subsequently the TM domains , causing the channel opening . The gating of P2XR by ATP and other orthosteric agonists can be broken down into three distinguishable phases , activation , desensitization , and deactivation , defined by their ionic current kinetics in whole-cell recordings [9 , 10] . Activation is a rapid phase of channel opening that corresponds to increasing inward current subsequent to agonist application . This is usually followed by the desensitization phase , a decay of current amplitude in the maintained presence of an agonist , with an onset that is slower than that of activation . After agonist removal from the medium , a relatively rapid decrease in current amplitude , referred to as the deactivation phase , is observed . Receptors differ in both their sensitivity to agonists and the kinetics of the phases ( with desensitization being transient and controversial in P2X7Rs ) [9 , 11] . It has also been suggested that P2X2R and P2X7R are capable of exhibiting another phase in their gating , termed dilation , when the receptor pore was thought to progressively enlarge during sustained ATP application . Two observations were used as evidence for pore dilation: the ability of these receptors to become permeable to N-methyl-D-glucamine ( NMDG+ ) , a large organic cation ( ~7 . 3 Å in mean diameter ) , and the change in reversal potential ( Erev ) during a ramp protocol , when cells were bathed in a medium containing only NMDG+ with pipette containing only NaCl [12] . Recent investigations , however , have shown that changes in Erev during prolonged channel activation of P2X2R do not reflect pore dilation , but rather time-dependent alterations in the concentration of intracellular ions , specifically washout of intracellular Na+ and gain of NMDG+ through the initially opened large pore of P2XR [13 , 14] and that permeation of NMDG+ can also occur without pore dilation [15] . In addition to orthosteric regulation , P2XRs also exhibit allosteric regulation [9] ) , as is evident from the action of ivermectin ( IVM ) on P2X4R channels . Extracellularly applied IVM increases current amplitude at low concentrations , and increases the sensitivity of receptors to ATP and partial agonists at higher concentrations . IVM also decreases the extent of desensitization in the continuous presence of agonist and prolongs deactivation of the receptor after the removal of agonists [16–18] . Furthermore , P2X4R is not substantially permeable to NMDG+ natively [16 , 19] , but displays a shift in Erev in the presence of IVM , suggesting that the channel pore is permeable to NMDG+ in IVM-treated cells [20] . The action of IVM on P2X4R gating is also time-dependent; i . e . the cells must be exposed to IVM for at least 30 s , in the absence of ATP , to alter the P2X4R gating ( compared to ms for orthosteric activation ) [20] . The onset of IVM’s potentiating effect on P2X4R current amplitude is faster than the effects of IVM on deactivation kinetics [17 , 21] . Consequently it has been postulated that the two distinct effects of IVM are due to binding at two distinct sites [17] . Experiments with chimeric receptors containing domains from IVM-sensitive P2X4R and the IVM-insensitive P2X2R have provided evidence that TM domains play a critical role in this allosteric modulation by IVM [18] . The location of the IVM binding site has not yet been addressed in the context of the recent crystal structures of a zfP2X4R [2 , 3] . However , IVM apparently inserts between pairs of neighboring subunits of the P2X4R channel in the membrane and interferes with the molecular rearrangement in the TM domains involved in channel gating , similarly to glutamate-gated chloride channels crystalized with IVMs [22] . Accordingly , there should be three potential binding sites for IVM in the P2X4R , as there are three clefts between subunits . Such a topography of IVM binding sites provides rationale as to why receptors ( not previously stimulated orthosterically ) must be exposed to IVM for a prolonged period for it to be effective . Subsequently , we have used the term priming to describe the time and concentration dependence of IVM to occupy its binding sites and the resulting development of its varied allosteric effects . In recent years , mathematical modeling has begun to shed light on many aspects of P2XRs and to guide experimental designs to arrive at a more complete understanding of channel gating [19 , 20 , 23–25] . Biophysically detailed Markov models that describe individual orthosteric binding sites and their allosteric modulation , have been very successful in deciphering the kinetics of P2X homotrimers and succinctly explaining many phenomena [19 , 20 , 24–26] . They consider important biophysical details such as the conformational states of individual binding sites and other structural components of the receptor . One of these models for P2X4R is a simple Markov model that takes into account the sequential binding of ATP to its three subunits and assumes that IVM causes receptor sensitization upon the binding of three ATP molecules , that all ATP unbinding rates are decreasing functions of IVM concentration , and that IVM induces a change in ion selectivity caused by the assumed pore dilation [20] . In this model , the three allosteric effects of IVM on P2X4Rs are induced by a single IVM-dependent transition that allowed for generating the shift in Erev during the ramp protocol . However , the published model is unable to account for effects of pre-treatment with IVM before ATP application . The model also predicts a large ( > 150% ) increase in the unitary ( single-channel ) conductance of the receptor , in contrast to experimental evidence [17] indicating that there is at most 20% increase in unitary conductance . To satisfy these constraints , we developed two substantially larger models that not only fit the data more closely in more experimental circumstances but offer better insights into how the kinetics of ATP and IVM sequential binding to P2X4R affect P2X4R activation , desensitization , and deactivation . They also illustrate how changes in ion selectivity of these receptors are manifested , as well as predict the previously unappreciated existence of receptor states ( including the deeply inactivated and primed states ) that are not directly observable in the experimental current recordings .
When HEK293 cells expressing rat P2X4R were bathed in Ca2+-containing medium , where Na+ was substituted by NMDG+ and a voltage ramp from −80 mV to +80 mV was applied to the cell , Erev was not found to change during sustained applications of 100 μM concentrations of ATP ( S1 Fig , left ) . However , pretreatment with 3 μM IVM for 60 s caused a positive shift in Erev during sustained applications of 100 μM ATP ( S1 Fig , right ) . This is consistent with our earlier work [20] and the finding of others that IVM potentiates ATP-induced responses and increases permeability for NMDG+ , but cannot activate P2X4R channels on its own [16] . Because strategies that rely on changes in Erev to provide evidence for large pore formation during sustained stimulation with agonist were questioned [13 , 15] , we examined currents induced by ATP in Ca2+-free/NMDG+-containing medium . S2 Fig shows that in the absence of Ca2+ , a 40-s application of ATP ( 100 μM ) at −60 mV in bi-ionic NMDG+ out/Na+ in solution ( where the reversal potential of ATP-induced current is about -70 mV [27] ) evoked only outward Na+ current , whereas in the presence of IVM , outward Na+ current was followed by inward NMDG+ current . These experiments do not argue against findings with P2X2R presented recently [13] but provide evidence for the existence of two conductive pore states of P2X4R . These pore states , termed open1 and open2 , differ in their selectivity for organic cations ( a consequence of altered relative permeability PNMDG/PNa ) , and the priming of receptors by IVM is needed to switch from one conductive state to another . We will demonstrate later that the shift in Erev does not require an increase in unitary conductance associated with the open2 state ( s ) , but rather depends on the selectivity associated with Na+ and NMDG+ , as suggested by the Goldman-Hodgkin-Katz equation , Vrev=RTFlnPNa[Na+]out + PNMDG[NMDG+]outPNa[Na+]in + PNMDG[NMDG+]in , where R is the gas constant , T is the absolute temperature , F is Faraday’s constant , PNa ( PNMDG ) is Na+ ( NMDG+ ) permeability , and [Na+]out ( [Na+]in ) is Na+ concentration outside ( inside ) the cell , whereas [NMDG+]out ( [NMDG+]in ) is NMDG+ concentration outside ( inside ) the cell . Because the experimentally observed Erev shift is independent of the increase in unitary conductance , the term open2 state will be used to refer to both the ( small ) increase in unitary conductance and the ( large ) change in ion selectivity of the P2X4R pore . The previous paragraph proposed that P2X4R opens with the open2 pore state ( s ) in the presence of IVM , which may have an increase in unitary conductance of as much as 20% [17] . At the same time , the ramp protocol shows a decrease in the slopes of the I-V curves ( S1 Fig ) . To enforce such an outcome in any potential model of P2X4R with IVM-dependent allosteric transitions between open states , we require the rate of increase in the probability of open states due to allostery ( open1 → open2 ) to be slower than the rate of decrease in the probability of open states ( open1 → desensitized ) . We propose that the increased conductance of the open2 state ( s ) of the receptor pores is masked by desensitization in a time-dependent manner , similar to our previous finding with P2X2R [19] . We are thus led to assume that the probability of finding open receptors on the cell membrane , P ( open1 ) , is a strictly decreasing function of time ( P˙ ( open1 ) < 0 ) . On the other hand , we expect that the probability of finding a receptor whose pore is in the open2 state ( s ) , P ( open2 ) to be an increasing function of time ( P˙ ( open2 ) > 0 ) . Without specifying a Markov model to describe P2X4R kinetics , we may consider a generic equation for current production in these receptors , capable of distinguishing open1 and open2 states based on their conductances and reversal potentials established after washout of intracellular Na+ and gain of NMDG+ . According to the description above , we can write the equation for current as I=g1P ( open1 ) ( V−E1 ) +g2P ( open2 ) ( V−E2 ) ( 1 ) where g1 is the maximum conductance of the open1 state ( s ) , g2 ( > g1 ) is the maximum conductance of open2 state ( s ) , and E1 and E2 are the reversal potentials associated with the open1 and open2 states , respectively . The current equation can be rewritten in a standard form to isolate the total conductance and reversal potential of the cell , as follows I=gtot ( V−Etot ) , ( 2 ) where gtot and Etot are the total conductance and reversal potential of the cell , respectively . By equating Eqs ( 1 ) and ( 2 ) , we obtain gtot=g1P ( open1 ) +g2P ( open2 ) . The requirement for the slope of the I-V curves to decrease during the ramp protocol can be met if the total conductance of the receptor population decreases over time , i . e . , g˙tot<0 . Taking the time derivative of gtot and rearranging the terms , we obtain g˙tot=g1 ( P˙ ( open1 ) +g2g1P˙ ( open2 ) ) , which is strictly negative if we impose the condition −P˙ ( open1 ) >g2g1P˙ ( open2 ) . ( 3 ) It follows that |P˙ ( open1 ) |>g2g1P˙ ( open2 ) . This result implies that the total conductance of the cell will decrease if the fraction of open receptors decreases more rapidly than the ratio of the open1-to-open2 maximum conductances times the rate of increase of the open2 state ( s ) . Thus , in order to capture the decrease in the slopes of the I-V curves in any model development , we have to increase the rate of desensitization of the open states , reduce the rate of increase of open2 state ( s ) or decrease the ratio between the open2 and open1 conductances . As a first approximation , we can attribute the decrease in the fraction of open states to two processes , desensitization and priming of receptors , related by the equation −P˙ ( open1 ) =P˙ ( open2 ) +δ ( 4 ) where δ is the rate of change of open receptors due to desensitization . Furthermore , letting g2 satisfy g2=g1 ( 1+f ) where f is the fractional increase in unitary conductance , we can substitute this expression into Eq ( 3 ) to obtain δ>fP˙ ( open2 ) . ( 5 ) Inequality Eq ( 5 ) represents a new condition that can be used to produce the decrease in total conductance seen in the ramp protocol . For example , if we consider the experimentally observed value of ~0 . 2 for f in human P2X4R , then the rate of desensitization only needs to be one fifth the rate of the IVM-induced unitary conductance increase in order to mask its effect on the slopes of the I-V curves . The desensitization rate of naïve receptors is well characterized by the current recordings produced during prolonged application of ATP , which can be used to constrain δ as a fixed parameter . The IVM-induced increase in unitary conductance has not been determined for rat P2X4R , nor its time-course . We determine these in the Markov model ( discussed below ) by fitting the total current , imposing Inequality Eq ( 5 ) to ensure that the total conductance of the cell decreases during the ramp protocol ( due to desensitization ) . We next consider the effects of IVM on desensitization and deactivation . During the pulse protocol ( Fig 1A ) , where cells were repeatedly stimulated by 1 μM ATP for 2 s twice per min in the absence ( black trace ) and presence of 1 μM IVM ( colored traces ) , we observed an initial increase in desensitization rate of the receptor ( blue trace in Fig 1A ) , followed by a gradual decrease in desensitization rate at each subsequent ATP pulse ( see the Methods section for quantification procedure ) . By the fifth pulse ( green trace in Fig 1A ) , the desensitization rate reverted back to a value comparable to that seen in the absence of IVM ( black trace in Fig 1A ) . To assess if this phenomenon occurs consistently , we evaluated the statistical significance of the transient increase in desensitization rate . To quantify the amount of desensitization seen in the recordings , we used linear fitting to measure the rate of receptor desensitization normalized by the current amplitude of each pulse . As shown in Fig 1B , we did not see a significant change in the rate of desensitization at each ATP pulse in the absence of IVM ( filled circles ) ( n = 7 ) , suggesting that the desensitization proccess of receptors is far from equilibrium . However , in the presence of IVM , the first two ATP pulses following IVM application ( indicated by the small arrows ) exhibited a significant ( p < 0 . 005 and p < 0 . 05; n = 7 ) increase in desensitization rates . The desensitization rate of current recordings in subsequent ATP pulses gradually drifted back to its original value before IVM was applied , further suggesting that the open state , exhibiting an increased desensitization rate , has reached an equilibrium with its corresponding desensitized state . At higher IVM concentrations , however , these transient effects were not observed , but an increase in non-desensitized current amplitude was found [20] . Thus , while the binding of IVM potentiates P2X4R , it also increases both the apparent rate of desensitization at low ATP concentrations and the rate of recovery from desensitization ( i . e . , it lowers the Gibbs free energy barrier for these transitions ) . To assess the deactivation kinetics ( i . e . , decay of current amplitude following washout of agonist ) of P2X4R , we used the same pulse protocol of 1 μM ATP for 2 s twice per minute ( Fig 1C ) . In the absence of IVM ( filled circles ) , receptors underwent fast deactivation with a time constant that remained roughly the same at each pulse , whereas in the presence of 1 μM IVM ( open circles following the small arrow ) the deactivation time constant progressively increased with incubation time , indicating a decrease in receptor deactivation rates . This effect became even more pronounced at higher IVM concentrations . At IVM concentrations greater than or equal to 10 μM , deactivation following washout of IVM was not always complete ( see Fig 2A in [20] ) , suggesting that complex physiological processes might be initiated at these concentrations . These results are consistent with the idea that IVM increases the sensitivity of the receptor to ATP and decreases the rate of agonist unbinding following its washout from medium [16–18] . Our previous study showed concentration response curves for rat P2X4R stimulated by ATP in the presence and absence of 3 μM IVM ( S3 Fig ) ; ATP alone was found to produce a concentration response curve with an EC50 of 2 . 3±0 . 4 μM ( blue line ) , and with 30-s pretreatment with IVM , the EC50 was 0 . 5±0 . 1 μM ( green line ) [20] . A similar conclusion was reached with human P2X4R [17] . A pretreatment period of 10 s was also considered and , whereas it did produce the same maximal current amplitude , an intermediate EC50 of 1 . 6±0 . 3 μM was measured ( maroon line ) [20] . This suggests that there are at least two distinct priming effects associated with IVM with separate time scales of action . First , IVM primes receptors by increasing the maximal whole-cell current response . Second , after prolonged exposures to IVM , receptors become further primed by an increased sensitivity to ATP ( previously called sensitization ) . The model in [20] was only partially able to account for these behaviors and specifically was not able to account for their dependence on the duration of pre-treatment because there were no kinetics associated with IVM binding in the absence of ATP . The concentration response curves of P2X4R ( S3 Fig ) reveal that not only do 10- and 30-s pretreatments with IVM increase sensitivity to ATP ( maroon and green lines , respectively ) , but they also increase the maximum current amplitude evoked by ATP [17 , 20] . The two hypotheses that can explain this behavior are: ( i ) the unitary conductance of individual channels increases; or ( ii ) the number of open receptors is rising ( i . e . , the maximal open probability increases ) . Although there is evidence that the former hypothesis holds [17] , this does not preclude a change as well in the maximal open probability with IVM application [17 , 20 , 23] . In fact , it was reported that the maximal open probability in the absence of IVM is ~0 . 2 compared to ~0 . 8 in the presence of IVM [17 , 23] . This phenomenon was previously explained by the Markov model in [20] , which assumes that IVM modifies the connectivity between open and desensitized states , but that model required a large increase in unitary conductance . This assumption on the conductance is inconsistent with a study of human P2X4R [17] , which showed that IVM produces a roughly 5-fold increase in maximal current amplitude while only inducing a 20% increase in unitary conductance . Those authors posited that , in the absence of IVM , desensitization plays a large role in reducing the current amplitude , whereas when IVM is applied , desensitization is greatly reduced , enhancing the observed current . While this is a plausible explanation , we are not aware of any receptors that function in this manner . Moreover , no quantitative analysis was made to assess to what extent such a mechanism produces the observed effect . In order to test this hypothesis , we constructed a simple and generic mathematical scheme ( hereafter referred to as a gating scheme ) of a desensitizing ligand-activated receptor ( Fig 2A ) . It consists of two rows: a naïve row comprised of two closed states ( C1 , C2 ) and two conducting states ( Q1 , Q2 ) , and a desensitized row comprised of four nonconducting desensitized states ( D1 , D2 , D3 , D4 ) . As was done in [20] , we assume that channels open from states with two or three bound ATP molecules . This is in accordance with the finding that a single-bound receptor state does not lead to activation of P2X7 channels [28] . This is also consistent with previous models of P2XRs and the notion that a single kinetic model underlies the functioning of all receptor subtypes [19 , 20 , 24–26 , 29] . Forward ( backward ) transitions between two states along each row represent a single ATP binding ( unbinding ) with rates k2 , k4 , k6 ( k1 , k3 , k5 ) , respectively , whereas upward ( downward ) transitions represent desensitization ( recovery ) with a rate kd ( kr ) . Concentration response curves were generated for this gating scheme , each with a progressively increasing rate of desensitization kd ( see Figs 2A and S5A ) . It was found that although reduced desensitization rates are capable of increasing the current amplitude at a given agonist ( such as ATP for P2X4R ) concentration , the mechanism proposed in [17] is unable to significantly increase the maximal current amplitude evoked by the agonist . Rather , it shifts the EC50 of the concentration response curves leftward as well as increases the Hill coefficient in such a way that the saturating phase of the concentration response curves are shifted by many orders of magnitude . A leftward shift in EC50 and modulation of the Hill coefficient by IVM have been observed experimentally [17 , 20] . This was , however , consistently associated with an increase in the maximal current amplitude , which the desensitization mechanism cannot produce at saturating agonist concentrations ( see Imax in the legend of S5A Fig ) . A mathematical model introduced by Silberberg et al . also used an IVM-dependent transition rather than modulation of desensitization by IVM in order to produce the increase in maximal current [23] . Therefore , the mechanism suggested in [17] seems unable , at least on its own , to explain the effects of IVM on the concentration-response relationship for the peak current of P2X4R . After having ruled out decreased desensitization as a cause for the increased maximal current amplitude in the presence of a modulator , we tested an alternative hypothesis , that the closed states exist in equilibrium with a deeply inactivated state ( C0 ) for which the agonist is not effective ( Fig 2B ) . This mechanism has previously been used in Markov models of sodium channels [30] . Transitions linking the two states must be slow , but the equilibrium mixture of the closed-inactivated subsystem ( C0 ↔ C1 ) establishes an upper bound on the maximal open probability in the absence of IVM , given by POMax=H1H1+H2 ( 6 ) even at the highest agonist concentration . In other words , before agonist application , only some fraction of receptors are in C1 and are susceptible to agonist-induced activation but in the presence of IVM , more can be recruited ( from C0 ) into C1 . Evidence of such recruitment was first seen during prolonged application of ATP in the absence and presence of IVM [20] and was obtained also by application of IVM to fully desensitized receptors ( S4 Fig ) . To see how effective this mechanism is in producing the observed effects in [17] , we tested the gating scheme of Fig 2B quantitatively , by progressively increasing the transition rate H1 ( between C0 and C1 ) , and plotting the concentration response curves ( S5B Fig ) . Increasing H1 decreased inactivation , which increased the fraction of receptors in C1 and thus POMax . Whereas reducing the occupancy of the deeply inactivated state is highly effective at increasing the maximal current amplitude , it does not significantly shift the concentration response curves or alter the Hill coefficients . Therefore , in order to match the experimental findings that IVM pretreatment of P2X4R not only increased maximal current but also shifted the EC50 leftward ( S3 Fig ) and increased the Hill coefficients , both reduced desensitization and rescue from a deeply inactivated state seem to be required . A model incorporating both features is described in the next section . Based on the above considerations , we designed a one-layer Markov state model that describes the full kinetics of ATP and IVM binding to P2X4R and tested it against experimental data . For a detailed description of the model , see S6 Fig , Table A and Appendix A in S1 Text . Briefly , it is a revised version of the model of Zemkova et . al . [20] that now assumes 3 IVM binding sites , that the binding of IVM acts on P2X4R independently of ATP binding , and that IVM can bind to any ATP-bound state , not just the 3-ATP bound naïve state . Sequential binding of IVM causes three stages of receptor priming , depending on number of IVM molecules bound to receptor: primed-1 , primed-2 , and primed-3 . Primed-1 receptors respond to ATP application with increased current amplitude , reflecting increased open probability . Primed-2 receptors exhibit modestly increased unitary conductance for Na+ and significantly increased unitary conductance for NMDG+ , whereas primed-3 receptors show increased ATP binding affinity . The model also incorporates rescue from the deeply inactivated state by IVM , and therefore has a maximal open probability given by Eq ( 6 ) in the absence of IVM . Although our analysis of this model ( and several variations of it ) revealed that it possesses many of the necessary ingredients to capture the gating properties of P2X4R and several aspects of its current recordings ( S7 and S8 Figs ) , it includes the implausible assumption that receptors in the primed states must lose all bound IVM molecules in order to desensitize . This assumption led to two major issues in the performance of the model: ( i ) it did not capture accurately the short timescales of activation and desensitization robustly; and ( ii ) it produced discrepancies in current amplitudes when compared to experimental data during the pulse protocol . That motivated us to design a more accurate model of P2X4R kinetics .
Data-based Markov state models that describe the processes of ligand binding/unbinding to ligand-gated receptor are powerful tools to understand orthosteric and allosteric regulation of these channels . P2X4Rs are prototypical examples of such receptors with orthosteric and allosteric binding sites for ATP and IVM , respectively . They are associated with ion channels that are permeable to small cations , including Na+ , Ca2+ , Mg2+ and K+ . The binding of ATP leads to receptor activation and channel opening , while IVM binding increases receptor unitary conductance and sensitivity to ATP . Here we analysed the kinetics of ATP and IVM binding/unbinding to P2X4R , and determined its gating properties using two detailed Markov models labelled the one-layer and two-layer models . The one-layer model extended a previously developed , simple Markov model of P2X4R by taking into consideration a deeply inactivated state , nonresponsive to ATP but responsive to IVM ( existing in equilibrium with the naïve ATP-unbound state ) , along with three additional gating schemes ( per each ATP binding ) representing the three IVM binding sites . The model also assumed that the IVM and ATP binding are independent of one another and that sequential binding of IVM can occur at any ATP-bound or unbound states . Our analysis revealed that the deeply inactivated state was essential for capturing the increase in the maximum response ( Imax ) in the ATP-dependent concentration-response curves ( in the presence of IVM ) , with only a small increase in conductance between the open1 and open2 states . Although the model was able to capture many of the essential features of P2X4R recordings ( S7 and S8 Figs ) , it assumed that IVM bound states can only desensitize by first becoming completely free of IVM . That made the model unable to robustly capture the short timescales of activation and desensitization , and it also produced current amplitudes incompatible with experimental data during the pulse protocol . By allowing the IVM-bound states to desensitize , we were able to show that a simple gating scheme is able to capture the profiles of the first pulse ( in the absence of IVM ) or the last pulse ( in the presence of IVM ) of the pulse protocol very accurately when the scheme is fitted individually to each pulse , but not both simultaneously . However , when comparing the entire pulse protocol recording to the outcome of the scheme for each case , there was a gradual increase in discrepancy between them , suggesting that a mixture of gating schemes must coexist to be able to capture all aspects of P2X4R kinetics . That led us to propose the two-layer model , which assumes that IVM-bound states can desensitize . The two-layer model was successful in capturing every aspect of P2X4R kinetics very accurately , including the short and long time scales of activation and desensitization , particularly the changes in the desensitization rate observed during the pulse protocol of Fig 1 , as well as the current amplitudes . The observed shift in the EC50 along with the increase in the maximum current , during pre-stimulation with IVM , were also reproduced by the model ( through the presence of the deeply inactivated state ) . Moreover , these gating schemes can be used to understand why ATP binding mutants with low amplitude of response tend to have significantly larger fold-increases in maximal current in the presence of IVM [21] . If we view such mutants as disproportionately populating the deeply inactivated state , where they cannot bind ATP , then their rescue by IVM from this state will produce a much larger fold-increase in maximal current . The existence of this deeply inactivated state was probed experimentally , by applying IVM to a cell whose receptors were almost completely desensitized from prolonged applications of 100 μM ATP . Upon IVM application , we observed an increase in the maximal current amplitude to about half of the initial maximal current ( see S4 Fig ) , suggesting that IVM rescued receptors from a ( deeply inactivated ) pool corresponding to about one third of all receptors . The two-layer model consisted of 4 gating schemes linked together through ATP/IVM binding/unbinding . Two of these gating schemes ( the primed-2 and -3 ) contained conducting states exhibiting a 15% increase in unitary conductance compared to that of the open states in the naïve and primed-1 states . This increase is within the 20% limit seen experimentally [17] , and is not required to produce IVM’s increase in maximal current amplitude within our model . Instead , the effect of IVM on maximal current amplitude is produced mainly by an increase in open probability . According to Eq ( 6 ) , the two layer model predicts a maximal open probability of approximately 0 . 53 in the absence of IVM ( with 47% of receptors in a deeply inactivated state ) , while it can easily reach values greater than 0 . 9 in the presence of IVM . It was suggested in [13] that the ionic conditions in the medium and the pipete are responsible for producing electrochemical effects which were long presumed to be evidence for pore dilation , particularly in P2X2R . While temporal changes in ionic gradients play a significant role in producing a shift in Erev associated with the I-V curves during the ramp protocol , our results suggest that the transition to open2 ( which is permeable to NMDG+ ) is an intrinsic property of the pore in P2X4R , is independent of the increase in unitary conductance ( S12B Fig ) and is induced by priming with IVM . The two-layer model assumes that the increase in unitary conductance associated with this transition is masked by desensitization . This results in the shift in Erev being accompanied by a decrease in the slope of the I-V curves ( due to desensitization ) . The observed decline in the slope of the I-V curves was a consequence of Inequalities Eqs ( 3 ) and ( 5 ) . The previously developed model in [20] was capable of fulfilling these conditions and producing the decrease in the slope of the I-V curves , but it required a large increase ( >150% ) in unitary conductance to achieve it while simultaneously producing the increase in current amplitude induced by IVM . The inclusion of the primed-1 row with conductance g1 ( and reversal potential E1 ) was an essential element for reproducing the shift in the I-V curves . Without this intermediate step , the model required a very positive reversal potential for the open2 state ( E2 ) , which does not reflect its loss of selectivity . Both the one-layer and two-layer models proposed here keep the increase in unitary conductance within 20% and produce the current growth with IVM pretreatment through IVM-induced transitions from the inactivated state C0 ( by increasing POMax , as given by Eq ( 6 ) , rather than increasing the maximal conductance ) . The two-layer model , however , is more plausible because it does not assume that desensitization necessitates the unbinding of IVM . Moreover , according to this model , IVM is able to transition receptors to the primed-3 states in the absence of ATP , allowing pretreatment with IVM to produce sensitization independently of ATP . It is important to note that effectively removing the 15% increase in unitary conductance associated with the primed states only slightly altered model simulations and did not abolish any experimental phenomena in symmetric ionic conditions ( S11 and S12 Figs ) . Thus , the results of the two-layer model are independent of an increase in unitary conductance , but require a change in selectivity in order to capture the shift in the I-V curves of the ramp protocol . Recently , the pore dilation hypothesis has become increasingly disputed . Molecular dynamics simulations indicate that NDMG+ is capable of permeating the open state of P2X4R pore , provided it is maintained in an open state long enough for the slow permeation event of NDMG+ to take place [15 , 32] . One of the primary effects of IVM application on the single channel kinetics of P2X4R is to shift the distribution of open times from the sub-millisecond timecale to tens of millisecond [17] . We hypothesize that the drastic change in P2X4R’s permeability for NMDG+ upon IVM application results from the priming of receptors in such a way that their pores remain in the open2 state for long enough for the slow permeation of NMDG+ . This increase in the permeability for NMDG+ ( via an IVM-dependent transition to the open2 state ) allows for its influx into the cell . Together with the efflux of Na+ , these fluxes produce a more positive Erev as determined by the Goldman-Hodgkin-Katz equation . While the two-layer model may seem to be a large departure from both the previously developed model in [20] and the one-layer model , it should be noted that in the absence of IVM , the remaining blocks of the models ( or submodels ) are identical , and that the P2X2R Markov model developed in [19] has a similar structure; it included a corresponding desensitized state for each of its closed and open states , although the desensitization pathway for primed ( sensitized ) states was calcium dependent . The increase in the number of states and number of kinetics parameters in the two-layer model was necessary to capture all the observed features of P2X4R , which previous models , including the one-layer model presented here , failed to do . A step by step validation of such an increase in complexity was provided through the use of coupled gating schemes , and the design of an extensive MCMC fitting algorithm that combined parallel tempering approaches with the t-walk method to estimate the kinetic parameters of the model efficiently . The two major allosteric effects of IVM’s on P2X4Rs , observable from whole cell currents as an increase in maximal current amplitude or the deactivation time constant , exhibit distinct concentration dependencies [17 , 20] . This suggests that they are likely caused by two independent processes . This existence of two distinct allosteric effects with differing concentration dependencies have also been reported for other P2XRs [11 , 19] , although for these receptors , ATP alone was sufficient to induce such effects . The models presented here reduce the concentration dependence of IVM’s allosteric effects on P2X4R to a single sequential binding process , and thus represent a major simplification of a more realistic model where all effects are assumed to arise from independent binding events . Despite this simplification , the model is quite capable of capturing all aspects of allosteric modification by IVM . An important item for future work is cooperativity in the ATP and IVM binding , which has been investigated in the one-layer model but not yet in the two-layer model . By assuming correlations between the binding/unbinding parameters of ATP and IVM in the one-layer model , we were able to reduce the number of estimated parameters in that model significantly and found that there was negative cooperativity in the ATP binding in the naïve and primed-1 rows . Investigating if such cooperativity exists in the two-layer model is also warranted . This can be done by imposing correlations between the kinetic parameters of the two-layer model , which will again reduce the number of estimated parameters , and testing for cooperativity in ATP binding , IVM binding and between ATP and IVM binding . These variations of the model can then be compared to each other using Bayesian approaches to determine which is most likely . In conclusion , here we present two novel models , one of which ( the two-layer model ) effectively mimics all experimental observations . In this model , receptors go through four stages of activation cycle during ATP and IVM binding: transitioning from functional to desensitized , from desensitized to internalized , from internalized to deeply inactivated and from deeply inactivated to functional . Functional and desensitized stages each exist as 16 distinct states , determined by the progressive saturation of three ATP and three IVM binding sites , whereas internalized and deeply inactivated receptors are single states . Binding of IVM influences ATP-induced gating properties of receptors , i . e . the rates of activation , desensitization and deactivation , open probability of channels , and the sensitivity of receptors to ATP . The channel pore state , open1 is predominantly permeable to small cations and open2 is permeable to large organic cations .
Experiments were performed on human embryonic kidney 293 cells ( HEK293; American Type Culture Collection ) , which were grown in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum , 50 U/mL penicillin , and 50 μg/mL streptomycin in a humidified 5% CO2 atmosphere at 37°C . Cells were cultured in 75-cm2 plastic culture flasks ( NUNC , Rochester , NY , USA ) for 36–72 h until they reached 80–95% confluence . Before the day of transfection , ~150 000 cells were plated on 35 mm culture dishes ( Sarstedt , Newton , NC , USA ) and incubated at 37°C for at least 24 h . For each culture dish of HEK293 cells , transfection of wild-type P2X4R was conducted using 2 μg of DNA with 2 μl of jetPRIME reagent in 2 ml of Dulbecco modified Eagle’s medium , according to the manufacturer’s instructions ( PolyPlus-transfection , Illkirch , France ) . After 24–48 h of incubation , the transfected cells were mechanically dispersed and re-cultured on 35 mm dishes of Corning 3294 CellBIND Surface for 1–4 hours before recording . Transfected cells were identified by the fluorescence signal of enhanced green fluorescent protein using an inverted research microscope with fluorescence illuminators ( Model IX71; Olympus , Melville , NY ) . Currents were recorded in a whole-cell configuration from cells clamped to −60 mV using an Axopatch 200B patch clamp amplifier ( Axon Instruments , Union City , CA , USA ) . All currents were captured and stored using Digidata 1550A and pClamp10 software package . Patch electrodes were pulled from borosilicate glass tube with a 1 . 65 mm outer diameter ( type GB150F-8P; Science Products GmbG , Hofheim , Germany ) using a Flaming Brown horizontal puller ( P-97; Sutter Instruments , Novato , CA ) . The tip of the pipette was heat-polished to a final tip resistance of 3–5 MOhm . During the experiments , the dishes with cell cultures were perfused with an extracellular solution containing: 142 mM NaCl , 3 mM KCl , 2 mM CaCl2 , 1 mM MgCl2 , 10 mM HEPES and 10 mM D-glucose , adjusted to pH 7 . 3 with 10 M NaOH . The osmolarity of solution was 290 mOsm as determined by a vapor pressure osmometer ( Model VAPRO 5520; Wescor , Logan , UT , USA ) . Experiments were done on single cells with an average capacitance of about 10 pF held at membrane potential of -60 mV . Patch electrodes used for whole-cell recording were filled with an intracellular solution containing: 145 mM NaCl , 10 mM EGTA and 10 mM HEPES; the pH was adjusted with 10 M NaOH to 7 . 2 . The osmolality of the intracellular solution was 293 mOsM . Current-voltage relations were obtained by voltage ramps from −80 mV to +80 mV twice per second and used to estimate changes in reversal potential during 10–30 s of agonist application ( Yan et al . , 2008 ) . Under a ramp protocol , cells were bathed in extracellular solution containing: 155 mM N-Methyl-d-glucamine ( NMDG+ ) , 3 mM KCl , 2 mM CaCl2 , 1 mM MgCl2 , 10 mM HEPES and 10 mM D-glucose , adjusted to pH 7 . 3 with HCl , if not otherwise stated . IVM was dissolved in dimethyl sulfoxide , stored in stock solutions at 10 mM , and diluted to required concentration in extracellular solution prior to experiments . The control , ATP-containing and IVM-containing solutions were applied via a rapid perfusion system ( RSC-200 , BIOLOGIC , Claix , France ) consisting of an array of 5 glass tubes each approximately 400 μm in diameter . The application tube was routinely positioned at about 500 μm distance and about 50 μm above the recorded cell . A complete change of the solution around the cell took between 5–20 ms , depending on the speed of the solution expelled . The one-layer Markov state model developed here ( S6 Fig ) , is a revised version of a previously developed Markov model describing P2X4R orthosteric activation by ATP and allosteric modulation by IVM [20] , whereas the two-layer model ( Fig 3 ) is more complex in nature , taking into consideration all processes involved in ATP and IVM binding . Both models assume the presence of three ATP and three IVM binding sites and were tested against current recordings to compare their performance in capturing the physiological properties of P2X4R . The following symbols were used to describe the various states of the model: C for closed , Q for conducting ( open1 and open2 ) , D for desensitized and Z for internalized states , each representing the fraction of receptors in a given state . The transition rates between the various states are in Tables A and C in S1 Text . Detailed balance was not explicitly incorporated into these models because it was assumed that P2X4R never reach absolute equilibrium during ATP and IVM stimulation . Stimulation with ATP during the pulse protocol ( repetitive stimulation with 1 μM ATP for 2 s twice per minute in the absence and presence of various concentrations of IVM applied to the bath medium after two ATP pulses ) and the prolonged protocol ( stimulation with 100 μM ATP for extended periods , longer than 1 min , in the absence and presence of 3 μM IVM ) were modeled as a square wave and a rectangular function for the duration of ATP application , respectively; stimulation with IVM was modeled as a rectangular function for the duration of IVM application . The ramp protocol was modeled as a sawtooth-like sequence of upstrokes with slope 320 mV/s ( rising from −80 mV +80 mV over 500 ms ) . Detailed descriptions of the two Markov models along with their differential equations are provided in Appendices A and B in S1 Text . Concentration-response data points were fit to a hill function y=Imax1+ ( EC50/x ) n , where y is the amplitude of the current evoked at a particular ATP concentration x , Imax is the maximum current observed at 100 μM ATP , EC50 is the ATP concentration producing 50% of the maximum current , and n is the Hill coefficient . Deactivation kinetics of the current decay after agonist washout were fitted to a single exponential y=A1exp ( −t/τ1 ) +C or to a sum of two exponentials y=A1exp ( −t/τ1 ) +A2exp ( −t/τ2 ) +C , where A1 and A2 are the amplitudes of decay for the first and second exponentials , τ1 and τ2 are their decay time constants , and C is the baseline current . In the case where the sum of exponentials fits the data better than a single exponential , we report the weighted time constant τoff=A1τ1+A2τ2A1+A2 . In either case , we labeled the derived time constant of deactivation as τoff . Statistical significance ( **p<0 . 01 and *p<0 . 05 ) was assessed using the Wilcoxon signed rank test . MATLAB ( MathWorks , Natick , MA ) was used to solve the differential equations of the models numerically , fit the models to the data and apply statistical tests . In order to quantify the rate of desensitization from current traces of the pulse protocol ( see Figs 1 and 4 ) which do not show complete desensitization , we note that , at low concentrations of ATP ( 1 μM ) , desensitization is a mono-exponential process and thus employ a mono-exponential model I ( t ) =Ae−tτ , where A is the magnitude of the current at the onset of desensitization and τ is the time constant of desensitization . By first normalizing the current and then evaluating its derivative , with respect to time , at the onset of desensitization ( t = 0 ) , we obtain I˙ ( 0 ) A=−1τ . Thus the first derivative at time t = 0 of each desensitizing current , I˙ ( 0 ) , normalized by its maximum current ( as has been plotted in Fig 1B ) , yields information about the time constant of desensitization ( i . e . , a small normalized initial desensitization rate corresponds to a large time constant of desensitization ) . Due to the simultaneous activation and desensitization of multiple receptors , I˙ ( 0 ) was estimated using a linear fit for the small window of time ( 1–1 . 5 s ) after cells achieved their maximal current and before agonist was removed . This window is relatively small compared to the 6 s time constant of desensitization for P2X4R , and therefore this approximation method provides a relevant estimate of I˙ ( 0 ) /A , which serves as a proxy for desensitization time constant τ . Parameter estimation was performed using MCMC techniques . Model simulations were generated using ode solvers in MATLAB and then fit to experimental recordings . Generally , MCMC produces Markov chains Λ = {x1 , x2 , ⋯ , xM} of model parameters xm= ( p1m , p2m , ⋯ , pNm ) , where N is the number of parameters ( pi ) and m = 1 , 2 , ⋯ , M is the mth iterate of the Markov chain . The iterates represent samples from the posterior distribution π ( x ) determined using Bayes’ theorem as follows π ( x ) ∝L ( x ) P ( x ) , where L ( x ) = P ( data | x ) , the likelihood function , is the probability of observing the data given the parameter values of x , and P ( x ) is the prior distribution of x , which reflects any prior knowledge about the parameter values independent of observed data . Proportionality , indicated by ∝ , is sufficient—therefore there is no need to normalize the posterior . In order to increase mixing of modes in parameter space , we used the parallel tempering algorithm which produces Markov chains in the product space Xm = {xm ( 1 ) , xm ( 2 ) , ⋯ , xm ( L ) } , where each chain xm ( l ) , l = 1 , 2 , ⋯ , L , was sampled from a tempered distribution πβ ( l ) and β ( l ) is the inverse temperature of each chain . Parameter sets were stochastically swapped between chains according to the swap kernel of Miasojedow et al . , and their strategy of adaptively updating the inverse temperature of each chain [33] was adopted . Because Metropolis-Hasting move-kernels can be difficult to tune for continuous-time Markov models of ion channels [34] , we used the adaptive move kernel of the t-walk sampler [35] instead . Since the t-walk samples from the product distribution π ( x ) π ( x′ ) , the composite MCMC method samples from the product distribution πβ ( X ) πβ ( X′ ) =πβ ( 1 ) ( x ( 1 ) ) πβ ( 1 ) ( x′ ( 1 ) ) ∫πβ ( 1 ) ( x ) πβ ( 1 ) ( x′ ) dxdx′×…×πβ ( L ) ( x ( L ) ) πβ ( L ) ( x′ ( L ) ) ∫πβ ( L ) ( x ) πβ ( L ) ( x′ ) dxdx′ . Given that we have a set of discretely sampled whole-cell current recordings , we initially adopted the likelihood function from Gregory [36] , defined by L ( x ) =P ( Iμ , σ|x ) =exp{−∑i=1K ( Iμ , i−Ix , i ) 22σi2} , where the index i refers to the ith discretely sampled data point , K is the number of data points in the experimental recording , Iμ , i and σi are the ith samples , respectively , of the mean current and current standard deviation estimated from the data set , and Ix , i is the ith sample of the current produced by the model given the set of parameters x . To circumvent ( i ) sampling inefficiency from the posterior distribution , which is exacerbated by the use of high data sampling rates , and ( ii ) very poor fitting of rapid transient behaviour , due to the value of the likelihood being dominated by slower portions of the signal with more data points , we opted to fit ( using the least-squares method ) both experimental and simulation data to appropriately chosen functions and to compare the fit parameters of the experimental and simulated data . For example , we have used exponential functions to measure deactivation kinetics ( as described above ) . This results in the likelihood function L ( x ) =P ( τoff , μ , στ , Aoff , μ , σA|x ) =exp{− ( τoff , μ−τoff , x ) 22στ2− ( Aoff , μ−Aoff , x ) 22σA2} , ( 7 ) where τoff , μ and στ are the mean deactivation time constant and its variance , respectively , Aoff , μ and στ are the mean deactivation amplitude and its variance , respectively , and τoff , x and Aoff , x are the deactivation time constant and amplitude produced by the model corresponding to the parameter values x . Using this description-based approach ( rather than the distance from data ) , we were able to simultaneously fit numerous aspects of P2X4R activation kinetics and their allosteric modulation . This was done by comparing experimental data and model predictions of ( i ) time dependence of the activation time and normalized rate of desensitzation in the absence and presence of 1 μm IVM ( Fig 1A and 1B ) ( ii ) maximal current , deactivation time constant , and desensitization at 1 μM ATP with increasing IVM concentrations ( Fig 4E and 4F ) ( iii ) . Insensitivity to ATP removal at 10 μM IVM ( Fig 4H ) ( iv ) decay of current amplitude deactivation time constant following IVM washout ( Fig 4A–4D ) ( v ) activation , desensitization , and recovery after washout of 100 μM ATP in the absence and presence of IVM ( Fig 5C and 5D ) ( vi ) EC50 and Hill coefficient ( n ) of the ATP concentration-response curves for peak current ( Fig 5A ) . The degree of cooperativity in ATP binding was determined from the Markov chain xm of 5000 samples associated with each ATP binding and unbinding rate ki , i = 1 , 2 , ⋯ , 24 , that was generated from data fitting , followed by calculating the chains of ATP binding affinities 3k6n+2k6n+1 , k6n+4k6n+5 , k6n+63k6n+5 , n=1 , 2 , 3 along each of the non-desensitized rows of the one-layer model ( including the four lower rows ) and the two-layer model ( including the four rows in the upper layer ) . The posterior distributions associated with these affinities were used to compare the values of the most frequently sampled points along each row ( n = 1 , 2 , 3 ) . In the presence of a specific cooperativity between ATP bindings , correlations between the different binding affinities were detected and reported . | Ligand-gated ion channels play a crucial role in controlling many physiological and pathophysiological processes . Deciphering the gating kinetics of these channels is thus fundamental to understanding how these processes work . ATP-gated purinergic P2X receptors ( P2XRs ) are prototypic examples of such channels . They are ubiquitously expressed and play roles in numerous cellular processes , including neurotransmission , inflammation , and chronic pain . Seven P2X subunits , named P2X1 through P2X7 , and several splice forms of these subunits have been identified in mammal . The receptors are organized as homo- or heterotrimers , each possessing three ATP-binding sites that , when occupied , lead to receptor activation and channel opening . The P2XRs are non-selective cation channels and the gating properties differ between the various receptors . Previously , we have used biophysical and mathematical modeling approaches to decipher the kinetics of homomeric P2X2aR , P2X2bR , P2X4R , and P2X7R . Here we extended our work on P2X4R gating . We developed two mathematical models that could capture the various patterns of ionic currents recorded experimentally and explain the particularly complex kinetics of the receptor during orthosteric activation and allosteric modulation . This was achieved by designing a computationally efficient , inference-based fitting algorithm that allowed for parameter optimization and model comparisons . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"allosteric",
"regulation",
"ion",
"channel",
"gating",
"medicine",
"and",
"health",
"sciences",
"markov",
"models",
"ions",
"enzymology",
"social",
"sciences",
"electrophysiology",
"permeability",
"neuroscience",
"simulation",
"and",
"modeling",
"ion",
"channels",
"mat... | 2017 | Deciphering the regulation of P2X4 receptor channel gating by ivermectin using Markov models |
Plant-associated microorganisms have been shown to critically affect host physiology and performance , suggesting that evolution and ecology of plants and animals can only be understood in a holobiont ( host and its associated organisms ) context . Host-associated microbial community structures are affected by abiotic and host factors , and increased attention is given to the role of the microbiome in interactions such as pathogen inhibition . However , little is known about how these factors act on the microbial community , and especially what role microbe–microbe interaction dynamics play . We have begun to address this knowledge gap for phyllosphere microbiomes of plants by simultaneously studying three major groups of Arabidopsis thaliana symbionts ( bacteria , fungi and oomycetes ) using a systems biology approach . We evaluated multiple potential factors of microbial community control: we sampled various wild A . thaliana populations at different times , performed field plantings with different host genotypes , and implemented successive host colonization experiments under lab conditions where abiotic factors , host genotype , and pathogen colonization was manipulated . Our results indicate that both abiotic factors and host genotype interact to affect plant colonization by all three groups of microbes . Considering microbe–microbe interactions , however , uncovered a network of interkingdom interactions with significant contributions to community structure . As in other scale-free networks , a small number of taxa , which we call microbial “hubs , ” are strongly interconnected and have a severe effect on communities . By documenting these microbe–microbe interactions , we uncover an important mechanism explaining how abiotic factors and host genotypic signatures control microbial communities . In short , they act directly on “hub” microbes , which , via microbe–microbe interactions , transmit the effects to the microbial community . We analyzed two “hub” microbes ( the obligate biotrophic oomycete pathogen Albugo and the basidiomycete yeast fungus Dioszegia ) more closely . Albugo had strong effects on epiphytic and endophytic bacterial colonization . Specifically , alpha diversity decreased and beta diversity stabilized in the presence of Albugo infection , whereas they otherwise varied between plants . Dioszegia , on the other hand , provided evidence for direct hub interaction with phyllosphere bacteria . The identification of microbial “hubs” and their importance in phyllosphere microbiome structuring has crucial implications for plant–pathogen and microbe–microbe research and opens new entry points for ecosystem management and future targeted biocontrol . The revelation that effects can cascade through communities via “hub” microbes is important to understand community structure perturbations in parallel fields including human microbiomes and bioprocesses . In particular , parallels to human microbiome “keystone” pathogens and microbes open new avenues of interdisciplinary research that promise to better our understanding of functions of host-associated microbiomes .
Hosts and their associated microbial communities are increasingly seen as inseparable entities ( metaorganisms ) whose ecology and evolution are inseparably entwined [1 , 2] . For example , the phyllosphere ( above-ground portions ) and rhizosphere ( below-ground portions ) of living plants are niches for myriad microorganisms that can determine the fate of plants by influencing fitness [3] and growth [4 , 5] , protecting from herbivores [6] , or driving the evolution of multidisease resistances [7] . Understanding the plant holobiont ( the plant and the organisms that live in and on it ) , therefore , will have immense implications for human food security , biodiversity [8] , and ecosystem functionality [9] . Given the broad range of microbes that colonize above-ground parts of plants such as bacteria , yeasts , filamentous fungi [10] , and protists [11] , there is poor understanding of the entire diversity of those plant-associated microbes as well as factors that shape complex plant microbial communities from host colonization to plant senescence . Current analyses point towards soil [12] and air [13] as important sources of leaf and root microbial inoculum . How defined microbial communities get selected by different plant organs from highly variable and complex inoculum communities [14 , 15] is under strong debate . Still , since plant phenotypes and fitness depend on the associated microbiome , such knowledge is critical to enable plant microbiome management , that is , reaching the full potential of using microbes and microbial communities to promote beneficial plant–microbe interactions [2 , 16] . Generally , three mechanisms contribute to microbial community structures: random colonization; species sorting by local factors ( e . g . , nutrient availability , host availability , and microbial interactions ) ; and isolating factors such as dispersion and distance [17 , 18] . Previous work has identified neutral , abiotic , and host factors that sort and contribute to differences in plant bacterial or fungal communities [13 , 19–23] . Such studies are likely to reflect adaptations of microbes that enable them to colonize specific plant host environments [24 , 25] . While these adaptations can link abiotic and biotic host factors to colonization efficiency , they cannot be understood in isolation , since the host as a holobiont is simultaneously colonized by a multitude of prokaryotes and eukaryotes [26] . Phyllosphere colonization proceeds via mechanisms that fundamentally alter the host , since some microbes participate in what can be described as niche construction . For example , many symbionts ( including pathogens ) deliver effector proteins to suppress , activate , or alter host defense [27 , 28] , and some are able to completely reshuffle host metabolism [29 , 30] . These host alterations can cause changes to microbiome structure since some microbes can take advantage of new conditions while others cannot . In fact , the niche of some microbes specifically rely on others . For example , primary colonizers can protect secondary from abiotic selection factors such as desiccation [31] or can increase secondary colonizers’ competitive advantage by providing secondary metabolites [32] . Further examples of direct microbe–microbe interactions include hyperparasitism of primary colonizers [33] and opportunists that exploit a weakening of plant defenses to colonize their hosts [34 , 35] . Such interactions explain why certain colonizers can affect establishment of even distantly related microbes on the host [32 , 36] and suggest an important role for interactions in determining microbiome structures . Most studies implicitly assume that abiotic and host factors differentiate microbial communities because of variable microbial adaptations . Research in the animal field has shown that for example , variation in the Major Histocompatibility class II ( MHC ) genotypes contributes to microbial variation among hosts [37] . In human populations , the gut microbiome is significantly influenced by the host genetics and in turn , the microbiome has a significant impact on host metabolism [38] . How microbe–microbe interactions fit in colonization models remains , however , largely unknown , not least because of limitations to the robustness and depth of taxonomic resolution . To begin to move towards a more holistic understanding of forces shaping microbiomes in general and the phyllosphere microbiome in particular , we have measured diversity and community composition of three major groups of microbes representing key branches of life ( fungi , bacteria , and oomycetes as a representative of the heterogeneous group of protists ) in both epiphyte ( surface microbe ) and endophyte ( interior microbe ) leaf compartments of individual samples . Complementary approaches of wild sampling and a common garden experiment confirmed combinatorial mechanisms of species isolation and sorting due to abiotic and host factors that manipulate A . thaliana phyllosphere microbiomes . A systems biology approach documented highly interactive “hub” microbes , and in controlled laboratory experiments we confirmed that one , Albugo laibachii , strongly affects phyllosphere communities and found evidence for direct interactions by a second , Dioszegia sp . The results demonstrate that hub microbes mediate between sorting factors and microbial colonization , effectively amplifying sorting effects in the phyllosphere and stabilizing populations of specific microbes on individual plants . Our findings provide insights into the complexity of multikingdom interactions in the phyllosphere and improve the understanding of the dynamics of plant microbiome colonization .
To identify how several factors ( Table 1 ) control phyllosphere microbiome assembly , we selected five sites near Tübingen in southern Germany with stable A . thaliana populations that have been studied for several years [39] ( WH , JUG , PFN , EY , ERG; S1 Table ) . We collected plants in the fall , covering the early growth phase of A . thaliana under short day conditions before its resting stage in winter , and in spring , just before its reproductive stage during increasingly longer days ( Experiment 1 ) . Microsatellite markers [40] confirmed that there is more A . thaliana genetic variation between sites than within sites , with no overlap of multilocus haplotypes between sites ( S2 Table ) [39] . We therefore grouped factors into “sampling time , ” which includes differences between fall and spring , and “sampling location , ” covering differences between sites such as soil , local climate , and plant genotypes ( Table 1 ) . Importantly , a major phenotype observed at all sites except PFN was the presence of white rust caused by the obligate biotrophic oomycete pathogen Alb . laibachii . From each sample , we recovered epiphytic and endophytic microbes , extracted genomic DNA , and generated six amplicon libraries: two from rRNA gene regions of bacteria ( 16S rRNA V3/V4 and V5/V6/V7 regions ) and two from each of fungi and oomycetes ( internal transcribed spacers 1 and 2 [ITS 1 and 2] of the large subunit rRNA complex ) . We included multiple amplified regions to address the fact that differences arise due to primer specificity and bias and due to differential gene region variability . Therefore , we treated the two amplified regions from a single microbial group complementarily , presenting findings generated by either dataset as well as differences between the datasets . Generally , amplicon-based microbial abundances reported are relative within each gene region . We measured how factors correlated to microbial community structure by performing constrained ordination ( canonical correspondence analysis ) on log-transformed microbial abundances . For epiphytic and endophytic bacteria and fungi , location was correlated to up to 25%–30% of community variation , and sampling time about 10%–15% ( most correlatations are significant at p < 0 . 05 based on random permutations , Fig 1A and S1A Fig ) . To further clarify variation , we calculated location- and sampling time-specific enrichment of each microbial genus based on whether it was more abundant at a specific sampling site compared to any other site or in spring or fall ( Tukey’s honest significant difference test [HSD] p < 0 . 01 , i . e . , the genus contributes to distinguishing between locations or sampling times ) . A median of one and four enriched bacterial genera per location ( endophytes and epiphytes , respectively ) suggests that relatively few species contributed to observed variation between sampling sites ( S3 Table ) . The location PFN , however , was unique because 25 and 16 bacterial genera ( endophytes and epiphytes , respectively ) were significantly enriched there ( S3 Table ) . Enrichment of many taxa at PFN explains why samples there consistently had some of the highest endophytic and epiphytic bacterial alpha diversities ( S2 Fig ) . Many fungal taxa were enriched in abundance at PFN and JUG ( 15 and 12 genera , respectively , S3 Table ) , compared to an average of 2 . 7 at each of ERG , WH , and EY . Only site PFN had significantly enriched endophytic fungal genera . Generally , bacterial locational variation was more quantitative than fungal: eight abundant fungal genera ( each > 500 total observations ) were only observed at < 5 sites , while all abundant bacteria were detectable at five or more sites ( S3 Fig ) . Sampling time was also important , with many taxa at higher abundance in fall ( 122 total taxa compared to 25 in spring ) ( S4 Table ) . The large fall/spring difference can mostly be attributed to bacteria: 16 and 14 fungal taxa were more abundant in fall and spring , respectively , while the rest of the enriched taxa were bacteria . Interestingly , while 90 taxa were more abundant in at least one sampling location and 146 at one sampling time , only two taxa were both location- and season-enriched . For both bacteria and fungi , epiphytic alpha diversity was higher than endophytic ( S2 Fig ) , and abundant genera differed between epiphytic and endophytic compartments ( S4 and S5 Figs ) . Oomycete communities presented a very different picture . Here , while sampling time still was correlated to about 10% of community variation , sampling location was correlated to 35%–80% ( depending on leaf compartment and dataset , Fig 1A , S1A Fig ) . Oomycete alpha diversity was extremely low ( S6 Fig ) and the obligate biotrophic pathogen Albugo was dominant , comprising up to 100% of observations in some samples ( S6 Fig ) , agreeing with observations of extensive white rust symptoms . Overall , we did not observe that sites physically more close to one another ( S1 Table ) were more similar in terms of observed microbial communities ( Fig 1 and S1 and S2 Figs ) . Endophytic Albugo detected via quantitative polymerase chain reaction ( qPCR ) in some of the samples scored as white rust-free was easily detectable ( S2 Table and S7 Fig ) , indicating some extent of asymptomatic endophytic Albugo growth [11] . Considering the striking symptoms on hosts affected by Albugo and its ubiquity , we decided to examine distribution of this organism at the strain level . Two Albugo species , Alb . candida and Alb . laibachii , have previously been described causing white rust on A . thaliana [41] . ITS amplicon data suggested absence of Alb . candida in our samples , and Alb . candida-specific primers confirmed this ( S2 Table ) . For strain determination , we thus focused on Alb . laibachii , using newly developed microsatellite-based markers . Although the Albugo genus was widespread , we found no strain overlap between sites but instead that each site was dominated by a stable major strain over multiple host generations ( S2 Table ) . The second most common oomycete pathogen in plants was Hyaloperonospora sp . ( Hpa ) . While only four of 19 tissue samples with observed white rust contained appreciable levels of Hpa , we found high relative levels in five of the ten tissue samples where white rust was not observed ( more present when white rust was not observed at p = 0 . 022 , one-tailed Fischer’s exact test ) . Hpa-relative abundance was not necessarily dependent on Albugo , since some of the highest observed levels were in samples with high levels of measured endophytic Albugo ( S7 Fig ) . Host genotype was not separable from location as a factor in wild samples . To determine whether it could uniquely affect microbial communities , we planted three natural A . thaliana accessions with differential resistance to Albugo sp . strains ( Ws-0 , Col-0 , and Ksk-1: see S8 Fig for qPCR quantification of endophytic Alb . laibachii Nc14 and Alb . candida Nc2 levels in susceptible versus resistant accessions ) in randomized plots in a common garden in Cologne , Germany ( CG ) and sampled their microbiomes just before flowering time ( Experiment 2 ) . Constrained analysis suggested that plant genotype affected microbial community variation , correlating to 25%–30% of bacterial and fungal community variation and 45%–55% of oomycete community variation ( Fig 1B and S1B Fig ) . Because of the low number of samples , only a few of the correlations were significant , and therefore some of the constrained variance could be due to chance . Therefore , we tested each genus for genotype-enrichment based on whether they were enriched on a specific host accession compared to any other accession ( Tukey’s HSD p < 0 . 05 , i . e . , the genus contributes to distinguishing the accession from other accessions , S5 Table ) . Indeed , even with only three samples per accession , multiple bacterial and fungal taxa were detected as enriched on each . The higher correlation of oomycete variation to host genotype was due to the low diversity of oomycete communities in the garden experiment—these were nearly completely dominated by Alb . laibachii ( S6 Fig ) . We observed significantly more white rust on A . thaliana accessions Ws-0 and Col-0 than on the partially resistant accession Ksk-1 ( S9 Fig ) , and this agreed with qPCR measurements of endophytic Albugo ( S2 Table: There was significantly less Albugo , using oomycete levels as an Albugo proxy , in accession Ksk-1 than in Ws-0 or Col-0 based on one-sided t test of 10/5/13 samples at p < 0 . 1 or p < 0 . 05 using both 5/5/13 and 10/5/13 samples ) . Additionally , we delineated three different Alb . laibachii strains in the field: the dominant strain 1 was observed on 5 A . thaliana Col-0 and Ws-0 samples , but only one Ksk-1 sample , strain 2 grew on one sample each of A . thaliana Col-0 and Ws-0 , while a third strain was found in a second A . thaliana Ksk-1 sample only ( S9 Fig ) . Taken together , our results based on the phyllosphere of the model host A . thaliana indicate that the factor’s location , sampling time , and host genotype are important determinants for plant colonization patterns of bacteria , fungi , and oomycetes . Up to ~40% of observed phyllosphere microbial community variation in constrained ordination models of wild samples could be explained by location and sampling time together ( Fig 1 and S1 Fig ) . We hypothesized that microbe–microbe interactions could contribute to the remaining variation and reasoned that the most important microbes and microbial relationships could be discovered by looking for “hubs”—highly connected microorganisms in scale-free correlation networks [22 , 42] . Therefore , we generated a co-occurrence network by measuring abundance correlations between 90 , 524 pairs of microbes grouped at the genus level ( Computational Experiment 3 , Fig 2A—important terms related to network analysis are defined in Box 1 ) . Correlations were based on samples from Experiment 1 and Experiment 2 . We did not seek to detect binary interactions ( where the presence of one microbe depends on another regardless of abundance ) , which would be distorted , because even single leaf samples pool leaf areas that are very large and diverse in terms of microbial habitats [43] . Using a cutoff that removed correlations with either a low r-squared value or that were based on microbes found only in limited samples ( see S1 Text for details ) , the resulting edges represented correlations that are widespread among locations we sampled , since most ( 86% ) were supported in at least 50 of 100 randomly subsampled datasets ( S10 Fig ) . Within kingdoms , we found that correlations were usually positive ( 86 . 5% , n = 630 ) and were dominated by interactions between bacteria . Correlations between microbes from different kingdoms were overwhelmingly negative ( 76 . 6% , n = 141 ) , driven by a disproportionate number of correlations to a few microbes ( relatively more interactions with oomycetes than random X2 = 169 . 4 p < 2 . 2 x 10−16 , S6 Table ) . We found that the cutoff used to identify “good” correlations could strongly affect identification of the most-connected microbes ( S11 Fig ) . Therefore , we used several cutoffs to identify genera with significantly ( p < 0 . 1 based on fitting a log-normal distribution ) higher betweenness centrality , closeness centrality , or degree , all of which are measures of how connected a node is in the network ( defined in Box 1 ) . Taking the intersection of significant taxa from the three connectivity parameters , genera representing each kingdom ( Albugo sp . , Udeniomyces sp . , Dioszegia sp . , Caulobacter sp . , a genus of Comamonadaceae and a genus of Burkholderiales , the last two of which could not be identified at the genus level ) were highly connected ( S11 Fig ) . We performed the same analysis on operational taxonomic units ( OTUs ) grouped at order , family , and species levels , and all genera except Udeniomyces sp . were supported at > 1 taxonomic level ( S7 Table ) . Albugo sp . was supported at all tested taxonomic levels . Hub microbes are not necessarily keystones in the community , or taxa which are responsible for significant amounts of the observed microbial community network structure [44] , but ecologically relevant hub microbes are likely to be . To check if our analysis identified keystone hubs , we computationally analyzed three of the “hub” microbes ( Dioszegia sp . , Albugo sp . , and the Comamonadaceae genus ) identified in the genus-level network ( Fig 2A ) . Together , these three hub microbes are direct correlates of most other nodes in the network ( 100 of 191 nodes , Fig 3B ) . We first generated a “spring-loaded” network view in which tightly correlated microbes form clusters ( Fig 3A ) . The main observed cluster was of high-degree epiphytic bacteria and was formed by the large number of positive correlations between them . Interestingly , most microbes with high degree were first neighbors of the three hubs and were negatively correlated to Albugo sp . or Dioszegia sp . ( Fig 3A ) , suggesting these microbes could be responsible for observed high positive correlations between many epiphytic bacteria . Next , we computationally “removed” each of the hub microbes to test their influence on network structure by building networks with partial correlations that account for their abundances . We also tested positive control keystone genera ( with high degree and low betweenness centrality ) or negative control species ( with high abundance , low degree , and low centrality ) ( Fig 3C ) . Hub microbes affected less of the network structure than positive controls , but more edges were dependent on them than on negative controls . We can conclude from this that our three main “hub” microbes are likely “keystone” species with an important role in determining network structure for the leaf microbial community . The hub microbes Albugo and Dioszegia were strongly negatively correlated to many of the bacteria in the microbial community networks but are themselves affected by abiotic and host factors . For example , Albugo is affected by host resistance encoded by single A . thaliana genes [45] , and Dioszegia , although widespread , was seasonal , being significantly more abundant in spring samples ( S4 Table ) . In light of the effects of abiotic and host factors on microbial community structure and the presence of central hubs in the microbial network , we hypothesized that there is a specific mechanism whereby microbial hubs act as “receptors” of abiotic and host factors and “regulatory units , ” amplifying or dampening effects of microbiome perturbations . To test our hypothesis , we examined the effect of the presence of isolates of Albugo ( Experiment 4 ) and Dioszegia ( Experiment 5 ) on other phyllosphere microbiota . Axenic isolation of Albugo is difficult because of its obligate biotrophic lifestyle , but several characteristics make this a good model system for testing our hypothesis . First , Albugo is easily propagated with associated microbes that would also be propagated in nature ( i . e . , spore- or leaf-associated microbes ) by washing infected leaves ( where infection refers to susceptible plants treated with live Albugo spores and visible white rust ) and reinoculating . Second , its presence is easily controlled independently of other microbes by introducing a resistant host . We selected two strains ( Alb . laibachii Nc14 and Alb . candida Nc2 ) that had been propagated continually for > 1 yr , giving any associated microbiome time to acclimate to lab conditions . We individually inoculated both strains using leaf washes ( containing the Albugo strain spores and the strain-associated microbial community ) onto the three host accessions that we had used for the garden experiment . As expected , susceptible plants displayed strong symptoms , while resistant plants were asymptomatic and had very low detected levels of Albugo sp . ( S8 Fig ) . To additionally simulate an abiotic factor limiting Albugo ( e . g . , a distribution limitiation in the wild ) , we included a second set of Albugo-free controls by removing Albugo spores from leaf washes by filtering ( < 6 μm ) . This set of controls also allowed us to account for noise in controls due to host genotype background . Filtering could have affected the abundance of other microbes , so we confirmed observed trends in one replicate of an experiment in which Albugo filtering was replaced by inactivation using a combination of the oomycete inhibitors metalaxyl and benalaxyl . In all cases all tested conditions were grown together in growth chambers and the communities were allowed to adapt over two cycles of reinoculation before sample collection for microbial community profiling ( S12 Fig ) . If Albugo is indeed a “hub” that transmits , e . g . , host factors , affecting colonization of many microbes , we expected the following: 1 ) decreased alpha diversity as a consequence of infection ( following from the observed strong negative correlations to many bacterial taxa ) , 2 ) less variability between replicates of infected plants ( since other microbes in the inoculum were cocultivated with Albugo and many are presumably reliant on its presence ) , 3 ) divergence of control from infected communities , and 4 ) stronger differences between genotypes in the presence of Albugo . Compared to the resistant host A . thaliana Ksk-1 , epiphytic bacterial communities on Alb . laibachii-infected plants had significantly lower alpha diversity ( Fig 4A , S13 and S14 Figs ) and significantly more similar beta diversity between replicates ( within-replicate distance , Fig 4A and S15 Fig ) . Bacterial communities on plants with Alb . laibachii infection were more similar to each other than to uninfected controls , although this effect was mostly apparent in the bacterial V3/V4 dataset ( between-treatment distance , Fig 4A and S16 Fig ) . Effects with abiotic A . laibachii control ( regardless of filter removal or chemical inhibition ) were the same but less significant than due to host resistance ( filtering [Fig 4A and S13 , S15 and S16 Figs]/chemical inhibition: [S14 Fig]—significance for within-replicate and between-treatment distances were stronger for the V3/V4 dataset ) . Alpha diversity or within-replicate distance differences between the three A . thaliana accessions were strongly increased in the presence of active Alb . laibachii ( S13 , S14 and S15 Figs ) , confirming that Alb . laibachii can amplify host genotype-specific bacterial community differences . The effect of Alb . laibachii on fungal communities were less consistent and not as clear . Most apparent was a slight depression in fungal alpha diversity with infection , but without statistical significance ( Fig 4A and S13–S16 Figs ) . For Alb . candida infections , a similar and consistent trend of relatively low bacterial and fungal alpha diversity was observed on infected A . thaliana Ws-0 ( Fig 4A , S13 and S14 Figs ) , but it was not significant . We also did not observe more similar bacterial or fungal communities between replicates or between treatments with Alb . candida infection ( Fig 4A , S15 and S16 Figs ) . These results , combined with much higher numbers of epiphytic bacteria on Alb . candida-infected leaves than Alb . laibachii ( S17 Fig ) suggests a comparatively weak impact on the bacterial community in A . thaliana caused by Alb . candida infection . Not only epiphyte communities were disturbed by Albugo infection . Numbers of endophytic bacterial or fungal reads ( which have been used as a proxy for the amount of endophyte microbes [22] ) were lower in the absence and higher in the presence of Albugo ( S18 Fig ) . We identified a subset of bacteria that were significantly enriched as endophytes during infection but not in any control , and most of them only due to Alb . laibachii infection , indicating that Albugo enabled their colonization of endophytic space ( Fig 4B and S18 Fig ) . These bacteria were not likely to simply have been found because of higher epiphyte numbers in infected plants , since not all abundant taxa were enriched during infection ( S19 Fig ) . Next , we tested the hub fungus Dioszegia sp . , which we isolated from the endophytic compartment of A . thaliana at site EY . Unlike Albugo , Dioszegia can be axenically cultivated , making possible tests of direct , one-on-one interactions with other microbes in the phyllosphere . In short , we spray-inoculated 3-wk old axenically-grown A . thaliana seedlings with Dioszegia sp . After 3 d , we inoculated isolates of one of six bacterial genera ( all of which were isolated on or near A . thaliana and which we observed in phyllosphere samples , S8 Table ) . Colony forming units ( CFUs ) of Dioszegia and the bacteria were counted at the starting time and after one week of coculture ( S20A and S20B Figs ) . For four isolates ( Janthinobacterium , Caulobacter , Flavobacterium , and Agromyces ) , negative correlations to Dioszegia had been observed in the network analysis , while for two isolates ( Pseudomonas and Rhodococcus ) we had observed no correlation ( S20C Fig ) . Of the latter two , only Rhodococcus , an isolate from Alb . laibachii Nc14 spores , interacted by reducing Dioszegia growth . Rhodococcus generally grew to high epiphytic abundances in lab conditions ( S21 Fig ) , and thus reduced growth was probably due to spatial competition . Of the other four isolates , Janthinobacterium did not survive on the leaf and we did not observe any effect of Flavobacterium . Agromyces caused slightly reduced Dioszegia growth , but itself grew poorly in the phyllosphere . Of the negatively correlating genera , the Caulobacter isolate grew the best alone on plants and was strongly inhibited by Dioszegia ( ~100-fold lower CFU counts ) . Caulobacter was also identified as a hub at the genus and species level ( S7 Table ) . Taken together , our findings confirm that the microbial “hub” Albugo is a strong interactor in the phyllosphere , and that its presence limits alpha diversity and affects plant microbial communities . They also support the hypothesis that Albugo could stabilize plant microbial communities , for example on hosts in a single wild population . Negative correlations between the fungal hub Dioszegia and bacteria in the phyllosphere are due to both antagonistic effects by other bacteria on Dioszegia ( e . g . , due to spatial competition ) and direct antagonism on specific bacteria . Therefore , host or abiotic signatures that affect the abundance of the hubs Albugo and Dioszegia can also have disproportionately large effects among phyllosphere microbiota . To look closer at the mechanism of how hub microbes select phyllosphere microbiota , we asked which endophytic taxa were enriched in the field in samples with high measurable levels of endophytic Albugo sp . , and whether these were also enriched in lab experiments . Generally , in wild samples , no single bacterial genus dominated endophytes . Several taxa identified at the genus level were enriched ( >10% of reads ) in individual samples , including Pseudomonas ( up to 93% of reads at many sites ) , Sphingomonas ( up to 29% of reads at Cologne , ERG , and EY ) , Methylobacterium ( up to 22% of reads at Cologne and PFN ) , Deinococcus ( up to 12% of reads at Cologne ) , and Flavobacteria ( 10% of reads JUG ) ( S4 Fig ) . Of these , Pseudomonas sp . was also the genus that colonized the endophytic compartment during Alb . laibachii Nc14 infection most efficiently ( Fig 4B ) . Therefore , while specific bacterial genera seem to benefit from Albugo infection , these seem to be location-specific rather than Albugo-specific . Interestingly , we calculated interkingdom correlations of microbes to alpha diversity indices and found that the hub taxa Albugo and Dioszegia are strongly negatively correlated overall to bacterial diversity , as are two other epiphytic yeast-like fungal genera ( Leucosporidiella and Udeniomyces—the latter was identified as a hub at the genus level , S7 Table ) and a genus of Pleosporales fungi ( S22 Fig ) . Only the epiphytic fungal genus Heterobasidion was positively correlated with bacterial diversity based on support from both bacterial amplicon datasets; in addition , one dataset supported positive correlations for several other fungal classes and the genus Aspergillus . Additionally , several epiphytic bacterial classes positively correlated with fungal epiphyte diversity and endophytic Pseudomonas negatively correlated to it ( S22 Fig ) . Negative correlations of hubs to bacterial diversity ( also observed in lab experiments for Albugo sp . ) correspond to the network observation of extensive negative correlations to epiphytic bacterial genera . Thus , as hubs , Albugo and Dioszegia decrease bacterial diversity and thereby increase relative abundances of a few groups of abundant and location-specific bacteria . Significant correlations of other genera to alpha diversity of bacteria and fungi suggest that other bacterial and fungal taxa are also important and will be detected as hubs with different sampling strategies ( e . g . , within single host populations ) . Besides affecting relative abundances of specific bacterial groups , hub microbe abundance is itself affected by abiotic or host factors like climate , distribution , or host resistance alleles . Therefore , we used constrained ordination to ask to what extent external factors and microbial hubs are responsible for observed beta diversity variation . The external factors location and sampling time together correlated to ~40% of total epiphytic or endophytic bacterial variation . The hub microbes Albugo sp . and Dioszegia sp . together correlated to about 15%–20% of variation ( Fig 5 ) . External factor and hub microbe effects were not completely independent , since up to 34% of variation correlated to external factors overlapped with variation correlated to hub microbes ( ~14 . 3% of 41 . 8% for bacterial epiphytes based on V5/V6/V7 amplicons , Fig 5 ) . Therefore , the external factors location and sampling time have important independent effects on phyllosphere microbiome structures , but up to one-third of their observed effects could be due to variation of two hub microbes .
Evidence has mounted that the holobiont is the unit on which evolutionary selection acts , but a full understanding of this concept , especially in plants , is missing complete explanations of how the metaorganism forms and is structured [1 , 2] . To elucidate principles enabling identification of mechanisms relevant for formation of the microbial fraction of the plant holobiont , we have generated an unprecedented high-resolution microbiome “map” showing a significant impact of biotic and abiotic factors . Analysis of three of the most important phyllosphere taxa ( oomycetes , fungi and bacteria ) addresses a lack of data with a broad taxonomic resolution which has prevented identification of specific mechanisms of microbial community differentiation . Our results suggest that mechanisms contributing to observed abundances are taxa-specific and are mediated by complex interactions between abiotic factors and taxa , between taxa and the host , and between multiple taxa . Sampling location ( correlating to ~25%–35% of community variation ) and sampling time ( season , correlating to ~10% of community variation ) were correlated to robust patterns of diversity variation caused by taxa that were not evenly distributed among sampling sites or times or which were completely specific to certain sites . Amplicon sequencing results in taxonomic resolution below the strain level , so microorganisms might be even less evenly distributed than our data suggests . This is illustrated by the oomycete genus Albugo , which was dominant and widely dispersed but which had site-specific strains that might be adapted to local conditions or hosts . Consistently , previous work on both phyllosphere and rhizosphere bacterial communities showed that location is a strong determinant of microbial community structures , which then vary to a lesser extent between different host species and genera [19 , 46] . Site-localized microbial taxa could result from both poor dispersal between sites ( e . g . , uneven distribution of microbial inocula like different soil conditions which differentially inoculate plants [12 , 14] ) or local sorting mechanisms that completely exclude species in specific locations ( e . g . , local conditions or host plant effects [47 , 48] ) [17] . Indeed , colonization of the phyllosphere and rhizosphere has been suggested to proceed with an ordered effect of inocula distribution followed by species sorting [12 , 49] . Distribution , dispersal , and species sorting to some extent go hand in hand , since strong adaptation of microbes to specific hosts have reproductive costs [50] and in some cases can limit their transmissibility [51] . In our study , it was clear that species sorting occurred at least in part due to hosts . For example , abundance of the genus Albugo was reduced ( in the CG experiment ) or eliminated ( in lab experiments ) due to ( partial ) resistance in the accession A . thaliana Ksk-1 . We also observed much lower endophytic bacterial and fungal diversity than epiphytic caused by the “gateway” between the leaf surface ( epiphytes ) and interior ( endophytes ) . This level of sorting occurs since endophytes and pathogens need to specialize and coevolve with hosts [52 , 53] to avoid or evade an arsenal of host self-defense mechanisms such as callose deposition [54] , antimicrobial peptides [55] , and reactive oxygen species ( ROS ) bursts [56] . Recent studies seeking to more generally connect host adaptation to microbiota colonization have utilized mutant plants [57 , 58] and genome-wide association studies ( GWAS ) [59] to demonstrate that plant genotypes sort their associated microbiota . Most direct allele or host accession effects , however , have only been minor and on specific taxa [21] . Comparatively , we observed significant effects on many diverse taxa in the CG experiment in this study , raising the question of what leads to broader host genotype impacts on microbial colonization . We proposed that one mechanism can be via microbe–microbe interactions . For example , despite high A . thaliana diversity between sites , we observed the genus Alternaria inside plant samples at almost all sites . We did not expect this because it is a plant pathogen with a necrotrophic lifestyle and host specificity on certain A . thaliana accessions [60] . This could be indicative of diverse strains with different host adaptations ( i . e . , broad compatibility as a genus ) , but an effectively expanded host range could also result from taking advantage of already broken down host barriers . For example , wide cooccurrence of Alternaria has been observed with Alb . candida [61] . We propose that microbe–microbe interactions generally increase host effects due to the community correlation network topology we observed in which many microbes are weakly connected , while only a few “hubs” are highly connected , dominant interactors . In other words , many genotype effects ( or other factors ) will only perturb the activities of less influential microbes . If , on the other hand , an external pressure “hits” a hub microbe , the disturbance can be expected to cascade through the microbial community ( Fig 6A ) . In this study , Albugo , the causal agent of white rust , was identified as an important hub . To show experimentally its hub status , we performed microbial “knockout” experiments in a CG experiment and in the lab , by introducing a range of different A . thaliana accessions carrying functional resistance alleles [45 , 62 , 63] or by physical/chemical removal/inhibition of Albugo in the lab ( simulating an abiotic elimination of Albugo infection ) . We could show that , regardless of how Albugo is removed from the system , the associated microbial community is more stable in the presence of the pathogen and significantly changes in its absence . Albugo functions as a hub from the “bottom up” by limiting bacterial diversity and increasing relative abundance of major taxa in the phyllosphere . This supports the hypothesis that it also stabilizes abundance of site-specific taxa in the wild since hubs will promote deterministic host-associated taxa selection at affected sites ( either directly or by modifying host phenotypes ) ( Fig 6B ) . On the other hand , community stability in the absence of major hubs is functionally based and occurs from the top down such that many observed taxa vary stochastically ( Fig 6B ) . Here , abundant taxa are the target of perturbations that eliminate them or reduce their abundance and rare community members are required to fill the resulting open niches or functional voids [64] . Therefore , Albugo absence can plausibly explain strongly differentiated bacterial communities at the wild site PFN and significant accession-correlated microbial community differentiation in our garden experiment . With the apparent importance of specific microbes for local species sorting in the phyllosphere , the question arises as to what makes these microbes hubs ? One possibility is that they can exert strong indirect effects on other microbes via the host . On plants where host genetic diversity is a result of selection under pathogen pressure [65] , pathogens can cause phenotypic expression of that diversity . At least some of the observed effects by Albugo sp . probably occur in this way , since it has been shown to , for example , alter host metabolism [30] , which could lead to community differentiation . Therefore , from the point of view of its transformative effects on the host , the hub status of Albugo is not too surprising . Other more specific antibiosis selection mechanisms such as direct interactions and inhibition likely occur through , for example , ecological effectors [52 , 66] . No such pathways have been identified in Albugo genomes [67] , but single protein effectors cannot be excluded . Our results suggest that the hub microbe Dioszegia directly inhibits some taxa , since it affected colonization efficiency of only specific bacteria . Not only Dioszegia but also other basidiomycetous yeasts [68] can directly interact with microbiota , and these are also likely to operate as microbial hubs . Plant-associated species like Rhodotorula and Pseudozyma , for example , are known for compounds secreted that are effective in “biocontrol” of bacteria and fungi , respectively [69 , 70] . Such direct effects would then be expected to cascade through the interconnected community . Thus , hub microbes can influence diversity by acting indirectly via the host or directly on colonization efficiency of other microbes . Both indirect ( via host [71] ) and direct effects ( via metabolites [72] ) have been suggested for mechanisms of action of some microbes that cause abnormal human microbiomes . The hub status of these pathogens ( and the dysbiotic microbiota they cause ) is suggested to benefit them by promoting disease in the host [73] . Due to their disproportionally high impact on the metacommunity , these hubs are called “keystone” species [74] . Therefore , being a hub with a high level of “keystoneness” [44] , as we have detected specifically for Alb . laibachii may be a critical part of host colonization . This might explain why Alb . candida was absent on wild A . thaliana ( even where compatible strains were found on nearby Capsella sp . plants ) since it was only weakly able to control A . thaliana microbial communities in lab experiments . Therefore , while hub interactions can occur indirectly through the host , where benefits of being a hub microbe can be identified , researchers should consider that strong selection exists for the ability of hubs to directly select cocolonizing microbiota . Not all pathogens , however , share the hub microbe or even keystone status , so pathogenicity cannot be taken as a rule to detect “hub” or “keystone” species . The second most abundant oomycete genus that we recorded on A . thaliana was another obligate biotrophic pathogen , Hpa . Although Hpa was common , it was not a hub at the broad geographic and host diversity scales that we tested since it was not a strong interactor in the network and did not significantly correlate to bacterial diversity ( S7 Fig , adjusted r2 value of 0 . 27 for correlation to epiphytic bacterial diversity ) . We still cannot exclude that Hpa might act as a hub within a specific A . thaliana population or by interacting with an A . thaliana genotype not in our survey . However , we hypothesize that the lack of hub status across broad scales compared to Alb . laibachii reflects fundamental differences in the biotrophic strategies of the pathogens . For example , a disproportionate number of hybrid incompatibility ( HI ) loci in A . thaliana encode leucine-rich repeat containing ( NLR ) resistance proteins to Hpa [65 , 75] . This evidence of active , strong selection at HI loci suggests that Hpa must have significant consequences for host fitness . While alleles are known that encode NLR proteins conferring resistance to Alb . laibachii [63 , 76] , there are comparatively few , and none have been implicated as HI loci . Thus , we hypothesize that the “hub” characteristic of Alb . laibachii that leads to a low diversity and a stable phyllosphere microbiome is part of an “under-the-radar” approach to biotrophy . We therefore hypothesize that pathogens like Hpa , which thrive by participating in an extremely active evolutionary “arms race , ” should exhibit less microbiome control . Dioszegia and Albugo were functionally redundant with regard to decreasing bacterial diversity . Functionally redundant hubs in networks are characteristically stable , because the loss of one hub minimally interrupts function [77] and so this may suggest a relationship between these organisms . Interestingly , even after many generations of almost continuous subculturing in the lab ( > 8 yr [67] ) of Alb . laibachii Nc14 , the basidiomycetous yeast Pseudozyma sp . is by far the most abundant associated fungus ( S23 Fig ) . Associations of basidiomycetous yeasts including Dioszegia with other eukaryotes on plants such as arbuscular mycorrhiza fungi ( AMF ) and their spores [78] is known . Therefore , a close association and even beneficial relationship could exist between yeast and Albugo by limiting growth of complementary sets of microbes . Other relationships may also exist: other hub microbes ( e . g . , Caulobacter and a Comamonadaceae genus ) seemed to have opposite effects on phyllosphere bacterial diversity . In at least one study , host manipulation of bacterial diversity has been suggested to affect its resistance to pathogens [79] . Thus , diversity manipulation might be a key battleground where hosts and various hubs cooperate or compete with one another . In this case , hubs with complementary or opposite microbial community functions are attractive targets for biocontrol studies in plants . Since pathogens have been identified as influential hubs in human hosts as well [71 , 72] , a similar approach can be used there to find new targets for disease therapies . Taken together , our results demonstrate that phyllosphere colonization by bacteria , fungi , and oomycetes is determined by various mechanisms of species sorting . These include seasonal effects , partitioning between epiphytic and endophytic leaf compartments , and host genetic differences . Most effects so far attributed to these factors have implicitly assumed their direct effect on microbes . Our broader-resolution study , however , strongly suggests that “hub” microbes are important intermediaries between abiotic , temporal , and host factors and colonization of many other microbes in the phyllosphere . Although previous studies have postulated the existence of keystone microbes in the phyllosphere [80 , 81] or suggested their existence based on bacterial network analyses [82 , 83] , this is the first study to identify and confirm hubs from various kingdoms , to show their effects across kingdoms , and to identify hub microbes as direct targets of abiotic or host factors and mediators of observed microbiome variation . Because of complementary or antagonistic functions of these hubs , their resolution in plant , human , and other host contexts will improve understanding of what a holobiont is and how it functions . Specifically , if indeed hubs select cocolonizing microbiota to improve their own fitness , the host holobiont has to be understood in the context-colonizing hubs which themselves are holobionts . For example , host–pathogen coevolution can be expected to occur both on the molecular and microbial level . Thus , identifying hub interactions will reveal central targets to quickly revolutionize how we understand host–microbe–microbe relationships and to enable better future management of plant microbiomes—a crucial tool for biocontrol and resource saving food security .
We selected five sites near Tübingen in southern Germany for collection from wild populations of A . thaliana ( S1 Table ) . These sites were selected because plants grew in open conditions in discrete populations with minimal disturbance from other plants such as grasses . Genotypic diversity within these populations was previously studied [39] . At two time points , in the spring and fall ( 5/7/13 and 11/26/13 ) , we harvested several samples from each site . Because white rust caused by Albugo sp . was an extremely common phenotype at most sites , we recorded whether or not it was observed on collected plants . Where we recorded visible white rust ( S2 Table ) , all leaves in the pool had visible white rust . Commonly , plant leaves were extremely small , in which case we pooled leaves from multiple plants , and otherwise we pooled multiple collected leaves from single plants ( S2 Table ) . When otherwise healthy plants had leaves that were extremely dirty or where >50% of the leaf area exhibited lesions ( most likely through mechanical wounding , insects or other factors ) , these were avoided . In the garden experiment , three A . thaliana accessions ( Ws-0 , Col-0 , and Ksk-1 ) were planted in nine plots . Each plot consisted of 30 plants , 10 of each accession , that were ordered randomly in 5 rows and 6 columns . The plants had been germinated from sterile seeds sown on Jiffy seed pellets ( Jiffy Products International BV ) , initially watered with 2 mL / 1 L of WuxAl Liquid Foliar nutrient ( AgNova Technologies Pty Ltd ) . After 10 d in a long-day greenhouse ( 12-h light / 12-h dark ) , when the plants had the second set of true leaves , the peat pellets and plants were transferred to the field site on 10/18/12 . On 5/5/13 and again on 5/10/13 , three leaf samples ( pooled leaves from single plants ) from rosettes of different A . thaliana Col-0 , Ws-0 and Ksk-1 plants were harvested ( see compartmentalization protocol below ) . We harvested from plants from various locations in the nine plots , selecting plants that were setting seeds but on which visible symptoms of senescence were not observed . We conducted experiments with the lab strains of Alb . candida Nc2 and Alb . laibachii Nc14 that had previously been kept growing on A . thaliana Ws-0 or Col-TH0 , respectively , > 1 yr . In short , leaf washes from infected A . thaliana Ws-0 plants were collected and they or controls in which spores were either filtered from the solutions or chemically inhibited were sprayed on the A . thaliana accessions Ws-0 , Col-0 , and Ksk-1 ( 2 pathogens x 2 inocula x 3 hosts = 12 treatments ) . In each experimental replicate , all treatments were kept together in the same growth chamber under identical conditions . After 12 d , leaf washes were collected from all 12 treatments and were used to reinoculate a second round of plants . After another 12 d , the epiphytic ( leaf surface ) and endophytic ( intra/intercellular ) microbial communities were recovered from collected leaves from each of the 12 treatments ( S12 Fig ) . The infection experiments were performed in three replicates . Further details can be found in the supporting materials and methods ( S1 Text ) . Interaction between Dioszegia sp . and individual bacterial isolates was observed on A . thaliana Ws-0 seedlings grown under sterile conditions on 1/2 MS media . Bacteria and Dioszegia ( see S8 Table for strain information ) were grown in liquid 10% TSB and PD media [84] until they reached an OD600 of 0 . 6 . The microbes were pelleted , suspended in 10 mM MgCl2 , and 200 μl microliter were sprayed on the individual plants using an airbrush pistol ( Conrad Electronics , Germany ) . Three-week-old seedlings were sprayed with Dioszegia , and 2 d later the bacterial isolate was sprayed . After one week , leaf discs ( 0 . 07 cm2 ) were punched out from single leaves , crushed with a mortar and pestle and suspended in 50 μl of water . The CFU's for bacteria and Dioszegia were determined by growing on 10% TSB plates containing Nystantin and PDA plates containing antibiotics , respectively . From each leaf sample ( wild collection , garden experiments , and lab experiments ) , leaf epiphytic and endophytic microorganisms were collected using the same protocol . In short , the collected leaves in a 15 mL tube were first rinsed with water by gentle agitation for 30 sec , from which an aliquot was taken and stored . Next , 3–5 mL of epiphyte wash ( 0 . 1% Triton X-100 in 1x TE buffer ) was added to the tube , agitated for about 1 min , and filtered through a 25 mm , 0 . 2 μm nitrocellulose membrane filter ( Whatman , Inc ) . The filter containing epiphytic microorganisms was placed in a screw-cap tube and frozen in dry ice . Next , the same leaves were surface sterilized first using 15 sec washes of 80% ethanol followed by 2% bleach ( sodium hypochlorite ) . Leaves were then rinsed three times with sterile autoclaved water and the resulting leaves containing endophytic microorganisms were frozen on dry ice for further processing . We extracted DNA with a custom protocol and prepared amplicon libraries for ten samples from each of two wild collection events ( always two samples from WH , two from ERG , three from EY , two from JUG , and one from PFN ) and three samples of each A . thaliana accession from the garden experiment collected on 5/10/2013 . From the controlled lab experiments , we prepared libraries for triplicates of each of 12 treatments . In total , bacteria , fungi , and oomycete amplicon libraries were prepared from 65 epiphyte and 65 endophyte leaf fraction samples ( see S1 File for samples and index sequences ) . A two-step amplification protocol was implemented , and the first step was prepared in triplicate . Primers consisted of a concatenation of the Illumina adapter P5 ( forward ) or P7 ( reverse ) , an index sequence ( reverse only ) , a linker region , and the base primer for the region being amplified . For each region , we used 20 different reverse primers that were identical except for the 12-bp index [85] that would be used later to identify sequencing products in combined libraries . Information for all primers used can be found in S2 File . Amplicon libraries were quantified fluorescently , and products of 120 amplicon libraries ( the six targeted amplicon regions from epiphyte and endophyte templates from ten samples ) were combined in equimolar concentrations in seven combined libraries . The combined libraries were concentrated and quantified via qPCR and were sequenced on an Illumina MiSeq lane using a mixture of custom sequencing primers complementary to the linker/primer region of the concatenated primers ( S2 File ) . Sequencing was performed for 500 cycles to recover 250 bp of information in the forward and reverse directions . Additional details can be found in supporting materials and methods ( S1 Text ) . Raw sequence data is publicly available online through MG-RAST project number 13322 [http://metagenomics . anl . gov/linkin . cgi ? project=13322] . We developed a custom pipeline to simultaneously process reads from bacteria , fungi , and oomycetes for downstream analysis . In short , for data from each Illumina lane , we de-multiplexed and quality filtered reads , split sequence files into the six amplicon groups , and separated reads that were still paired or were orphans after filtering . We then trimmed adapter sequences and aligned paired reads . Next , we placed all aligned paired , unaligned , and orphan reads together and checked for chimeras then combined reads from which the first 125 or last 125 bases were identical ( since all orphan reads were at least this long ) . We then combined the prefix/suffix combined reads from all sequencing runs and picked OTUs at 97% similarity and picked representative sequences for each OTU . Finally , OTUs were assigned taxonomy , and filters were applied to remove low abundance OTUs and nontarget amplicons . For downstream analyses , OTU tables were rarefied to an even depth of reads per sample and summarized to a specific taxonomic level ( usually genus except where noted ) . More details on softwares used and processing parameters can be located in supporting materials and methods ( S1 Text ) . Data and code used to generate the main figures in the text are being made available on GitHub ( https://github . com/magler1/HubMicrobes ) . For further details on downstream statistical analyses and other details not included in the main text , please refer to the Supporting Materials and Methods ( S1 Text ) . | Under natural conditions , plant growth and behavior strongly depend on associated microbial communities called the microbiome . Much research has been performed to evaluate how the environment and plant genes help to determine the structure of the microbiome . Here , we show that interactions between microorganisms on plants can be responsible for large portions of observed microbial community structures on leaves . Importantly , particular microbes , termed “hub microbes” due to their central position in a microbial network , are disproportionally important in shaping microbial communities on plant hosts . We discovered fungal and oomycete hub microbes that act by suppressing the growth and diversity of other microbes—even across kingdoms—and several candidate bacterial hubs , which largely positively control the abundance of other bacteria . We also showed that factors impacting the microbial community—such as plant genotype—are strongest if they affect colonization of a hub microbe because the hub in turn affects colonization by many other microbes . Our results further suggest that hub microbes interact directly or via the microbial community . Hub microbes are thus promising targets for better understanding the effects of host genomic engineering and for future work in controlling disease-associated and beneficial host-associated microbial communities . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"plant",
"anatomy",
"oomycetes",
"medicine",
"and",
"health",
"sciences",
"ecology",
"and",
"environmental",
"sciences",
"microbiome",
"pathology",
"and",
"laboratory",
"medicine",
"community",
"structure",
"brassica",
"microbiology",
"fungi",
"plant",
"science",
"model... | 2016 | Microbial Hub Taxa Link Host and Abiotic Factors to Plant Microbiome Variation |
Granule cells ( GCs ) are the major glutamatergic neurons in the cerebellum , and GC axon formation is an initial step in establishing functional cerebellar circuits . In the zebrafish cerebellum , GCs can be classified into rostromedial and caudolateral groups , according to the locations of their somata in the corresponding cerebellar lobes . The axons of the GCs in the caudolateral lobes terminate on crest cells in the dorsal hindbrain , as well as forming en passant synapses with Purkinje cells in the cerebellum . In the zebrafish mutant shiomaneki , the caudolateral GCs extend aberrant axons . Positional cloning revealed that the shiomaneki ( sio ) gene locus encodes Col4a6 , a subunit of type IV collagen , which , in a complex with Col4a5 , is a basement membrane ( BM ) component . Both col4a5 and col4a6 mutants displayed similar abnormalities in the axogenesis of GCs and retinal ganglion cells ( RGCs ) . Although type IV collagen is reported to control axon targeting by regulating the concentration gradient of an axonal guidance molecule Slit , Slit overexpression did not affect the GC axons . The structure of the BM surrounding the tectum and dorsal hindbrain was disorganized in the col4a5 and col4a6 mutants . Moreover , the abnormal axogenesis of the caudolateral GCs and the RGCs was coupled with aberrant BM structures in the type IV collagen mutants . The regrowth of GC axons after experimental ablation revealed that the original and newly formed axons displayed similar branching and extension abnormalities in the col4a6 mutants . These results collectively suggest that type IV collagen controls GC axon formation by regulating the integrity of the BM , which provides axons with the correct path to their targets .
The cerebellum is involved in various brain functions , including motor coordination and motor learning [1–3] . Since the structure of the cerebellum is basically conserved among vertebrates [4 , 5] , the zebrafish cerebellum provides a good model for understanding cerebellar development and functioning [6–8] . In both the mammalian and zebrafish cerebellum , granule cells ( GCs ) are the major glutamatergic neurons . The teleost cerebellum contains at least two different types of GCs ( Fig 1 ) that have different locations and developmental processes and contribute to distinct neural circuits [9–11] . The GCs in the rostromedial lobes , the valvula cerebelli ( Va ) and corpus cerebelli ( CCe ) , form a layer that is deep to the molecular layer . These GCs are derived from neuronal progenitors located in the rostral part of the rhombic lip , which migrate ventrally . Each GC sends its axon to meet the dendrites of Purkinje cells ( PCs ) in the molecular layer . In contrast , the GCs in the caudolateral lobes , the eminentia granularis ( EG ) and lobus caudalis cerebelli ( LCa ) , are derived from neuronal progenitors in the caudal and lateral parts of the rhombic lip , and their somata lie superficial to the molecular layer . They send their axons to PCs in the cerebellum , and also extend them caudally to the dendrites of crest cells , which are Purkinje-like cells , in the dorsal hindbrain . While rostromedial GC circuits involving PCs are likely to be involved in motor learning and classical conditioning , the caudolateral GC circuits , involving both PCs and the crest cells are may control motor coordination in response to vestibular information [12 , 13] . The mechanisms responsible for the formation of these two distinct GC circuits remain unknown . The extracellular matrix ( ECM ) controls neural circuit formation in various ways [14] . Collagen proteins are widely expressed in the ECM of the developing nervous system and its surrounding tissues [15 , 16] , and have been suggested to control axon extension and axon guidance in vertebrates and invertebrates [17–21] . Among the collagens , type IV collagen is a heterotrimeric protein complex , whose protomers can include various combinations of subunits ( Col4a1-6 ) ; this complex is a component of the basement membrane ( BM ) that lines epithelial cell sheets [22] . In humans , type IV collagen gene mutations , e . g . , COL4A5 , lead to Alport syndrome [23] , of which most symptoms , including renal dysfunction and auditory disturbance , are attributed to defects in the BM structure [22 , 23] . The zebrafish col4a5 mutant , dragnet , shows abnormal targeting of the retinal ganglion cell ( RGC ) axons; in wild-type animals , individual retinal axons project to single layers in the optic tectum , while the axons in dragnet aberrantly pass between layers or terminate on multiple layers [24] . Because type IV collagen binds to the axonal guidance molecule Slit1 [25] , one proposed explanation for the aberrant RGC axogenesis in these mutants is the presence of an abnormal Slit concentration gradient in the optic tectum [25] . In addition to regulating guidance molecules , collagens regulate neurite growth through integrin-family receptors or discoidin domain receptor 1 ( DDR1 ) [26–29] , and they control cell migration or tissue morphogenesis through adhesive G protein-coupled receptors ( aGPCRs ) [30 , 31] . However , it is still largely unknown how collagens control the axogenesis of individual types of neurons . We previously isolated zebrafish mutants with defects in the formation of cerebellar neurons and their neurites [9] . Among them , shiomaneki ( sio ) larvae show aberrant axogenesis of the GCs in the caudolateral lobe . In this study , we identified col4a6 as the causative gene for the sio mutation , and demonstrated that type IV collagen controls GC axon formation by regulating the integrity of the BM .
First , to reveal the normal axonal structures of zebrafish GCs , we performed single-cell labeling of GCs using GC-specific Gal4 Tg lines , gSA2AzGFF152B and gSAIGFF23C [32] . We took two approaches for mosaic labeling: the use of a partially silenced UAS reporter line and the transposon Tol1-mediated incorporation of reporter DNA [32] . Larvae were obtained from a cross between the gSA2AzGFF152B line and the partially silenced UAS:Kaede reporter fish ( Fig 1A-1C ) or by injecting UAS:Kaede reporter DNA and Tol1 transposase RNA into cleavage-stage embryos of the gSAIGFF23C line [32] . Larvae expressing Kaede in one or a few GCs were selected at 5 days post fertilization ( dpf ) , and the structure of the GCs in the larval cerebellum was examined ( Fig 1A-1D ) . We found that GCs in the Va and CCe ( rostromedial lobes ) extended their axon dorsally . It bifurcated in the superficial domain ( molecular layer ) , and then extended bilaterally to the dendrites of PCs ( parallel fibers , Fig 1A , 1B and 1Ea ) . This T-shaped axonal structure was similar to that of GCs in the mammalian cerebellum [4] . The GCs in the EG displayed a similar T-shaped axonal structure and sent their axonal branches ipsilaterally and contralaterally; however , both branches extended caudally to the dorsal hindbrain after reaching the lateral edge of the cerebellum ( Fig 1Eb ) . The GCs in the LCa extended their axon only ipsilaterally; it then extended caudally to the dorsal hindbrain , like the GC axons from the EG ( Fig 1Ec ) . Thus , the GCs from the EG and LCa ( caudolateral lobe ) extended a long axon that first synapsed with the dendrites of PCs in the cerebellum , and then with the dendrites of crest cells in the crista cerebellaris , which is part of the dorsal hindbrain ( Fig 1E ) . These data are consistent with previous findings obtained from the retrograde labeling of the dendrites of crest cells , which indicated that the GCs in the LCa and EG project ipsilaterally and bilaterally to the crest cells , respectively [11] . Our results revealed that there are three types of GCs ( Va/CCe , LCa , and EG ) ; that the axonal structure of the GCs in the rostromedial lobes ( Va/CCe ) is different from those of the GCs in the LCa and EG; and that the axons of the caudolateral lobes ( LCa and EG ) extend caudally to the crest cells . Thus , the rostromedial and caudolateral GCs form two distinct neural circuits . To elucidate the molecular mechanisms controlling GC axon development , we investigated the zebrafish mutant sio , which shows abnormal GC axon projections to the crest cells [9] . In the mutant larvae , the Vglut1+ axons of the GCs in the caudolateral lobes were aberrantly branched or misoriented in the crista cerebellaris ( n = 10/10 ) ( Fig 2A-2D ) . Labeling of the GCs with GC-specific Gal4 lines confirmed that axogenesis was affected for the GCs in the caudolateral lobes ( hspDMC90A , n = 4/4 , Fig 2E and 2F; gSA2AzGFF152B , n = 2/2 , S1 Fig ) . Sparse cell labeling with the GC-specific Gal4 lines further revealed that most of the GC axons from the LCa and EG displayed misoriented axon ( s ) ( LCa: 100% , n = 3; EG: 73 . 3% , n = 15 ) , whereas only a portion of the GC axons in the CCe showed aberrant axons in the mutant larvae ( 14 . 7% , n = 34 , S2 Fig , the statistic analysis is shown in S1 Table ) . Expression of the differentiated GC marker Neurod was not affected in the sio mutants ( S3 Fig ) , suggesting that the sio locus is involved in GC axogenesis but not in their differentiation ( Fig 2G ) . To reveal the molecular nature of the sio locus , we performed positional cloning , and mapped the locus to a region in chromosome 7 that contains the type IV collagen genes col4a5 and col4a6 ( Fig 2H ) . Sequence analysis identified a T-to-C point mutation in the splicing donor on the 3’ side of the 42nd exon of the col4a6 gene in the sio mutant genome ( Fig 2I ) . Reverse Transcription ( RT ) -PCR revealed that three aberrant col4a6 transcripts were generated by abnormal splicing in the sio mutants ( Fig 2J and 2K ) . Two of the aberrant transcripts encoded a C-terminally truncated Col4a6 , and one encoded a mutant Col4a6 containing an internal deletion ( Fig 2L ) . All of the mutant proteins had a deletion in the Noncollagenous Domain ( NC1 ) , which is required for the assembly of type IV collagen [33] . Col4a6 forms a heterotrimeric complex with Col4a5 [22] , and col4a5 is expressed in the epidermal cells surrounding the tectum in zebrafish [24] . We therefore examined the expression of col4a5 and col4a6 in the sio mutant hindbrain region . In wild type , both col4a5 and col4a6 transcripts were detected in the epidermal cells surrounding the hindbrain ( Fig 2Ma , 2Mb , 2Oa and 2Ob ) . Although the expression of col4a5 was not affected in the sio mutants ( Fig 2Pa and 2Pb ) , the expression of col4a6 was prominently reduced ( Fig 2Na and 2Nb ) , suggesting that the col4a6 transcripts in the mutants underwent nonsense-mediated mRNA decay . The injection of col4a6 RNA suppressed the formation of aberrant axons of the GCs in the sio mutants ( Fig 2Q-2W ) . In this experiment , exogenous col4a6 was provided ubiquitously , but Col4a6 may function only in epidermal cells in which its heteromeric partner , Col4a5 , is present . These data collectively indicated that the sio locus encodes Col4a6 ( hereafter we refer to the sio mutant as the col4a6 mutant ) . Zebrafish col4a5 mutants are reported to exhibit abnormal axon targeting of RGCs to the tectum [34] . We therefore examined the GC and RGC phenotypes in the col4a5 and col4a6 ( sio ) mutants ( Fig 3 ) . Mosaic labeling of RGCs with the pou4f3:Gal4 , UAS:GAP-GFP reporter line showed that some of the RGC axons trespassed between tectal layers in the col4a6 mutant larvae ( n = 4/4 ) ( Fig 3H and 3I ) , as in the col4a5 mutants [24] ( n = 3/3 ) ( Fig 3M and 3N ) . Similarly , we detected aberrant Vglut1+ GC axons in the caudolateral lobes of the cerebellum in both the col4a5 ( n = 3/3 ) ( Fig 3K and 3L ) and col4a6 mutants ( n = 5/5 ) ( Fig 3F and 3G ) . Expressivity and penetrance of the aberrant RGC and GC axons were not significantly different between col4a6 and col4a5 mutant larvae ( S2 Table ) . These data indicate that both Col4a5 and Col4a6 are required for the proper axogenesis of RGCs and caudolateral GCs , and imply that the Col4a5/Col4a6 complex plays a role in the axogenesis of these neurons in zebrafish . We next examined how type IV collagen controls RGC and GC axogenesis . Type IV collagen proteins are generated in epidermal cells and deposited into the BM underlying the epidermal cells . Type IV collagen is proposed to regulate the gradient of the axon guidance molecule Slit1 [24] . Neither the slit nor the robo genes , which encode Slit receptors , are strongly expressed in the antero-dorsal hindbrain , with the exception of robo3 , which is expressed in the cerebellum and thought to function as a negative regulator of Slit signaling [35] ( S4 Fig ) . We therefore examined whether overexpressing Slit protein would affect the formation of GC axons , by using the transgenic line hsp70l:Slit2-GFP [36] ( Fig 4A ) , which was previously used to analyze the Slit-mediated axon targeting of RGCs [24] . As reported previously [24] , the overexpression of Slit2-GFP induced aberrant RGC axons ( n = 2/2 , Fig 4E-4G ) . However , overexpressing Slit2-GFP at 3 , 4 , or 5 dpf , when the GC axons extend to the crest cells , did not significantly affect the GC axogenesis ( for 3 dpf , n = 1/10 , Fig 4J and 4K; for 4 dpf , n = 0/6 , S5C and S5D Fig; for 5dpf , n = 0/5 , S5E and S5F Fig; the statistic analysis is shown in S3 Table ) . These data suggest that Slit signaling is not involved in the axogenesis of caudolateral GCs , and thus that type IV collagen regulates the axogenesis of caudolateral GCs by a mechanism other than controlling Slit proteins . A triple helix complex of Col4a5 and Col4a6 exists in the BM of human skin , smooth muscle , and kidney , and is required for proper BM formation [22] . We therefore examined the structure of the BM surrounding the dorsal hindbrain and optic tectum of the col4a5 and col4a6 mutants ( Fig 5 ) . Immunostaining with an anti-laminin–1 antibody revealed a linear structure of the BM in the tectum and hindbrain of 5-dpf wild-type larvae ( Fig 5A , 5D and 5E ) . Although the BM initially adheres directly to the skin cells ( Fig 2Mb and 2Ob ) , the BM was separated from the skin by the otic vesicles that grew rapidly at the early larval stage ( Fig 5A , 5N and 5O ) . In the dorsal hindbrain of col4a5 and col4a6 mutants , the laminin–1 signals were split or intermittently disrupted ( Fig 5B and 5C , S4 Table ) . In the tectal BM of these mutants , the laminin-1-positive BM was split into two layers: one attached to the skin and the other more internal , with intermittent disruption ( arrowheads , Fig 5G and 5I; S4 Table ) . Some RGC axons were observed in the interspace of the split BM structures ( arrows , Fig 5G and 5I ) . We also examined the distributions of two tectal BM components , zn12 ( HNK–1 ) glyco-epitope and heparan sulfate proteoglycans ( HSPGs ) , in the col4a6 mutants , because their distributions are altered in col4a5 mutants [24] . We found that compared to wild type ( Fig 5J and 5L ) , the HNK–1 epitope and HSPGs at and near the hindbrain BM were more sparse in the col4a6 mutants ( Fig 5K and 5M , S4 Table ) . The HNK–1+ ( zn12+ ) region was also thicker in the tectal BM region of the col4a6 mutants ( S6 Fig , S4 Table ) . Electron microscopic ( EM ) analysis in wild-type larvae showed intact BM surrounding the tectum and dorsal hindbrain , where the GC axons run ( Fig 6A and 6B ) . In contrast , the electro-dense BM structure in the col4a6 mutant hindbrain contained branched and/or truncated regions ( Fig 6C-6F ) . Some axons that could be distinguished by the presence of presynaptic vesicles were observed outside the BM ( Fig 6F ) . Truncation of the tectal BM and ectopic localization of the RGC axons were also detected in both col4a5 and col4a6 mutants by the EM analysis ( S7 Fig ) . These data collectively indicated that type IV collagen is required for the integrity of both the tectal and dorsal hindbrain BM . To address the role of the BM in GC axogenesis , we carefully examined the relationship between the BM and caudolateral GC axons in the wild-type and col4a6 mutant hindbrain . The GC axons were labeled with the GC-specific Gal4 line hspGFFDMC90A and UAS:GFP [32] , and the BM was stained with the anti-laminin–1 antibody in the wild-type and col4a6 mutant dorsal hindbrain ( Fig 7 ) . The GFP+ axons of the caudolateral GCs ran along the linear BM in the wild-type hindbrain ( Fig 7A-7D and 7I ) . In contrast , in the col4a6 mutant , the BM split into two layers in the dorsal hindbrain ( Fig 7F and S8 Fig ) , and the GC axon bundle bifurcated at the split ( Fig 7F-7H and S8 Fig ) . After the bifurcation , some of the GC axons ran along the abnormal BM layer ( indicated by closed triangles , Fig 7 and S8 Fig ) but some of them did not run along the BM ( open triangles , Fig 7 and S8 Fig ) . Corresponding splits of the BM and the GC axon bundles were observed in all of the homozygous col4a6 mutant larvae observed ( 5 mutant larvae had 10 branches of the GC axon bundles; all of them were coupled with the BM splits ) , suggesting that the BM splits caused the GC axons to branch . We next ablated the axon bundles of the caudolateral GCs extending to the crest cells in 5-dpf wild-type larvae , by using a two-photon laser , and observed their regrowth . The re-growing axon bundles potentially contained both regenerated and de novo generated GC axons ( Fig 8C and S9 Fig ) . After two days , these axons had reconstructed the neural circuits to the crest cells ( Fig 8D and S9 Fig ) . We then performed this experiment using the col4a6 mutant . If the BM abnormality led to the aberrant GC axogenesis observed in this mutant , the newly formed ( and possibly regenerated ) axons should follow abnormal routes similar to those used by the original GC axons in the BM-affected areas . Following axon ablation in the col4a6 mutants , the caudolateral GC axons were re-extended as in wild type , although the extended axons were abnormal ( Fig 8E-8H and S9 Fig ) . Notably , at least some of the newly formed axons followed abnormal routes that were similar or identical to those used by the original axons ( marked by arrowheads in Fig 8J , 8K and S9 Fig ) . These findings suggest that the abnormal BM structure is directly linked to the abnormal GC axogenesis in the col4a6 mutant .
The GCs in zebrafish are known to form two different neural circuits . The GCs in the rostromedial lobes have a T-shaped axon that projects to the dendrites of PCs , whereas the GCs in the caudolateral lobes send their axons to PCs in the cerebellum and to crest cells in the dorsal hindbrain [9–11] . By mosaic analysis using GC-specific Gal4 lines , we confirmed these circuits and further revealed that , within the caudolateral lobes , the GCs in the EG and LCa exhibit different axonal structures: the GCs in the EG have a T-shaped axon that projects contralaterally and ipsilaterally to the crest cells , whereas the GCs in the LCa send their axon only ipsilaterally to the crest cells ( Fig 1 ) . The developmental processes of the GCs in the three lobes are also different [10 , 11] , and some development-related genes are differentially expressed between the GCs of the rostromedial and caudolateral lobes [9] , suggesting that distinct molecular mechanisms control the axogenesis of the three types of GCs . In this report , we found that the axons of the GCs in the caudolateral lobes were affected more strongly than those in the rostromedial lobes in col4a5 and col4a6 mutants ( Figs 2 , 3 and S2 , S1 Table ) , suggesting that Col4a5 and Col4a6 differentially control GC axogenesis in a cell-population-specific manner . The parvalbumin7-expressing crest cells in the medial octavolateral nucleus , which are axonal targets of the caudolateral GCs , were not significantly affected ( S10 Fig ) , indicating that type IV collagen controls the axogenesis of the GCs but not differentiation of their targets . Signaling from type IV collagen in the pial layer to the collagen receptor DDR1 on GCs is reported to be involved in GC axon extension in mouse [26] , implying that type IV collagen has a conserved role in GC axogenesis . The GC axons in mouse cerebellum display a T-shaped structure resembling that of the GCs in the rostromedial lobes of zebrafish cerebellum [4–7] . If the same signaling was used in mouse and zebrafish , the axons of the rostromedial GCs would be affected in type IV collagen zebrafish mutants . However only a small portion of the rostromedial GC axons were affected in col4a6 mutants ( Figs 2 , 3 and S2 , S1 Table ) . Furthermore , no ddr1 expression was detected in zebrafish GCs ( S11 Fig ) , indicating that collagen-DDR signaling may not be involved in zebrafish GC axogenesis . Nevertheless , Col4a5 and Col4a6 appear to be involved in the axogenesis of the caudolateral GCs in zebrafish . In the human genome , the COL4A5 and COL4A6 genes are located side by side on the X chromosome [22] . Two COL4A5 molecules and one COL4A6 form a heterotrimeric complex through their NC1 domain , and the complex is deposited into the BM [22] . Similarly , the zebrafish col4a5 and col4a6 genes are located side by side on chromosome 7 ( Fig 2 ) , indicating a conserved synteny of these genes in vertebrates . Both col4a5 and col4a6 are expressed in epidermal cells ( Fig 2 ) , and a mutation in either gene resulted in similar abnormalities of the RGC and caudolateral GC axogenesis , as well as in the structure of the BM surrounding the tectum and hindbrain ( Figs 3 and 5 ) . Thus , a loss of Col4a6 cannot be compensated for by Col4a5 , and vice versa , and both proteins have essential roles in RGC/GC axogenesis and BM integrity . Our data are consistent with the idea that Col4a5 and Col4a6 form a heterotrimeric complex in zebrafish , as they do in mammals . In humans , defects in the COL4A3/COL4A4/COL4A5 complex are the major cause of Alport syndrome , which is associated with a loss of BM integrity . Mutations in COL4A6 alone have not been related to Alport syndrome [22 , 23] , but a COL4A6 mutation is linked to hereditary hearing loss , which is a more restricted symptom compared to Alport syndrome [37] , implying that the Col4a5/Col4a6 complex has a more limited role in BM integrity in humans than in zebrafish . There are three possible mechanisms by which type IV collagen controls axon outgrowth and pathfinding [14 , 17–21] . First , collagens in the BM bind to collagen receptors on neurons and induce neurite outgrowth . The major receptors for type IV collagen are integrin-family transmembrane proteins , such as α1β1 and α2β1 [27–29] , and the binding of collagen to integrins initiates cytoplasmic signaling that activates focal adhesion kinase ( FAK ) [14] . Consistent with this mechanism , activated ( phosphorylated ) FAK was detected in the axons of the caudolateral GCs in wild-type larvae ( S12 Fig ) . However , it was not significantly altered in the col4a6 mutants ( S12 Fig ) . Furthermore , blocking FAK activation with a chemical inhibitor did not affect GC axogenesis ( S13 Fig ) . These data indicate that the Col4a5/Col4a6 complex may not initiate integrin signaling to activate FAK . Similarly , the morpholino-mediated knockdown of β1 integrin does not affect the axon targeting of RGC axons to the tectum [24] . Our data do not exclude the possibility that type IV collagen controls other aspects of neural development through unconventional collagen receptors , such as GPR56 and GPR126 , which are reported to function in cell migration and/or tissue morphogenesis [30 , 31] or the possibility that the BM components other than the Col4a5/Col4a6 complex play roles in the GC axogenesis . The second mechanism is that type IV collagen binds to guidance molecules and controls their concentration in and near the BM . Slit proteins are possible candidates , since type IV collagen binds to Slit1 , and Slit overexpression or a mutation in the Slit receptor gene , robo2 , affects the axon targeting of RGCs to the tectal layers [25] . However , overexpressing Slit did not affect the axogenesis of caudolateral GCs ( Fig 4 ) , suggesting that the Slit-Robo system is not involved in this axogenesis and that type IV collagen does not control GC axons by regulating the Slit gradient . Therefore , type IV collagen might have a different role in the axogenesis of RGCs versus GCs . However , we cannot exclude the possibility that type IV collagen controls the concentration gradient of guidance molecules other than Slits , and we note that still other mechanisms may be involved in the type IV collagen-mediated RGC axogenesis . The third mechanism is supported by our findings , i . e . , that type IV collagen controls the axogenesis of both RGCs and GCs by regulating the integrity of the BM , as discussed below , although we cannot rule out a role for guidance molecules or other mechanisms in this process . Type IV collagen proteins in the BM are thought to confer tensile strength to the BM . We observed abnormalities in both the tectal and dorsal hindbrain BM of the col4a5 and col4a6 mutants , as revealed by laminin–1 staining , including branching , truncation , and thinning ( Fig 5 ) . The abnormal BM structure was further confirmed by EM analysis ( Fig 6 and S7 Fig ) . Our data indicate that the Col4a5/Col4a6 complex is required for the integrity of the BM . In addition to the visible BM abnormalities , the distributions of the ECM components HNK–1 glyco-epitope and HSPGs in the vicinity of the BM were affected in the col4a6 mutant hindbrain and tectum ( Fig 5 and S6 Fig ) . Abnormal HNK–1 and HSPG distributions were reported for the col4a5 mutants [24] , supporting a role for the Col4a5/Col4a6 complex in the deposition of ECM components into the BM . In a previous report , col4a5 mutants did not show clear abnormalities in the laminin–1+ BM structure in the tectum [24] . The reason for the difference between our observations and the previous data on BM structures is unclear . One possibility is that , as the BM abnormalities did not always occur in the same positions and the BM lesions were discontinuous in the type IV collagen mutants , they might have been missed by other researchers . Even in the absence of the Col4a5/Col4a6 complex , the BM structure was not entirely abrogated , because other collagens and ECM components were present . Consistent with our observations of the BM , the abnormal branching and misorientation of the caudolateral GC axons were observed in different positions in each col4a6 mutant larva ( and also differed on the left and right sides , see Figs 2 , 7 and 8 ) . Similarly , the col4a5 and col4a6 mutants showed variations in the abnormal RGC axon targeting ( Fig 3 ) [24] . It is possible that the Col4a5/Col4a6-defective BM is vulnerable to tensile force , and that its structure is disrupted by a physical force that accompanies larval growth . In any case , our data suggest that the BM abnormalities take place at relatively random positions , and that the abnormal BM structure led to abnormal axogenesis of the caudolateral GCs and possibly RGCs in the type IV collagen mutants . We found that the defective BM caused two types of aberrant axogenesis: ( 1 ) the presence of axons outside the nervous system , and ( 2 ) the branching and misorientation of axon bundles . In regions of BM disruption , some RGC axons wandered out of the tectum in col4a5 and col4a6 mutants , and some GC axons were observed outside the hindbrain in col4a6 mutants ( Figs 5 and 6 ) . This situation is similar to that of the nephrons in Alport syndrome patients , in which blood cells escape from blood capillaries to the Bowman’s capsule through the disrupted BM [38] . Branched and misoriented GC axon bundles were observed in both type IV collagen mutants ( Figs 2 and 7 ) . Importantly , we found a close correlation between the regions of BM disruption and GC branching in the col4a6 mutants . The axons of the caudolateral GCs ran along the BM in the hindbrain . At the branch points of the BM in the type IV collagen mutants , the axon bundles also branched , and many of the axon bundles ran along the abnormal BM ( Fig 7 ) , suggesting that the aberrant axons were guided by the abnormal BM structure . Furthermore , laser ablation of the GC axons revealed that the re-growing axons followed similar ( or the same ) abnormal routes as were used by the original GC axons in the col4a6 mutants ( Fig 8 ) , suggesting that the normal axon path was broken in the mutants , preventing the axons from following the correct path to their target . The BM may serve to guide these axons to their target . Some of the abnormal axon bundles did not run along the BM after the branching in the col4a6 mutant hindbrain ( Fig 7J and S8 Fig ) , suggesting that the BM breaks caused misorientation of the caudolateral GC axons but did not attract them in some cases . The data support the idea that the BM integrity is essential for the axogenesis of the caudolateral GCs . A similar role of BM integrity is also reported for the mouse cerebellum , in which a deficiency of laminin α1 , a subunit of laminin–1 , results in corresponding defects in BM integrity and cerebellum development [39] , suggesting that BM has a conserved role in cerebellum development . The question remains , if the abnormal axogenesis was attributable to abnormal BM structures . Why were only the axons of the RGCs and GCs affected in the BM-defective mutants ? An explanation may lie in the neuroanatomy of the developing cerebellar neural circuits . During the embryonic and larval stages , the major neural tracts run ventrally and are not located in the vicinity of the BM . The axons of the RGCs and GCs , however , form major axon tracts that are located most dorsally and are at least partially attached to BM . Therefore , the BM structure could play a pivotal and specific role in the axogenesis of these neurons . To reach their correct targets , axons require proper guidance signaling . In the col4a6 mutants , some GC axons still ran along the abnormal BM ( Figs 7 and 8 ) , suggesting that the BM itself , or a BM-associated molecule , functions to attract the axons . It is also likely that guidance molecules act on the GC axons independently of the BM . In any case , the BM and guidance molecules would need to cooperate to control the pathfinding of the GC axons . Eph-ephrin signaling plays a major role in guiding RGC axons [40 , 41] . More study is needed to clarify the signaling system that cooperates with the BM to guide the caudolateral GC axons . In summary , our data revealed that type IV collagen controls the axogenesis of caudolateral GCs and RGCs by establishing and/or maintaining the integrity of the BM . The role of type IV collagen in BM integrity and axogenesis may provide insight into the etiology and pathology of Alport syndrome .
The animal work in this study was approved by Nagoya University Animal Experiment Committee ( approval number: 2014020503 , 2015022304 ) and was conducted in accordance with “Regulations on Animal Experiments in Nagoya University ( Regulation No . 71 , March 12 , 2007 ) ” and the “Guidelines for Proper Conduct of Animal Experiments ( June 1 , 2006 , Science Council of Japan ) . Wild-type zebrafish ( Danio rerio ) with the Oregon AB genetic or TL background were used . In some experiments , mutant and Tg larvae were generated on the casper ( mitfaw2; roya9 ) background [42] . The mutants used in this report were siork18 [9] and col4a5s510 ( dragnet ) [24] . To genotype siork18 , genomic DNA at the mutation site was amplified by PCR using the primers 5’-ATGTGTCCTGAGGGAATGACCAGG–3’ and 5’- TTCATTGGCGGTGCAGTAGAGGA–3’ , followed by digestion with HphI . To genotype col4a5s510 , genomic DNA at the mutation site was amplified by PCR using the primers 5’-GCCTGGTTCACCTGAGAAT–3’ and 5’-GATTGCCAGGTCATTTCCTT–3’ , followed by digestion with NheI . The transgenic lines pou4f3:GAL4 ( s311t ) , and pou4f3:GAL4 , UAS:GAP-GFP ( s318t ) [34] , and hsp70l:Slit2-GFP ( rw015a ) were previously reported [36] . The Gal4 trap lines gSA2AzGFF152B , gSAIGFF23C , and hspGFFDMC90A , which express a modified Gal4 ( GFF ) in GCs , were described previously [32] . gSA2AzGFF152B and hspGFFDMC90A express Gal4 in GCs of the Va , CCe , LCa , and EG , whereas gSAIGFF23C expresses Gal4 in GCs of the LCa and EG . The UAS:Kaede and UAS:GFP fish ( rk8Tg and nkuasgfp1aTg in ZFIN: http://zfin . org/ ) were previously reported [43 , 44] . The zebrafish were maintained in an environmentally controlled room at the Bioscience and Biotechnology Center , Nagoya University . Tol1-mediated single-cell labeling was carried out as described previously [32] . Briefly , 5–25 pg of UAS:Kaede plasmid DNA and 50 pg of Tol1 transposase RNA were co-injected into 4-to-8-cell stage granule-cell-specific Gal4 Tg embryos . Larvae expressing Kaede in one or a few GCs were selected and observed under an epi-fluorescence microscope ( MZ16A , Leica ) . For immunostaining , anti-GFP ( 1:1000 dilution , rat , Nacalai ) , anti-parvalbumin 7 ( Pvalb7 , 1:1000 , mouse monoclonal , ascites ) , anti-Vglut1 ( 1:1000 , rabbit , affinity purified ) , anti-Neurod ( 1:500 , mouse monoclonal , ascites ) [9] , anti-laminin ( 1:150 , rabbit , Sigma ) , anti-phosphorylated FAK ( anti-FAK[pY397] , 1:200 , rabbit , Life Technologies ) antibodies were used . Immunostaining was performed as described previously [9 , 10] . Alexa Fluor 488 goat anti-rabbit and Alexa Fluor 555 goat anti-mouse IgG ( H+L , Molecular Probes , Life Technologies ) were used as secondary antibodies . To inhibit FAK , zebrafish were treated with FAK inhibitor PF–573228 ( Sigma-Aldrich ) in 1% dimethyl sulfoxide ( DMSO ) [45 , 46] . For single-cell labeling and immunohistochemistry , embryos and larvae were treated with 0 . 005% phenylthiourea from 12 hpf to prevent pigmentation . Optical clearing of some fixed samples was carried out with SeeDB reagent as previously reported [47 , 48] . Fluorescent images were captured with an LSM700 confocal laser-scanning microscope equipped with a 20x/0 . 8 numerical aperture ( NA ) or 40x/1 . 3 NA oil-immersion objective ( Zeiss ) . To construct images , a series of optical sections was collected in the Z dimension ( Z-stack ) and projected as a single image or reconstructed in three dimensions to provide views of the image stack at different angles using the 3D projection program associated with the microscope ( Zen , Zeiss ) or by Imaris ( Bitplane ) . The figures were constructed using Adobe Photoshop and Adobe Illustrator . The sio mutant was initially isolated from an AB strain [9] . siork18 heterozygous fish were mated with TL fish to generate F1 families . Homozygous siork18 mutant larvae were raised from the F1 crosses and selected by immunohistochemistry with the anti-Vglut1 antibody . We used samples of their genomic DNA for segregation analysis . Primers for the simple sequence length polymorphism ( SSLP ) markers used for positional cloning ( Fig 2 ) were: 5’-TCATGTTGCTACAAGGCAAAA–3’ and 5’-TTGGGAAATGATTTGCAGTTT -3’ for 41 . 69 Mb , 5’-CAGAGGTCCTGATGGATTTGA–3’ and 5’-CAGACCCTGTGGAGGAAGAT–3’ for 42 . 5 Mb , 5’-CGCTCGTGGTGAGACAAATA–3’ and 5’-GGCGTCTGCATTGATGTTTA–3’ for 43 . 12 Mb , and 5’-AAGTCACATCTGGGTACGGC–3’ and 5’-TGCATCACTGAAAATGTGCA–3’ for 45 . 54 Mb ( Z8584 ) . Electron microscopic analysis was carried out as described previously [49 , 50] using JEM–1010 and JEM-1400Plus ( JEOL ) . Laser ablation of the GFP-labeled GC axons was carried out using an LSM780-DUO-NLO laser-scanning inverted microscope ( Zeiss ) equipped with a Ti-sapphire femtosecond pulse laser ( Chameleon Vision II , Coherent ) as described previously [37] . A Ti-sapphire laser tuned to 880 nm was used for the ablation . Before and after laser irradiation , time-lapse images were captured with an LSM700 confocal laser-scanning microscope . Statistic analyses are performed with Fisher’s exact test ( S1 , S2 , S3 and S4 Tables ) and Student’s t-test ( S10 Fig ) . | The cerebellum is involved in motor coordination and motor learning . Granule cells are the major excitatory neurons in the cerebellum . It is largely unknown how the formation of cerebellar neural circuits , including the elaboration of granule cell axons , is controlled . We investigated a zebrafish mutant shiomaneki , in which some of the granule cells have abnormal axons . We identified collagen ( col ) 4a6 as the gene responsible for the mutant phenotype . Col4a6 forms a complex with Col4a5 , which is a component of the basement membrane . We found that mutants of both col4a5 and col4a6 showed similar axonal abnormalities in both the granule cells and the retinal ganglion cells , and that the basement membrane structure surrounding the central nervous system was disrupted in these mutants . Furthermore , the abnormalities in granule cell axon formation were coupled with aberrant basement membrane structures in the col4a6 mutants . These data suggest that type IV collagen controls the axon formation of some types of neurons by establishing and/or maintaining the integrity of the basement membrane , which provides axons with the correct path to their targets . These findings may explain some aspects of a human disorder , Alport syndrome , which is caused by mutations in type IV collagen genes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Type IV Collagen Controls the Axogenesis of Cerebellar Granule Cells by Regulating Basement Membrane Integrity in Zebrafish |
Most bacterial glycoproteins identified to date are virulence factors of pathogenic bacteria , i . e . adhesins and invasins . However , the impact of protein glycosylation on the major human pathogen Staphylococcus aureus remains incompletely understood . To study protein glycosylation in staphylococci , we analyzed lysostaphin lysates of methicillin-resistant Staphylococcus aureus ( MRSA ) strains by SDS-PAGE and subsequent periodic acid-Schiff’s staining . We detected four ( >300 , ∼250 , ∼165 , and ∼120 kDa ) and two ( >300 and ∼175 kDa ) glycosylated surface proteins with strain COL and strain 1061 , respectively . The ∼250 , ∼165 , and ∼175 kDa proteins were identified as plasmin-sensitive protein ( Pls ) by mass spectrometry . Previously , Pls has been demonstrated to be a virulence factor in a mouse septic arthritis model . The pls gene is encoded by the staphylococcal cassette chromosome ( SCC ) mec type I in MRSA that also encodes the methicillin resistance-conferring mecA and further genes . In a search for glycosyltransferases , we identified two open reading frames encoded downstream of pls on the SCCmec element , which we termed gtfC and gtfD . Expression and deletion analysis revealed that both gtfC and gtfD mediate glycosylation of Pls . Additionally , the recently reported glycosyltransferases SdgA and SdgB are involved in Pls glycosylation . Glycosylation occurs at serine residues in the Pls SD-repeat region and modifying carbohydrates are N-acetylhexosaminyl residues . Functional characterization revealed that Pls can confer increased biofilm formation , which seems to involve two distinct mechanisms . The first mechanism depends on glycosylation of the SD-repeat region by GtfC/GtfD and probably also involves eDNA , while the second seems to be independent of glycosylation as well as eDNA and may involve the centrally located G5 domains . Other previously known Pls properties are not related to the sugar modifications . In conclusion , Pls is a glycoprotein and Pls glycosyl residues can stimulate biofilm formation . Thus , sugar modifications may represent promising new targets for novel therapeutic or prophylactic measures against life-threatening S . aureus infections .
Although usually being a common inhabitant of the human skin and mucous membranes , Staphylococcus aureus is a human pathogen that can cause diseases ranging from mild skin infections to serious and life-threatening infections , such as endocarditis , osteomyelitis , pneumonia , meningitis , and sepsis [1 , 2] . Especially due to the increasing use of various medical devices and implants in modern medicine , the number of nosocomial S . aureus infections is constantly rising [3 , 4] . Furthermore in the past three decades , the emergence of antibiotic-resistant staphylococci , such as methicillin-resistant S . aureus ( MRSA ) represents an increasing problem in the treatment of S . aureus infections . Thus , alternative therapeutic or prophylactic measures against S . aureus infections are urgently required . Until recently , it has been considered a dogma that bacteria are unable to glycosylate proteins , because they lack the equivalent cellular structures involved in protein glycosylation in eukaryotes . Now , it is widely accepted that bacteria can glycosylate proteins . Most bacterial glycoproteins identified to date are virulence factors of pathogenic bacteria , i . e . adhesins and invasins [5–7] . Bacteria have two basic systems to glycosylate proteins: N-linked and O-linked glycosylation [8–10] . The sugar transfer is carried out by glycosyltransferases ( Gtfs ) [10] . The N-linked glycosylation pathways have been well characterized in Gram-negative bacteria [5–7 , 9] . Known O-linked glycoproteins include serine-rich repeat ( SRR ) surface proteins from Gram-positive cocci , such as the 286-kDa platelet-binding protein GspB from Streptococcus gordonii and the homologous 227-kDa serine-rich adhesin for platelets ( SraP ) from S . aureus [11–14] . Very recently , the serine-rich S . aureus clumping factor A ( ClfA ) has also been identified as a glycoprotein [15 , 16] . Generally , adherence of S . aureus to components of the extracellular matrix or host tissue , i . e . endothelial and epithelial cells or platelets , is a prerequisite for tissue colonization and the initiation of an infection , such as infective endocarditis . S . aureus harbors an armamentarium of surface ( covalently linked to the peptidoglycan ) and surface-associated ( non-covalently attached to the surface ) adhesins that mediate adherence to extracellular matrix or plasma proteins acting as bridging molecules or directly to host cell receptors [17] . SraP and ClfA belong to a family of staphylococcal surface proteins characterized by common features , such as an N-terminal signal peptide , a ligand-binding A region , a repeat region , and a C-terminal cell wall anchor [18] . The C-terminal anchor domain consists of an LPXTG-motif that is involved in covalent linkage of the protein to peptidoglycan , followed by a stretch of hydrophobic amino acids ( aa ) , and a short charged tail [18] . SraP and GspB have very similar features including their large size , an atypically long putative N-terminal signal peptide , two SRR domains , srr1 and srr2 , that are separated by a non-repeat region , and the LPXTG cell wall anchor [13 , 14] . Furthermore , both genes , gspB and sraP , are located in operons that additionally encode accessory secretion ( Sec ) proteins and Gtfs [19 , 20] . Within the accessory sec system , gtfA and gtfB are located downstream of the sraP structural gene and have been reported to be required for the glycosylation of SraP [20 , 21] . The SraP protein domain containing srr1 and the non-repeat region was found to directly bind to platelets and the expression of sraP correlates with an increased virulence in a rabbit model of experimental infective endocarditis [14] . SRR glycoproteins have also been associated with increased virulence in animal models of meningitis [22 , 23] and blood stream infection [16 , 24] . In contrast to SraP , ClfA is not part of an operon that also contains the genes encoding the Gtfs . Instead , ClfA becomes glycosylated by the novel Gtfs SdgA and SdgB , whose genes are located downstream of the tandemly arranged genes encoding the SD-repeat ( Sdr ) proteins SdrC , SdrD , and SdrE [15 , 16] . The potential role of posttranslational protein glycosylation in adherence or in the pathogenesis of staphylococcal infections in general is largely unknown . Therefore , the aim of this study was to identify S . aureus surface proteins that are posttranslationally modified by carbohydrate moieties , the underlying glycosylation machinery and their potential role in the pathogenesis of staphylococcal infections . We found that the plasmin-sensitive surface protein Pls previously characterized as a virulence determinant in mouse septic arthritis and associated with the staphylococcal cassette chromosome ( SCC ) mec type I [25 , 26] is a glycoprotein and identified two open reading frames downstream of the pls structural gene that encode novel Gtfs ( termed GtfC/GtfD ) involved in Pls glycosylation . Functional characterization indicated that Pls carbohydrate moieties can stimulate biofilm formation , while they are not apparently involved in other Pls properties .
To identify glycosylated proteins in S . aureus , surface proteins from the MRSA strains COL and 1061 were analyzed ( strains are listed in Table 1 ) . Covalently linked surface proteins were prepared from cultures grown to exponential or stationary growth phase by lysostaphin treatment . Subsequently , the proteins were separated by SDS-PAGE and glycosylated proteins were detected by periodic acid-Schiff’s ( PAS ) staining . In the strain S . aureus COL , four glycosylated surface proteins with molecular masses of approximately >300 , 250 , 165 , and 120 kDa were detected in lysostaphin lysates from overnight-grown cultures ( Fig 1AI ) . Protein bands with the same molecular masses were present in lysostaphin lysates from S . aureus COL cultures grown to the exponential growth phase although to a lesser extent ( 1A II ) . In the strain S . aureus 1061 , only two glycosylated surface proteins with molecular masses of >300 and ∼175 kDa were identified in lysostaphin lysates from overnight-grown cultures ( Fig 1A ) . For comparison , no glycosylated surface protein could be detected in lysostaphin lysates from the apathogenic strain Staphylococcus carnosus TM300 ( Fig 1A ) . No additional glycosylated proteins could also be detected from the preparations of surface-associated proteins of the strain COL ( Fig 1AIII ) . To identify the glycosylated proteins , the ∼250-kDa and ∼165-kDa proteins from strain COL and the ∼175-KDa protein from strain 1061 were excised and subjected to mass spectrometry ( MS ) . All three proteins were identified as the plasmin-sensitive protein Pls ( Fig 1B ) . Pls is a covalently cell wall-anchored protein of MRSA strains with a reported apparent molecular mass of 230 kDa that is sensitive to proteolysis by plasmin leading to 175-kDa and 68-kDa cleavage products [26] . The predicted Pls polypeptide from strain COL has 1 , 548 aa and a calculated molecular mass of 165 kDa and thus is slightly smaller than Pls from strain 1061 , which consists of 1 , 637 aa and has a calculated molecular mass of 175 kDa . Thus , the reported apparent molecular mass of Pls is much higher than the calculated molecular mass [26] , which at least in part might be due to its glycosylation . To verify that Pls is a glycosylated protein , we analyzed the surface proteins prepared from the pls mutant strain S . aureus 1061pls and the complemented mutant S . aureus 1061pls ( pPLS4 ) by PAS staining ( Fig 1C ) . The previously reported plasmid pPLS4 [26] encodes the pls gene from strain 1061 ( see below , Fig 5A ) . The ∼175-kDa glycosylated surface protein was missing from the 1061pls mutant strain ( lane 3 ) , but present in the wild-type strain 1061 ( lane 2 ) and the complemented mutant strain ( lane 4 ) confirming that Pls is a glycosylated protein ( Fig 1C ) . To verify these observations and to exclude the possibility that glycosylated Pls produced by the strains S . carnosus ( pPlsGtfΔCD1061 ) and SA113sdgA/sdgB ( pPlsGtfΔCD1061 ) is below the detection limit , we purified the respective glycosylated proteins by using the lectin ConA . Lectins are carbohydrate-binding proteins that have high substrate specificity [37–39] . It has been reported before that Pls can be purified by using the lectin WGA [28] . Here , we successfully purified Pls from the strains 1061 ( Fig 4AI , lane 5 ) and COL ( Fig 4AII , lane 5 ) by using ConA . Moreover , we could purify Pls by using ConA , when heterologously expressed by S . carnosus TM300 ( pPlsGtfCDCOL ) ( Fig 4B , lane 3 ) , S . carnosus TM300 ( pPlsGtfCD1061 ) ( Fig 4D , lane 3 ) , SA113sdgA/sdgB ( pPlsGtfCDCOL ) ( Fig 4C , lane 3 ) or SA113sdgA/sdgB ( pPlsGtfCD1061 ) ( Fig 4E , lane 3 ) . Pls could also be purified by using ConA from S . carnosus TM300 ( pPlsGtfΔCDCOL ) , when only gtfD was coexpressed with pls ( Fig 4B , lane 5 ) . In contrast , Pls could not be purified from strain S . carnosus TM300 ( pPlsGtfCΔDCOL ) ( Fig 4B , lane 7 ) confirming our results presented in Fig 3B that suggested that GtfD is necessary for an initial glycosylation of Pls and GtfC is involved in further glycosylation , but dispensable . Very similar results were obtained , when Pls was purified from SA113sdgA/sdgB ( pPlsGtfΔCDCOL ) ( Fig 4C , lane 5 ) and from SA113sdgA/sdgB ( pPlsGtfCΔDCOL ) ( Fig 4C , lane 7 ) . As expected , it was not possible to purify Pls from the strains S . carnosus TM300 ( pPlsGtfCΔD1061 ) ( Fig 4D , lane 7 ) and SA113sdgA/sdgB ( pPlsGtfCΔD1061 ) ( Fig 4E , lane 7 ) . In agreement with our results from Fig 3D and 3E , it was not possible to purify Pls from strains S . carnosus TM300 ( pPlsGtfΔCD1061 ) ( Fig 4D , lane 5 ) and SA113sdgA/sdgB ( pPlsGtfΔCD1061 ) ( Fig 4E , lane 5 ) by using ConA thus confirming that GtfD from strain 1061 unlike that from strain COL does not seem to be sufficient for an initial glycosylation of Pls and requires the additional activity of GtfC . In order to identify the region of glycosylation in Pls , we transformed the strain 1061pls with the plasmid pPLS6 that encodes the pls gene with a deleted SD-repeat region [29] ( Fig 5A ) . Furthermore , we generated different subclones from plasmid pPLS4 in strain 1061pls that led to the production of truncated versions of Pls with 17 aa ( pPLSsub1 ) , 34 aa ( pPLSsub2 ) , or 130 aa ( pPLSsub3 ) of the SD-repeat region ( Fig 5A ) . SDS-PAGE of surface proteins revealed that the 1061pls strains harboring the plasmids pPLS4 ( lane 4 ) , pPLS6 ( lane 5 ) , pPLSsub1 ( lane 6 ) , pPLSsub2 ( lane 7 ) , or pPLSsub3 ( lane 8 ) all produced a large surface protein with the expected molecular masses in contrast to the control strain 1061pls ( lane 3 ) ( Fig 5B , upper panel ) . However , PAS staining revealed that only the strains 1061pls ( pPLS4 ) ( lane 4 ) and 1061pls ( pPLSsub3 ) ( lane 8 ) produced a glycosylated version of Pls ( Fig 5B , lower panel ) . This indicates that the SD-repeat region of Pls is modified by glycosyl residues with an apparent minimal requirement of > 34 aa of the SD-repeat region . The intensity of the glycostained protein bands produced by strain 1061pls ( pPLSsub3 ) ( lane 8 ) is markedly decreased in comparison to that produced by strain 1061pls ( pPLS4 ) ( lane 4 ) strongly suggesting a lower number of attached sugar moieties due to the shortened SD-repeat region with strain 1061pls ( pPLSsub3 ) . Pls preparations from the strains COL and 1061 were extensively digested by use of trypsin , chymotrypsin , endoproteinase Glu-C , and thermolysin and the proteolytic peptides were subsequently analyzed and sequenced by means of MS . Though leading to high sequence coverages of the non-SD repeat regions of the proteins , no hint on glycosylation was obtained . This further confirmed that glycosylation was restricted to the SD repeats that were , however , not susceptible to proteolysis . Even thermolysin ( supposed to cleave N-terminal to alanine ) and pronase ( yielding randomly cleaved short peptides down to single aa ) failed to produce ( glyco ) peptides derived from the SD repeats . Acid hydrolysis in the presence of 12 . 5% ( v/v ) acetic acid at 95°C was finally successful with respect to the preparation of the desired ( glyco ) peptides . The hydrolytic ( glyco ) peptides were analyzed by nanoESI MS and the spectrum resulting from the hydrolysate of Pls derived from strain 1061 is shown in Fig 6A . A number of species with aa compositions ( SD ) n and ( AD ) 1-2 ( SD ) m carrying 0-n and 0-m N-acetylhexosamine ( HexNAc ) moieties , respectively , as monosaccharides most probably attached to serine residues . For reasons of clarity only a few of them are labeled in the spectrum ( Fig 6A ) , but a summary of all detected species is given in S1 Table . For a closer inspection , selected glycopeptides ion species were subjected to collision-induced dissociation ( CID ) . The fragment ion spectra—an example ( [M+H]+ of ( SD ) 2-HexNAc ) is shown in Fig 6B—corroborating the assumed structures and confirming that acid hydrolysis was achieved by cleavage of peptide bonds C-terminal to aspartic acid as has been reported earlier [40] . Moreover , isobaric species , i . e . sequence isomers in alanine-containing glycopeptides could be identified . However , the positions of the HexNAc residues could not be unambiguously determined since only very few glycosylated fragment ions were detectable . Similar results were obtained for Pls derived from strain COL . The large number of observed ( glyco ) peptides and the fact that peptides carry independently of their length from no up to maximum number of HexNAc moieties indicate that glycosylation as well as hydrolytic cleavage are random processes . In order to get a clue whether SdgA/SdgB play a role in Pls glycosylation in MRSA strains , an estimation was made by comparing the ratios of the relative intensities of MS peaks corresponding to glycosylated and non-glycosylated SD repeat hydrolytic peptides obtained from Pls preparations of the S . aureus strain COL and its sdgA/sdgB mutant . The result is shown in Table 2 . Indeed , the intensity ratios I ( SD ) -HexNAc/I ( SD ) , I ( SDAD ) -HexNAc/I ( SDAD ) , I ( SDSD ) -HexNAc/I ( SDSD ) , and I ( SDSD ) -HexNAc2/I ( SDSD ) were lower by a factor of approximately 2 to 3 for hydrolysates of Pls from the COLsdgA/sdgB mutant compared to the wild-type Pls . This result gives some confirmation that SdgA/SdgB participate in the glycosylation of Pls . It has been reported before that MRSA strains that naturally express pls as well as MSSA strains that recombinantly express pls are internalized by non-professional phagocytes , such as host endothelial cells to a significantly lesser extent , which was also proposed to be due to steric hindrance [29 , 42] . To analyze , whether the Pls-mediated prevention of internalization of S . aureus strains by human host cells is dependent on its glycosylation , we analyzed the strains described above producing glycosylated or non-glycosylated Pls for their internalization by EA . hy 926 endothelial cells using flow-cytometric internalization assays . The control strain SA113 ( pCU1 ) was internalized by the endothelial cells at a similar rate like the strain Cowan 1 that is known to have a high capacity for internalization and was set to 100% internalization ( Fig 7B ) . Similarly , the control strain SA113sdgA/sdgB ( pCU1 ) was internalized by the endothelial cells at a high level , although its internalization rate seems slightly reduced compared to its parent . All strains producing Pls showed a significant and strong reduction of the internalization rate by endothelial cells in comparison to their control strains ( Fig 7B ) . However , there was no significant difference in the internalization rate among the strains producing a glycosylated version of Pls [SA113 ( pPlsGtfΔCΔD1061 ) , SA113 ( pPlsGtfΔCΔDCOL ) , SA113sdgA/sdgB ( pPlsGtfCD1061 ) , SA113sdgA/sdgB ( pPlsGtfCDCOL ) ] versus a non-glycosylated version of Pls [SA113sdgA/sdgB ( pPlsGtfΔCΔD1061 ) , SA113sdgA/sdgB ( pPlsGtfΔCΔDCOL ) ] ( Fig 7B ) . Moreover there was no significant difference in the internalization rate between the strains expressing the genes from strain 1061 or COL . Thus , the glycosylation of Pls does not seem to influence the internalization rate by human endothelial cells . To study the potential impact of Pls and its glycosylation status on the phagocytosis of S . aureus by professional phagocytes , we performed a flow-cytometric phagocytosis assay . The phagocytosis of the control strain SA113 ( pCU1 ) by PMNs was set to 100% phagocytosis ( Fig 7C and 7D ) . The control strain SA113sdgA/sdgB ( pCU1 ) was phagocytosed by PMNs at a similar level ( Fig 7C ) . All strains producing Pls showed a significant reduction of the phagocytosis rate in comparison to their control strains ( Fig 7C ) . However , there was no significant difference in the phagocytosis rate among the strains producing a glycosylated version of Pls [SA113sdgA/sdgB ( pPlsGtfCD1061 ) , SA113sdgA/sdgB ( pPlsGtfCDCOL ) ] versus a non-glycosylated version of Pls [SA113sdgA/sdgB ( pPlsGtfΔCΔD1061 ) , SA113sdgA/sdgB ( pPlsGtfΔCΔDCOL ) ] . There was also no difference between the strains expressing the genes from strain 1061 or COL . Similarly , there was no significant difference in the phagocytosis rate between the strains 1061 and the 1061pls mutant expressing the different subclones of pls ( Fig 7D ) . In contrast , the 1061pls mutant was phagocytosed at a significantly higher rate ( Fig 7D ) . Pls has been reported to mediate cell-cell interaction [43] . To address the question , whether Pls mediates biofilm formation in a glycosylation-dependent manner , we analyzed the biofilm forming capacities of strains harboring the different pls and gtfC/gtfD expression plasmids in a polystyrene microtiter plate . Strains SA113 and SA113sdgA/sdgB expressing pls did not show increased biofilm formation probably because these strains form a strong polysaccharide intercellular adhesin ( PIA ) -dependent biofilm [44] thereby masking other factors ( S2 Fig ) . However , we found that strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) producing Pls glycosylated by GtfC/GtfD formed significantly higher levels of biofilm ( P ≤ 0 . 001 ) than their negative controls carrying the empty vector ( Fig 8A ) . Wells of representative biofilms stained with safranin are shown in the supplemental S2 Fig . Moreover , they also formed significantly higher levels of biofilm ( P ≤ 0 . 001 ) than the respective strains producing Pls and non-functional GtfC/GtfD [Newman ( pPlsGtfΔCΔDCOL ) and NewmansdgA/sdgB ( pPlsGtfΔCΔDCOL ) ] ( Fig 8A ) . To ensure that increased biofilm formation is indeed due to GtfC/GtfD-glycosylated Pls , we also constructed strains expressing functional gtfC/gtfD , but not pls [Newman ( pΔPlsGtfCDCOL ) and NewmansdgA/sdgB ( pΔPlsGtfCDCOL ) ] , which produced significantly lower levels of biofilm ( P ≤ 0 . 001 ) than strains expressing the intact pls ( Fig 8A ) . Thus , the effect of increased biofilm formation in strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) clearly depends on Pls and its glycosylation by GtfC/GtfD . Strain Newman is known to harbor a variation of the SaeRS regulatory locus [45–47] . A nucleotide exchange within the saeS gene results in an exchange from leucine at aa position 18 ( present in other S . aureus strains , saeSL ) to proline ( saeSP ) [45–47] . This aa exchange leads to a constitutively expressed SaeRS system in strain Newman that has multiple consequences , one of them being reduced biofilm formation [47] . To verify that the increase in biofilm formation in strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) is due to the expression of GtfC/GtfD-glycosylated Pls and not due to a point mutation resulting in the saeSL allele , which would also lead to increased biofilm formation , we sequenced the saeRS locus in both strains and found no sequence alteration in comparison to the published S . aureus Newman genome sequence . Thus , we can rule out that a mutation within the saeRS regulatory locus caused the phenotype . To further exclude that any other mutation in the genome of strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) caused the observed phenotype of increased biofilm formation , we cured both strains from their plasmids generating strains Newmancured ( C ) and NewmansdgA/sdgBC . To enable equal growth conditions among all strains in the biofilm assay ( i . e . supplementation with antibiotics ) , we transformed the cured strains with the empty vector generating strains NewmanC ( pCU1 ) and NewmansdgA/sdgBC ( pCU1 ) . In the biofilm assays , strains NewmanC ( pCU1 ) and NewmansdgA/sdgBC ( pCU1 ) showed significantly lower biofilm formation ( P ≤ 0 . 001 ) than strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) and similar biofilm forming capacities like strains Newman ( pCU1 ) and NewmansdgA/sdgB ( pCU1 ) ( Fig 8A ) further verifying that indeed the production of GtfC/GtfD-glycosylated Pls is the cause of increased biofilm formation . Generally , biofilm formation proceeds in two steps: Rapid initial attachment of the bacteria to a surface is followed by a more prolonged accumulation phase , which requires intercellular adherence [17] . Intercellular adherence may be mediated by protein factors or PIA , a polysaccharide whose production is encoded by the icaADBC operon [17] . Another important structural component of S . aureus biofilms is extracellular ( e ) DNA [48] . To characterize the mechanisms involved in the increased biofilm formation mediated by GtfC/GtfD-glycosylated Pls , we analyzed the initial attachment of the bacteria to a plastic surface . We could not detect any significant differences in the number of attached bacteria among the strains tested suggesting that increased intercellular adherence must account for the observed differences in biofilm formation ( Fig 8B ) . In agreement , strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) producing GtfC/GtfD-glycosylated Pls also showed increased biofilm formation on a glass surface ( Fig 8C ) . To further dissect the mechanisms underlying the stimulated biofilm formation by GtfC/GtfD-glycosylated Pls , we treated preformed biofilms with proteinase K as well as with sodium metaperiodate ( NaIO4 ) . Proteinase K treatment completely abolished biofilm formation of all strains tested except for the control S . epidermidis RP62A , which is known to form a PIA-dependent biofilm , confirming the protein dependency not only of the biofilms mediated by GtfC/GtfD-glycosylated Pls , but also of strains Newman in general ( Fig 8D ) . Interestingly , treatment with NaIO4 , which oxidizes carbohydrates and is known to disintegrate PIA-dependent biofilms , significantly ( P ≤ 0 . 001 ) degraded only the biofilms of strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) producing GtfC/GtfD-glycosylated Pls to the levels of the remaining Newman strains strongly suggesting a direct involvement of the Pls sugar residues in biofilm formation ( Fig 8E ) . As expected , biofilms of the PIA-producing control S . epidermidis RP62A were also significantly degraded ( P ≤ 0 . 001 ) ( Fig 8E ) . Furthermore , growth in the presence of DNase I significantly ( P ≤ 0 . 001 ) reduced the biofilm levels of strains Newman ( pCU1 ) , NewmansdgA/sdgB ( pCU1 ) , Newman ( pΔPlsGtfCDCOL ) and NewmansdgA/sdgB ( pΔPlsGtfCDCOL ) , which all did not produce Pls , suggesting that eDNA is an important structural component of strain Newman biofilms ( Fig 8F ) . However , the biofilm levels of strains Newman ( pPlsGtfΔCΔDCOL ) and NewmansdgA/sdgB ( pPlsGtfΔCΔDCOL ) producing Pls not glycosylated by GtfC/GtfD were not noticeably altered when biofilms were grown in the presence of DNase I ( Fig 8F ) . In contrast , the higher biofilm levels of strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) producing GtfC/GtfD-glycosylated Pls were significantly ( P ≤ 0 . 001 ) reduced when grown in the presence of DNase I to the levels of strains Newman ( pPlsGtfΔCΔDCOL ) and NewmansdgA/sdgB ( pPlsGtfΔCΔDCOL ) and still remained significantly ( P ≤ 0 . 001 or P ≤ 0 . 01 ) higher than those of Newman strains not producing Pls when grown in the presence of DNase I ( Fig 8F ) . These results together strongly suggest that two distinct mechanisms are involved in biofilm formation mediated by GtfC/GtfD-glycosylated Pls , one depending on GtfC/GtfD-glycosylated Pls and potentially also on eDNA , while the other being independent of glycosylation by GtfC/GtfD as well as of eDNA . While with strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) both mechanisms can be observed , strains Newman ( pPlsGtfΔCΔDCOL ) and NewmansdgA/sdgB ( pPlsGtfΔCΔDCOL ) only display the latter .
In the past two decades , evidence has grown that bacterial glycoproteins play important roles in the physiology and pathophysiology of Gram-negative and Gram-positive bacteria , such as adherence to host cells , interaction with the host immune system , immune evasion , surface recognition , enzymatic activity , protein stability , and conformation [5–7 , 10] . The knowledge on glycosylated surface proteins in S . aureus , the underlying glycosylation machinery and their potential role in pathogenesis has been very limited so far . In a search for staphylococcal surface glycoproteins , we identified four glycosylated surface proteins from the MRSA strain COL and two from strain 1061 . The ∼250- and ∼165-kDa glycoproteins from strain COL and the ∼175-kDa glycoprotein from strain 1061 were identified as the plasmin-sensitive protein Pls by MS . The presence of the pls gene usually is associated with the SCCmec type I , but has also been found in one strain harboring the SCCmec type IV [42] . Pls is sensitive to proteolysis by plasmin leading to 175-kDa and 68-kDa cleavage products [26] . However , these cleavage products also occur in lysostaphin lysates without prior proteolytic treatment [28] . Cleavage could be prevented ( sometimes only partially ) by the addition of protease inhibitors suggesting that Pls is cleaved by an S . aureus protease at the same position ( between position R387 and A388 in Pls from strain 1061 [26] ) . This explains the presence of the large ∼250-kDa Pls glycoprotein in staphylococcal surface protein preparations or sometimes its absence due to proteolytic cleavage . The expression of pls encoded on plasmid pPLS4 in the MSSA strains SA113 , SH1000 , and Newman led to the production of a glycosylated version of Pls , which however was not produced by the respective sdgA/sdgB mutant strains ( Fig 2A and 2B ) demonstrating that SdgA/SdgB are capable of transferring glycosyl residues to Pls . In contrast GtfA , which mediates the glycosylation of SraP [20 , 21] , is not apparently involved in the glycosylation of Pls as there was no glycosylated Pls detectable in the sdgA/sdgB mutants and the glycosylation of Pls seemed to be unchanged in the SA113gtfA mutant . However , the COLsdgA/sdgB mutant still produced a glycosylated version of Pls ( Fig 2B ) thereby demonstrating that the genome of strain COL must contain additional gtf genes that are able to confer glycosylation of Pls . In a search for potential gtf genes in strain COL , we identified gtfC and gtfD encoded downstream of the pls gene on the SCCmec . Expression and deletion analysis in the strains S . aureus SA113sdgA/sdgB and S . carnosus TM300 revealed that both gtfC and gtfD are involved in the glycosylation of Pls ( Fig 3 and Fig 4 ) . Interestingly , we observed a difference among the genes cloned from strain COL in comparison with those cloned from strain 1061: While with strain COL , only gtfD is required for an initial glycosylation of Pls , with strain 1061 , both gtfC and gtfD are required . Nucleotide sequence analysis revealed that the deduced aa sequences of GtfCCOL and GtfC1061 are 100% identical and that there is only one aa exchange between GtfDCOL and GtfD1061 ( F208 ⇒ S208 ) potentially accounting for the observed difference . However , another possibility could be that sequence differences between Pls from strain COL and Pls from strain 1061 are responsible for the observed difference . Furthermore , it seems likely that in strain COL SdgA/SdgB additionally to GtfC/GtfD transfer carbohydrate residues to Pls . Indeed , our mass-spectrometric analysis indicate that SdgA/SdgB are also involved in the glycosylation of Pls , because Pls purified from the strain S . aureus COL is more heavily glycosylated than Pls purified from the strain COLsdgA/sdgB ( Table 2 ) . Our mass-spectrometric analysis demonstrated that modifying carbohydrates are N-acetylhexosaminyl residues . Future analysis of the modifying glycan moieties of Pls prepared from different gtf mutants will clarify , whether further Gtfs might be involved in the glycosylation of Pls and whether the modifying sugars consist of one or more than one species of N-acetylhexosamines . In analogy to GspB and other SRR surface proteins , we expect N-acetylglucosaminyl and/or N-acetylgalactosaminyl residues to be among the modifying carbohydrates [12 , 49 , 50] . Equivalent to other reported SRR proteins ( see below ) , we hypothesize a role for the glycan modifications in the function of Pls . As it was reported for Pls to promote bacterial cell aggregation [43] , a possible function of the Pls glycosyl residues is an involvement in Pls-mediated cell aggregation and biofilm formation . Indeed , we could identify a role for the Pls glycosyl residues in biofilm formation in strain Newman ( Fig 8 ) . Analysis of the participating factors in biofilm formation mediated by GtfC/GtfD-glycosylated Pls revealed a proteinase K-sensitive factor as expected , which is also in agreement with Hazenbos et al . ( 2013 ) , who found that proteinase K treatment of protein preparations resulted in a loss of reactivity with a monoclonal antibody that exclusively detected SD-repeat protein domains when they are glycosylated [15] . Most importantly , our further results strongly suggest that Pls mediates biofilm accumulation via two distinct mechanisms . The first mechanism requires Pls SD-repeat glycosylation by GtfC/GtfD and its sensitivity to NaIO4 indicates a direct involvement of the carbohydrate modifications in intercellular adherence . To our knowledge this is the first study to demonstrate the importance of sugar modifications of a staphylococcal surface glycoprotein in biofilm formation . This mechanism may involve the contribution of eDNA , because we observed a significant reduction of biofilm levels of strains Newman ( pPlsGtfCDCOL ) and NewmansdgA/sdgB ( pPlsGtfCDCOL ) by DNase I . However , we cannot exclude the possibility that this observation may at least partially be due to an overlap with the intrinsic eDNA dependency of strain Newman . The second mechanism of Pls-mediated biofilm formation seems to be independent of glycosylation as well as eDNA and becomes only evident upon biofilm growth of the strains Newman ( pPlsGtfCDCOL ) , NewmansdgA/sdgB ( pPlsGtfCDCOL ) , Newman ( pPlsGtfΔCΔDCOL ) , and NewmansdgA/sdgB ( pPlsGtfΔCΔDCOL ) in the presence of DNase I suggesting that otherwise this second mechanism is masked by the presence of eDNA . The second , glycosylation-independent mechanism likely involves the Pls G5 domains ( see Fig 3 ) . G5 domains are also part of the Pls-homologous proteins Aap from S . epidermidis and SasG from S . aureus and known to promote biofilm formation via a zinc-dependent self-association mechanism [51–55] . Aap and SasG however lack an SRR domain and therefore the G5 domain-mediated mechanism of biofilm formation must be independent of glycosylation . Taken together , the findings resulting from our cloning , expression , and biofilm studies enabled us to propose two distinct mechanisms involved in biofilm formation mediated by GtfC/GtfD-glycosylated Pls and it may be speculated that the bacteria depending on the actual environmental conditions may apply one or the other . Further studies are planned in the future to exactly decipher the mechanisms underlying biofilm accumulation mediated by GtfC/GtfD-glycosylated Pls . Strain Newman carries a variant saeSP allele instead of the saeSL allele resulting in an over-active SaeRS regulatory system [46 , 47 , 56] . Our findings suggest that the effect of increased biofilm formation mediated by GtfC/GtfD-glycosylated Pls depends on the saeSP allele . Although generally , the saeRS system seems to be quite conserved , it was recently reported that the saeSP allele is present in several uncharacterized S . aureus strains found at the Genomes OnLine Database ( GOLD ) ( https://gold . jgi . doe . gov ) ( IDs 53133–53147 ) [47] . Further analyses are required to determine , whether the saeSP allele might also be an occasional or even frequent feature of clinical MRSA . Similarly , variations can also occur in other regulatory loci: a substantial number of clinical S . aureus isolates have been found to be negative in the well-characterized accessory gene regulator agr [57] . Alternatively , it seems possible that an upregulated saeRS system via the saeSP allele may not be required in the in vivo situation . In support of this , several analyses revealed that saeRS is an in vivo-active and essential regulatory locus that plays a crucial role in S . aureus virulence , which has also been shown during human and mouse infection with MRSA [58–61] . However , the regulatory mechanisms underlying increased biofilm formation mediated by GtfC/GtfD-glycosylated Pls in strain Newman still have to be elucidated . It has been previously established that Pls decreases the S . aureus adherence to extracellular matrix and plasma proteins including Fg , Fn , IgG , and laminin and also its internalization by human host cells by the mechanism of steric hindrance [29 , 41 , 42] . To study whether the glycosyl residues of Pls mediate steric hindrance , we performed different functional ELISA and flow-cytometric assays . We could neither detect an impact of the sugar modification on the Pls-mediated decrease of S . aureus SA113 adherence to Fg , Fn , and endothelial cells , nor in the decrease of its internalization by endothelial cells or of its phagocytosis by PMNs . Thus , we can rule out the possibility that the steric hindrance is caused by the glycosyl residues masking other surface adhesins and leading to the observed effects of Pls . The >300 kDa glycosylated surface protein produced by strains COL and 1061 ( Fig 1 ) probably represents SraP ( SasA ) [20] , because we identified a glycosylated protein with the same size produced by the strain SA113sdgA/sdgB as SraP by MS ( S1 Fig ) . The ∼120-kDa glycoprotein identified from strain COL that was missing from the COLsdgA/sdgB mutant ( Fig 2B ) might either be SdrC , SdrD , or SdrE [62] , because sdgA/sdgB are located downstream of the sdrCDE locus and the genes encoding Gtfs are frequently encoded adjacent to the structural genes , whose products they glycosylate [12 , 49 , 50] . Alternatively , the ∼120-kDa glycoprotein might be ClfA [15 , 16] . The absence of the ∼120-kDa glycoprotein from strain 1061 might be explained by non-functional sdgA/sdgB genes in strain 1061 . Interestingly , our nucleotide sequence analysis revealed the insertion of an IS1181 element upstream of the sdgB gene in strain 1061 thereby potentially influencing sdgB transcription ( S3 Fig ) . Alternatively or additionally , a non-functional SdgA or SdgB might be explained by 6 and 3 aa exchanges found in SdgA and SdgB from strain 1061 , respectively , compared to SdgA and SdgB from strain COL ( which are identical to SdgA and SdgB from strain SA113 ) . Besides SraP and ClfA from S . aureus [14–16] and GspB from S . gordonii [63] , further members of the growing family of SRR surface proteins include Hsa from S . gordonii , which is homologous to GspB [64] , SrpA of Streptococcus sanguis [50] , PsrP of Streptococcus pneumoniae [24] , Srr1 and its homolog Srr2 from Streptococcus agalactiae [65] , and Fap1 from Streptococcus parasanguinis [11 , 12] . SRR proteins have been associated with different adhesive functions and with bacterial pathogenesis . Like SraP and GspB , SrpA binds to platelets and it has been shown in animal models of infective endocarditis that their expression is associated with a higher pathogenicity [14 , 50 , 66] . With GspB , it has been demonstrated that incorrect glycosylation leads to impaired binding to its platelet receptor [49] . Srr1 mediates binding to several types of human epithelial cell lines and interacts with cytokeratin 4 as an epithelial cell surface ligand , which seems to involve the glycosylated SRR domain of Srr1 [67] . Furthermore , it was shown that the extent of Srr1 glycosylation by GtfCDEFGH modulates the adherence and virulence of S . agalactiae in a rat model of neonatal sepsis [65] . The fimbria-associated protein Fap1 from S . parasanguinis that colonizes saliva-coated teeth thereby causing the formation of dental plaque mediates biofilm formation in an in vitro tooth model , which seems to involve the sugar residues [68 , 69] . PsrP from S . pneumoniae binds to keratin 10 on lung epithelial cells and mediates bacterial cell aggregation [70 , 71] . Similarly , it was shown that GspB and SraP promote bacterial aggregation [71] . Thus , in several SRR proteins , the glycan moieties of the proteins seem to be involved in or to modulate the functions of the respective adhesins , which is in line with our finding that Pls confers increased biofilm formation when glycosylated by GtfC/GtfD . Although there are several common features among the SRR proteins of Gram-positive cocci and the pls locus shares some of them , such as the structural gene encoding a large SRR surface protein and gtf genes that are located downstream of the structural gene and encode enzymes involved in posttranslational modification , there also seem to be marked differences . Like GtfA from the S . aureus sraP locus and GtfA from the S . gordonii M99 gspB locus , which share more than 40% identical aa with the poly ( glycerol-phosphate ) α-glucosyltransferase TagE of Bacillus subtilis [13] , GtfC also has a high degree of identical aa with poly ( glycerol-phosphate ) α-glucosyltransferases ( see above ) suggesting similar functions of GtfA and GtfC . Here , we found that S . carnosus and the SA113sdgA/sdgB mutant produced non-glycosylated , surface-anchored Pls upon pls expression , when gtfC/gtfD are either not present or deleted . In contrast , in a gtfA and orf4 ( later termed gtfB ) mutant of S . gordonii , GspB was not detectable [19] . This was not due to an altered gspB transcription in these mutants . Thus , the authors concluded that either GspB is not translated or quickly degraded intracellularly and thus the Gtfs may greatly affect the stability of GspB [13 , 19] . Similarly , GtfA and/or GtfB is essential for the production of Srr1 , while full glycosylation of Srr1 mediated by the six dispensable additional Gtfs ( GtfCDEFGH ) leads to the cell surface display of a protein that is protected from proteolysis [65] . Moreover , the non-glycosylated Fap1 protein is less stable and more sensitive to protein degradation upon inactivation of the gtf gene that mediates glycosylation of Fap1 [69] . Recently , modifying glycosyl residues have also been demonstrated to protect ClfA from proteolytic cleavage by host proteases and might therefore modulate its function as an adhesin [15] . However , in our preliminary experiments , we could not detect a difference in protein stability or secretion of glycosylated versus non-glycosylated Pls ( Fig 3 ) . In conclusion , Pls is a glycoprotein and GtfC/GtfD as well as SdgA/SdgB are involved in its glycosylation . The production of GtfC/GtfD-glycosylated Pls leads to increased biofilm , while glycosyl residues do not have an impact on other previously known Pls properties . Because Pls has been shown to be a virulence determinant in a mouse septic arthritis model [25] , it is reasonable to assume that glycosyl residues might contribute to in vivo biofilm formation . Future experiments are planned to clarify , if the sugar modifications of Pls may represent promising new targets for therapeutic or prophylactic measures .
Bacterial strains used in this study are listed in Table 1 . Staphylococcus and Escherichia coli strains were grown aerobically at 37°C in Tryptic Soy ( TS ) broth ( TSB , BD Bioscience ) and lysogeny broth ( LB , BD Bioscience ) , respectively . TS and LB agar plates contained 1 . 4% agar . Staphylococcal cultures were cultivated in TSB unless otherwise indicated . Antibiotics were added , when appropriate: Ampicillin ( Am; 100 μg/ml ) , chloramphenicol ( Cm; 10 μg/ml ) , erythromycin ( Em; 10 μg/ml ) , tetracycline ( Tc; 10 μg/ml ) , and kanamycin ( Kan; 25 μg/ml ) . For the cloning of pls , gtfC , and gtfD , the vector pCU1 was used [72] . The sdgA/sdgB-deficient S . aureus SA113 mutant was constructed by using the plasmids pEC2 and pBT2 [73] . The sdgA/sdgB-deficient S . aureus SH1000 mutant was constructed by using the plasmids pGL433 [74] and pMUTIN4 [75] . For the construction of the mutants SA113gtfA [20] , SA113gtfA/sdgA/sdgB , and SA113bgt , the vector pKOR1 was employed [76] . Pls subclones were constructed from the plasmid pPLS4 [26] . For the transduction of the sdgA/sdgB double mutation from S . aureus SH1000sdgA/sdgB into strains COL and Newman , the phage Φ11 was used [77] . For the internalization and adherence assays , the endothelial cell line EA . hy 926 ( ATCC CRL-2922 ) was employed [78] . Cultivation of the EA . hy 926 cells was performed as described [79] . Micro Bio-Spin P6 Columns were purchased from BIO-RAD ( Munich , Germany ) . Trypsin , chymotrypsin , endoproteinase Glu-C , and pronase were from Roche Diagnostics GmbH ( Mannheim , Germany ) . Thermolysin was purchased from Sigma-Aldrich Chemie GmbH ( Taufkirchen , Germany ) . Methanol , formic acid , and acetic acid were from Fluka ( Buchs , Switzerland ) . All solvents used were of HPLC grade purity . DNA manipulations and transformation of E . coli were performed according to standard procedures [80] . S . carnosus and S . aureus strains were transformed by protoplast transformation [81] or electroporation [82] . Plasmid DNA was isolated using the PrepEase MiniSpin Plasmid Kit ( USB , Staufen , Germany ) and staphylococcal chromosomal DNA was isolated with the PrestoSpin D Bug DNA purification kit ( Molzym , Bremen , Germany ) . PCR was carried out using the Phusion High-Fidelity DNA Polymerase ( Finnzymes , Vantaa , Finland ) according to the instructions of the manufacturers . The primers ( Table 3 ) were synthesized by Eurofins MWG Operon ( Ebersberg , Germany ) . DNA sequences were determined by Eurofins MWG Operon using the indicated primers ( Table 3 ) and an ABI 3730XL DNA sequencer . The DNA and deduced protein sequences were analyzed using the program JustBio at http://www . justbio . com . The deduced Pls and GtfC/GtfD sequences were compared using the programs BLASTP [83] and FASTA [84] and the alignments were done using the program ClustalW at the European Bioinformatics Institute ( EBI , Cambridge , UK ) . The signals obtained by MS were assigned to peptides of known proteins by using the MASCOT search engine and the SwissProt database at http://www . expasy . ch . The CAZy database used for the identification of putative Gtfs encoded by the S . aureus COL genome is available at www . cazy . org . The accession numbers of the deduced sequences of the UniProt and GenBank databases are: Pls ( SACOL0050 ) : Q5HJU7 , AAW38699; GtfC ( SACOL0051 ) : Q5HJU6 , AAW38700; GtfD ( SACOL0052 ) : Q5HJU5 , AAW38701; SdgA ( SACOL0611 ) : Q5HIB1 , AAW37720; SdgB ( SACOL0612 ) : Q5HIB0 , AAW37721; poly ( glycerol-phosphate ) α-glucosyltransferase from S . aureus C75S: ACZ59060; poly ( glycerol-phosphate ) α-glucosyltransferase from S . epider-midis ATCC 12228: NP_765949 . The nucleotide sequence accession numbers are for the sdgA/sdgB genes from strain S . aureus 8325–4: SAOUHSC_00547 ( sdgA ) and SAOUHSC_00548 ( sdgB ) , for gtfA: SAOUHSC_02984 [20] , and for the putative bactoprenol glycosyltransferase bgt: SAOUHSC_00713 . The GenBank nucleotide sequence accession number for the gtfC/gtfD genes from strain S . aureus 1061 is JX193902 and for the sdgA/sdgB genes including the adjacent sequence of the insertion sequence IS1181 from strain S . aureus 1061 is JX204384 . The Pfam accession number for the G5 domain is available at http://pfam . xfam . org/family/PF07501 . The sdgA/sdgB genes are colocalized in the same locus on the chromosome . The double mutant S . aureus SA113sdgA/sdgB was created by the replacement of the sdgA/sdgB genes with the antibiotic resistance marker ermB . Briefly , both DNA fragments of approximately 1 kbp flanking the sdgA/sdgB locus were PCR amplified with the primer pairs TE-P3-HindIII/TE-P1-PstI and TE-P1-XbaI/TE-P4-EcoRI , respectively ( Table 3 ) . Both , the upstream and downstream DNA fragments were restricted , purified , and ligated into the pBT2 vector together with a 1 . 1 kbp PstI-XbaI fragment encoding the ermB gene taken from the plasmid pEC2 . S . aureus SA113 was transformed with the resulting knock-out plasmid pBT-sdgA/sdgB by electroporation . By incubation at 42°C and subsequent screening for Em-resistant clones without the plasmid-encoded Cm resistance , the sdgA/sdgB mutant , was identified . Similarly , the mutants SA113bgt , SA113gtfA , and SA113gtfA/sdgA/sdgB were constructed by using the vector pKOR1 . The S . aureus SH1000sdgA/sdgB mutant was constructed using the primer pairs Rmgts4/Rmgts5 and Rmgts6/Rmgts7 and the Kan resistance cassette , which was PCR amplified from the plasmid pGL433 using the primer pair Fkan1/Rkan2 ( Table 3 ) . To introduce the sdgA/sdgB mutation from strain SH1000sdgA/sdgB into strains COL and Newman , phage transduction was performed using Φ11 as the transducing phage as described [77] . To analyze the potential of GtfC and/or GtfD to glycosylate Pls , the pls gene and the downstream located genes , gtfC and gtfD , including the ribosomal binding sites and putative promoter sequences were amplified by PCR from S . aureus COL and 1061 genomic DNA using the primers PlsGtfCD-F and PlsGtfCD-R ( Table 3 ) yielding a 9 . 81 kbp DNA and a 10 . 08 kbp fragment , respectively . The DNA fragments were cloned into the KpnI site of the shuttle vector pCU1 in E . coli , generating the plasmids pPlsGtfCDCOL and pPlsGtfCD1061 . To functionally delete gtfC on the plasmids pPlsGtfCDCOL and pPlsGtfCD1061 , the plasmid DNA was restricted by Eco47III and EcoRV and religated leading to a deletion of 478 bp and creating plasmids pPlsGtfΔCDCOL and pPlsGtfΔCD1061 . To functionally delete gtfD on the plasmids pPlsGtfCDCOL and pPlsGtfCD1061 , they were restricted by BglII . Then , the sticky ends were refilled by the Klenow fragment and religated generating a frameshift mutation and plasmids pPlsGtfCΔDCOL and pPlsGtfCΔD1061 . Plasmids pPlsGtfΔCΔDCOL and pPlsGtfΔCΔD1061 were constructed by introducing the frameshift mutation in gtfD in the plasmids pPlsGtfΔCDCOL and pPlsGtfΔCD1061 as described above . To functionally delete pls , plasmid pPlsGtfCDCOL was restricted by HpaI and XbaI resulting in a deletion of 4 , 736 bp , the XbaI sticky end was made blunt end by the Klenow fragment and the DNA fragment was religated yielding plasmid pΔPlsGtfCDCOL . The sequences of the pls and gtfC/gtfD genes and their deletion derivatives were verified by DNA sequencing of the respective plasmids using the primers listed in Table 3 . Subsequently , plasmids pPlsGtfΔCΔDCOL and pPlsGtfΔCΔD1061 were introduced into strain S . aureus SA113 , all plasmids except for pΔPlsGtfCDCOL were introduced into S . aureus SA113sdgA/sdgB and S . carnosus TM300 , and plasmid pΔPlsGtfCDCOL was introduced into strain S . aureus Newman and S . aureus NewmansdgA/sdgB . To analyze the involvement of the SD-repeat region in the glycosylation of Pls , we constructed different subclones from plasmid pPLS4 [26] by using inverse PCR and the primer pairs: Pls4Sub1-R/Pls4Sub1/2/3-F to generate pPLSsub1 ( 9 . 51 kbp fragment ) , Pls4Sub2-R/Pls4Sub1/2/3-F to generate pPLSsub2 ( 9 . 56 kbp fragment ) , and Pls4Sub3-R/Pls4Sub1/2/3-F to generate pPLSsub3 ( 9 . 85 kbp fragment ) ( Table 3 ) ( Fig 5A ) in E . coli . After passaging the plasmids in S . aureus SA113 , they were introduced into S . aureus 1061pls . Surface-associated proteins of staphylococcal strains were solubilized from the cell surface by heating with SDS-sample buffer essentially as described before [85] . Staphylococcal surface proteins covalently linked to the peptidoglycan were prepared by lysostaphin treatment of cultures that were grown overnight in TSB as described [86] . To prepare the lysostaphin lysates for the purification of Pls via lectins , staphylococcal strains were grown overnight in Todd-Hewitt broth ( BD Bioscience ) . Cells were harvested , washed in phosphate-buffered saline ( PBS ) , and resuspended in 40 ml PBS . Then , 200 μl of lysostaphin ( 5 μg/ml ) , 50 μl of DNase ( 1 mg/ml ) and protease inhibitors ( complete EDTA-free protease inhibitor cocktail; Roche ) were added and incubated at 37°C for 2 h . The lysates were centrifuged at 13 , 000 rpm for 20 min . The supernatant was heated to 80°C to stop the reaction , centrifuged again , and sterile filtered . A 1 ml column packed with ConA sepharose 4B ( GE Healthcare , München , Germany ) was equilibrated with binding buffer ( 20 mM Tris-HCl , 0 . 5 M NaCl , pH 7 . 4 ) according to the instructions of the supplier . Afterwards , the lysate was applied to the column , the flow-through was collected and reapplied to the column thrice . The column was then washed with 25 ml binding buffer and bound protein was eluted with 10 ml elution buffer ( binding buffer containing 15% methyl α-D-glucopyranoside ) ( Sigma Aldrich , München , Germany ) . The eluted fractions were separated by SDS-PAGE to check for the presence of protein . Fractions containing protein were passed through a NAP-10 G25 column ( GE healthcare ) to remove the small methyl α-D-glucopyranoside . Staphylococcal surface , surface-associated or purified proteins were separated by SDS-PAGE ( 10% or 7 . 5% separation gel , 4 . 5% stacking gel ) and stained with Coomassie Brilliant Blue G-250 . Glycoproteins were detected using Pierce Glycoprotein Staining Kit ( Thermo Scientific , Schwerte , Germany ) in accordance with the protocol of the supplier by staining the sugar moieties directly in the SDS gel . 50 μl of a Pls solution ( 5 to 15 pmol/μl , 50 mM Tris/HCL or 100 mM NH4HCO3 , pH 7 . 4 to 7 . 8 ) were transferred to distilled water by use of Micro Bio-Spin P6 columns according to the manufacturer’s instructions . Briefly , the column was equilibrated with distilled water , the sample was applied to the column , and the protein was eluted with distilled water . 25 μl aliquots of the eluate were adjusted to 12 . 5% acetic acid in a total volume of 50 μl and incubated for 2 h at 95°C . Subsequently , the solvent was evaporated in vacuo and the residue was redissolved in 40% methanol/0 . 5% formic acid for mass-spectrometric analysis . 50 μl of a Pls solution ( 5 to 15 pmol/μl , 50 mM Tris/HCL or 100 mM NH4HCO3 , pH 7 . 4 to 7 . 8 ) were rebuffered to 25 mM NH4HCO3 by use of Micro Bio-Spin P6 columns as described above . Aliquots corresponding to 100 to 200 pmol were incubated in the presence of trypsin , chymotrypsin , endoproteinase Glu-C ( 0 . 2 μg each ) or pronase ( 1 μg ) overnight at 37°C . For digests with thermolysin , 0 . 5 μg of the protease were added and the mixture was incubated overnight at 65°C . The bands containing the glycosylated proteins were excised from the polyacrylamide gel and prepared for MS . For this , proteins were digested tryptically in the gel and the peptides were extracted , desalted , and subjected to electrospray ionization on a Q-Tof Premier coupled to a Nano Acquity ( Waters Micromass , Eschborn , Germany ) at the Integrated Functional Genomics ( IFG ) Core Unit of the Interdisciplinary Center of Clinical Research ( IZKF ) at the University Hospital of Münster ( Germany ) . The obtained signals were assigned to peptides of known proteins by using the MASCOT search engine and the SwissProt database and the ProteinLynx Global SERVER ( PLGS ) software ( Waters Micromass ) . The products of acid hydrolysis and proteolytic cleavage were analyzed by nanoESI Q-Tof MS and MS/MS and chosen ( glyco ) peptide structures were deduced from fragment ion spectra derived from CID . NanoESI MS experiments were carried out by use of a quadrupole time-of-flight ( Q-Tof ) mass spectrometer ( Micromass , Manchester , UK ) equipped with a Z-spray source in the positive ion mode . The source temperature was kept at 80°C and the desolvation gas ( N2 ) flow rate at 75 l per h . The capillary and cone voltages were adjusted to 1 . 1 kV and 30 V , respectively . For low energy CID experiments , the ( glyco ) peptide precursor ions were selected in the quadrupole analyzer and fragmented in the collision cell using a collision gas ( Ar ) pressure of 3 . 0 × 10−3 Pa and collision energies of 30–60 eV ( Elab ) . The wells of 96-well microplates were coated with fibrinogen ( Fg , 20 μg/ml; Calbiochem ) , fibronectin ( Fn , 10 μg/ml; Roche ) , or as a negative control with blocking buffer ( protein-free blocking buffer , Thermo Fisher Scientific ) at 4°C overnight and subsequently blocked . To assess the adherence to endothelial cells , EA . hy 926 cells were grown to confluence in 96-well cell culture plates ( Greiner Bio-One ) , washed with PBS , fixed with ice-cold methanol ( Merck ) and blocked with blocking buffer . Then , the microplates were washed thrice and each well was incubated for 2 h at 37°C with 100 μl of a staphylococcal suspension , which was previously grown overnight , washed with PBS , sonicated using an ultrasonic cell disruptor ( Branson Sonifier 250 ) to separate cell aggregates , and adjusted to an optical density ( OD578 ) of 1 . 0 ( corresponding to approximately 5 x 108 cfu/ml ) . As negative controls , wells without bacteria were included . Unbound bacterial cells were removed by washing twice with 200 μl PBS . Bound S . aureus cells were detected by a polyclonal rabbit anti-S . aureus antibody ( previously raised in rabbits by Eurogentec , Liège , Belgium ) ( diluted 1:1 , 500 ) and alkaline phosphatase-conjugated goat anti-rabbit IgG ( diluted 1:2 , 000 [0 . 32 μg/ml] , Dako ) . SigmaFast p-Nitrophenylphosphate ( Sigma Aldrich ) conversion was detected by determination of the OD405 after 30 min of incubation . Overnight-grown staphylococci were washed , sonicated , fixed , and fluorescein isothiocyanate ( FITC isomer I; Invitrogen ) -labeled as described before [79] . Sample preparation and detection of internalized staphylococci by EA . hy 926 cells were performed by flow-cytometric internalization assays as described before [79] . The phagocytosis assay was performed as described [87] . Briefly , PMNs were freshly isolated from Na citrate-treated blood from healthy donors by density gradient centrifugation using Ficoll-Paque Plus ( Amersham Bioscience ) according to the manufacturer's instruction . FITC-labeled bacteria were added to the PMNs and incubated . Samples were analyzed on a FacsCALIBUR ( BD Bioscience ) . Electronic gating was used to analyze 5 , 000 PMNs in each sample . The FL1 photomultiplier ( transmittance at 500 nm ) was used to detect uptake of staphylococcal cells by PMNs . For quantification of the biofilm-forming capacity , a biofilm assay was performed essentially as described previously [88] . Briefly , strains were grown in TSB for 24 h at 37°C in 96-wells polystyrene microtiter plates ( cell star; Greiner , Frickenhausen , Germany ) . Afterwards , the plates were emptied , the wells were washed with PBS and adherent biofilms were stained with 0 . 1% safranin ( Serva ) . In some experiments , 24-h biofilms were washed with PBS and then treated with 0 . 1 mg/ml proteinase K ( Sigma ) in 20 mM Tris-HCl ( pH 7 . 5 ) or with 40 mM NaIO4 ( Applichem ) in double-distilled H2O for 2 h at 37°C . In the respective untreated controls , 24-h biofilms were incubated with 20 mM Tris-HCl ( pH 7 . 5 ) or double-distilled H2O for 2 h at 37°C . Furthermore , in some experiments 24-h biofilms were grown in the presence of 0 . 1 mg/ml DNase I as described [89] . Afterwards , the wells were emptied , washed with PBS and stained with 0 . 1% safranin . Absorbance was measured with a Micro-ELISA-Autoreader at 490 nm . Strains were tested at least in quadruplicates . Determination of biofilm formation on a glass surface was carried out essentially in the same way , except that 5 ml TSB were inoculated in glass tubes . Initial attachment of the bacteria to a plastic surface was tested essentially as described before with some modifications [88] . Briefly , diluted bacterial cell suspensions in 2 ml PBS were incubated in the wells of a Nunc Lab-Tek Chamber Slide-System ( Thermo Scientific ) for 30 min at 37°C and after two washing steps , attached bacteria were evaluated by phase-contrast microscopy , photographed and counted; the number of adhered cells per square millimeter was determined . All phagocytosis experiments were performed with the healthy adult blood donors giving written informed consent according to human experimentation guidelines . The study was conducted according to the principles expressed in the Declaration of Helsinki and was approved by the local ethics committee ( Ethikkommission der Ärztekammer Westfalen-Lippe und der Medizinischen Fakultät der WWU Münster ) ( reference number: Sitzung 19 . 05 . 1999 ) . Mean values of experimental data were compared with one-way ANOVA and , if adequate with subsequent Bonferroni’s posttest for multiple comparisons using GraphPad Prism 5 . P values ≤ 0 . 05 were considered statistically significant and are indicated with asterisks: * ( P ≤ 0 . 05 ) , ** ( P ≤ 0 . 01 ) , and *** ( P ≤ 0 . 001 ) . | Staphylococcus aureus is a serious pathogen that causes life-threatening infections due to its ability to attach to surfaces , form biofilms , and persist inside the host . One of previously identified virulence factors in S . aureus pathogenesis is the plasmin-sensitive surface protein Pls . We here identified Pls as a posttranslationally modified glycoprotein and characterized the domain within Pls that becomes glycosylated as well as the modifying sugars . Moreover , we found that the glycosyltransferases GtfC and GtfD carry out the glycosylation reactions . In a search for a role for the modifying sugars , we found that Pls can stimulate biofilm formation apparently via two distinct mechanisms , one being dependent on glycosylation by GtfC and GtfD the other being independent of glycosylation as well as eDNA . Moreover , we found that none of the already known Pls functions is mediated by the sugar moieties . Thus , we conclude that GtfC/GtfD-glycosylated Pls may contribute to MRSA pathogenicity via stimulation of biofilm formation and may serve as future target to combat or prevent infections with this serious pathogen . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bacteriology",
"biofilms",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"microbiology",
"staphylococcus",
"aureus",
"plasmid",
"construction",
"methicillin-resistant",
"staphylococcus",
"aureus",
"dna",
"construction",
... | 2017 | The Plasmin-Sensitive Protein Pls in Methicillin-Resistant Staphylococcus aureus (MRSA) Is a Glycoprotein |
Complex structural connectivity of the mammalian brain is believed to underlie the versatility of neural computations . Many previous studies have investigated properties of small subsystems or coarse connectivity among large brain regions that are often binarized and lack spatial information . Yet little is known about spatial embedding of the detailed whole-brain connectivity and its functional implications . We focus on closing this gap by analyzing how spatially-constrained neural connectivity shapes synchronization of the brain dynamics based on a system of coupled phase oscillators on a mammalian whole-brain network at the mesoscopic level . This was made possible by the recent development of the Allen Mouse Brain Connectivity Atlas constructed from viral tracing experiments together with a new mapping algorithm . We investigated whether the network can be compactly represented based on the spatial dependence of the network topology . We found that the connectivity has a significant spatial dependence , with spatially close brain regions strongly connected and distal regions weakly connected , following a power law . However , there are a number of residuals above the power-law fit , indicating connections between brain regions that are stronger than predicted by the power-law relationship . By measuring the sensitivity of the network order parameter , we show how these strong connections dispersed across multiple spatial scales of the network promote rapid transitions between partial synchronization and more global synchronization as the global coupling coefficient changes . We further demonstrate the significance of the locations of the residual connections , suggesting a possible link between the network complexity and the brain’s exceptional ability to swiftly switch computational states depending on stimulus and behavioral context .
Structural neural connectivity and its implications for brain function have been a long-sought subject in neuroscience . Many previous studies have been limited either to small networks of few cells or coarser connectivity among larger brain regions [1–9] , often binarized and without spatial information . Recent development of the Allen Mouse Brain Connectivity Atlas from anterograde fluorescent viral tracing experiments [10] provides us the unique opportunity to investigate precise weighted anatomical connectivity of the mammalian whole brain network . Combining the mesoscopic connectivity data with spatial information of the network , we seek a parsimonious representation of the weighted whole-brain network that captures salient network properties . Specifically , we investigate whether the network can be compactly represented solely based on the spatial dependence of the network topology . Biological networks are inherently spatially constrained . Recent studies have shown that geographic constraints play a critical role in generating graph properties of real-world neuronal networks [5 , 11–20] , which cannot be fully captured by classical generative network models such as the small-world network [2] and the scale-free network [21] . Yet many of the studies are limited to binarized networks [11 , 12 , 17 , 19 , 20] and are focused explicitly on comparing graph theoretical measures [11 , 13–20] . In this paper , we examine spatial embedding of a weighted whole-brain connectivity , and ask whether spatial dependence alone can depict the full computational capability of the brain network by studying dynamics of the network . By analyzing the latest connectivity data from a new mapping algorithm , we find that the network connectivity strongly depends on its spatial embedding , with spatially close brain regions strongly connected and distal regions weakly connected . We study the precise relationship between connectivity and distance , and investigate possible computational roles of positive residual connection strengths that are not captured by the spatial dependence . To probe the possible implications of the residual connections on the network dynamics , we construct a network of phase oscillators with the data-driven adjacency matrix and compare its dynamics to those of the oscillator network with the connections strictly dependent on distance . We analyze spatial structures of synchronization by measuring the order parameter for varied amounts of global coupling coefficient . We further examine the strong connections between distal brain regions by studying network dynamics when fractions of the strong residual connections are added to the spatially constrained network . Finally , we relocate the positive residuals either to connections between nearby brain regions or to different fractions of longest-range connections , thus increasing the connection strengths for the spatially close or distal brain regions while eliminating sparse , strong connections spread across different edge lengths . The networks restructured this way maintain overall connection strength of the brain network but have a connectivity topology different from that of the brain network . By comparing dynamics of such restructured networks and the data-driven whole brain network , we show that the spatial locations of the strong positive residuals are important . Specifically , our study reveals that strong connections distributed over the brain network across many length-scales enhance the capability of the system to switch between asynchronous and synchronous states , underlining the significance of the existence of these connections . The network without these long-range connections , as well as the network in which these long-range connections are shuffled , when pushed by perturbations or low coupling coefficient , lose global synchronization but maintain local synchronization over small spatial scales . In the same conditions , the data-driven network loses synchronization over all spatial scales . It is interesting to speculate that this phenomenon is necessary for the integrative processes necessary for global cognitive functions .
The mesoscopic mouse whole-brain connectivity was constructed based on viral tracing experiments available on the Allen Mouse Brain Connectivity Atlas [10] with a recently developed interpolative mapping algorithm [22] . This produced a weighted and directed structural connectivity matrix with 244 brain regions as source nodes and 488 brain regions as ipsilateral ( 244 ) and contralateral ( 244 ) target nodes . By combining the ipsilateral and contralateral connections for each hemisphere , we constructed a whole-brain connectivity matrix with 488 nodes . The data-driven mouse brain network is shown in Fig 1A , left column . We analyzed the relationship between connection strength and spatial distance between brain regions in the data set . In accordance with previous studies on brain networks [5 , 15–20] , the connectome strongly depends on the spatial embedding; connections are stronger between spatially close regions and weaker between distal regions . Specifically , the connection strengths decrease with distances between brain regions following a power law ( Fig 1B ) rather than an exponential relationship , in agreement with previous studies on Allen Mouse Brain Connectivity data [18 , 22] . Additional details on the fitting are available in Methods ( “Dependence of connection strengths on interregional distance” ) . We constructed adjacency matrices for the ipsilateral and the contralateral networks based on the power-law relationship , as shown in Fig 1A , middle column . While the general trend of decrease in connection strength with distance is clear and well-predicted by a power law , there are also a number of residual connection strengths that are not captured by the power-law relationship ( Fig 1A , right column ) . To understand the structure and effects of the residual connection weights that are not captured by the power-law dependence on distance , we had a closer look at these residuals . For both ipsilateral and contralateral connections , a long , positive tail is observed in the distribution of residual connection weights , suggesting strong distal connections above the power-law dependence on distance ( Fig 2A and 2B ) . The strongest 20 residual connections are plotted in Fig 2C . We observed that for the ipsilateral network , connections from preparasubthalamic nucleus ( PST ) to subthalamic nucleus ( STN ) , laterodorsal tegmental nucleus ( LDT ) to Barrington’s nucleus ( B ) , dorsal motor nucleus of the vagus nerve ( DMX ) to gracile nucleus ( GR ) , cuneate nucleus ( CU ) to gracile nucleus ( GR ) , and locus ceruleus ( LC ) to Barrington’s nucleus ( B ) are a few examples of the strong distal connections unexplained by the power-law dependence on distance . For the contralateral connectivity , on the other hand , many of the strongest residuals above the power-law relationship include connections between the same regions in different hemispheres as well as connections to and from hippocampal areas . Do these positive residual connections between distal regions have any computational significance ? In other words , can we capture the full computational capacity of the mesoscopic brain network with connectivity governed by strictly distance-dependent rules , with the residuals removed ? To test this , we compare dynamics of the data-driven brain network to those of an artificial , strictly distance-dependent network generated by the power-law relationship . Specifically , we built a network of coupled phase oscillators whose coupling strengths are described by the weighted adjacency matrix of the data-driven brain network or the power-law distance-dependent connectivity . Each of these Kuramoto-type phase oscillators corresponds to a brain area . Kuramoto-type coupled phase oscillators have been widely used to model oscillatory brain dynamics [23–25] . The phase of region i , represented by θi , is described by: θ ˙ i= ω i + k ∑ j = 1 N A i j sin ( θ j ( t - τ i j ) - θ i ( t ) ) + η i ( t ) ( 1 ) where ωi denotes the natural frequency , and k describes the coupling coefficient . Aij is the adjacency matrix of the network . For the case of the data-driven brain network , Aij = Jij where Jij indicates the adjacency matrix obtained from viral tracing data , for both ipsilateral and contralateral connections . For simulations of the artificial , distance-dependent network , Aij = Kij indicates the adjacency matrix constructed by making the connection weights strictly follow the power-law dependence on distance . The last term ηi ( t ) represents an additive Gaussian white noise with zero mean ( 〈ηi ( t ) 〉t = 0 ) and variance σ n 2 / T ( 〈 η i ( t ) η j ( t ′ ) 〉 t = δ i j δ ( t - t ′ ) σ n 2 / T ) , where δij is the Kronecker delta and δ ( ⋅ ) denotes the Dirac delta function . The standard deviation σn is in radians and T is a timescale , which is set to 1 second in our study . N denotes the number of nodes of the network , which is 488 in our whole-brain simulations . The natural frequencies ωi are randomly chosen from a symmetric , unimodal distribution g ( ω ) . In this paper , we used a Gaussian distribution with the mean at 40 Hz and the standard deviation σd for g ( ω ) , as done in other studies of modeling large-scale brain dynamics with phase oscillators [24–27] . Note that this falls within a frequency range of gamma rhythms ( 30-80 Hz ) that are frequently observed in oscillatory brain dynamics . Numerous previous studies have shown the importance of distance-dependent delays in networks of oscillators [28–33] . For example , time delays can destabilize synchrony in neuronal networks , leading to travelling waves [29–33] . To reproduce synaptic and axonal conduction delays dependent on connection distance , we incorporated distance-dependent time delays in our model as done in other studies [23–27 , 34–37] . In the rodent brain , the conduction velocity ranges from values as low as 0 . 5 ( m/s ) to much higher speed around 10 ( m/s ) depending on various factors such as axonal myelination [32 , 38 , 39] . Experimental studies show that the propagation speed distributions peak in between 2-5 ( m/s ) [32 , 39] . While time delays are heterogeneous over different regions in the brain , we simplified the model by using a fixed conduction speed at 3 . 5 ( m/s ) for the whole brain , which falls in the middle of the propagation speed distribution peak . In Eq 1 , the distance-dependent time delay between areas i and j is denoted by τij , which is computed by dividing the Euclidean distance dij between nodes i and j by the fixed conduction speed . We investigated the dynamics of the data-driven network and the power-law generated network using Eq 1 , and measured the network coherence by calculating the “universal” order parameter r , recently proposed in [40] as following: r≡1∑i=1Nki∑i , j=1NAij〈Re ( ei ( θi−θj ) ) 〉t=1∑i=1Nki∑i , j=1NAij〈cos ( θi−θj ) 〉t ( 2 ) where k i = ∑ j = 1 N A i j is the input strength of node i . Unlike the original order parameter which was proposed by Kuramoto [41 , 42] for all-to-all coupled phase oscillators ( see Eq 6 in Methods ) , the universal order parameter [40] was developed to quantify coherence in more general , weighted networks of oscillators . The universal order parameter accounts for the network topology and its influence on the phase coherence . Therefore , we can compare network coherence in topologically different weighted networks even when their total connections strengths are not the same . Furthermore , the universal order parameter captures partially phase-locked states accurately . To quantify different degrees of network coherence and to visualize localized and global synchrony , we measured the universal order parameter , both for the whole network of oscillators ( Eq 7 in Methods ) as well as for subnetworks of different spatial scales ( Eq 8 in Methods ) . To compute order parameters of subnetworks on the spatial scale d , we measured the averaged phase difference for each node i , with all the other nodes that are within the given spatial distance d from the node i . Thus computed averaged phase difference for each node is weighted by the inverse of input strengths to the given node i provided by its neighbors within the distance d from the node , and summed over all regions in the whole network . By thus computing the order parameter for the subnetworks , we describe the order parameter as a function of distance . Obtaining an explicit , analytical relationship between the order parameter and generalized network structures has been a challenging problem in studies of phase oscillators on complex networks [43 , 44] . While analytical expressions for the order parameter as a function of the adjacency matrix have been derived in previous works , these mean-field approaches are based on strong assumptions of a large network with sufficiently high average degree , valid only near the onset of synchronization [41 , 45–48] . Existing analytical approaches , therefore , are not applicable to the complex mesoscopic brain network of a finite size . We thus address the relationship between the network coherence and the network structure by computing the order parameter based on numerically obtained time series of the oscillators . Phases were initialized randomly , and Eq 1 was integrated numerically using the Forward Euler method , with a sufficiently small time step Δt = 10−4 ( s ) for 4 seconds ( Nt = 40000 steps ) , until a stationary state is reached . In our simulations , the time step size Δt = 10−4 ( s ) satisfies the condition Δ t ≤ 0 . 01 / max ( max ( k · A i j ) , 0 . 05 μ , σ n 2 2 T , 1 ) , as in [37] . The data from the first Nt/2 steps are discarded in measuring the order parameter . The order parameter , representing network coherence , can be modulated by the global coupling coefficient k , the standard deviation σd of the intrinsic frequency distribution , and the standard deviation σn in the additive Gaussian white noise . In this paper , we computed the order parameter using Eq 8 in Methods for varied global coupling coefficient k , with the standard deviation of the natural frequency distribution fixed at σd = 0 ( Hz ) and the standard deviation of the Gaussian noise fixed at σn = 2 ( rad ) . For each value of coupling coefficient k , we performed 10 independent runs , and plotted the average and the standard deviation of the order parameter as a function of distance between nodes ( Fig 3B ) as well as a function of global coupling coefficient k ( Fig 3C ) . We also show the order parameter as a function of the coupling coefficient k , with a nonzero standard deviation in the natural frequency distribution in the Supporting Information S2 ( C ) Fig . In this figure , the order parameter was averaged over 100 repeats with σd = 0 . 2 ( Hz ) and σn = 2 ( rad ) , to offset different effects of each configuataion of the intrinsic frequencies due to the nonzero σd . When the standard deviation of the white noise is held constant , increasing the coupling coefficient k with fixed σd has qualitatively the same effect as decreasing σd with k fixed , as the ratio of k/σd determines the network coherence . The same is true for decreasing the amount of σn . We show that varying σn and σd produces the same qualitative results as with varying k in the Supporting Information S2 Fig . When σn is varied , the intrinsic frequency distribution and the coupling coefficient are held constant , at σd = 0 and k = 3 , and the order parameter was averaged over 10 repeats . When σd is varied , the other two parameters are fixed at σn = 0 and k = 2 , and the order parameter was averaged over 100 repeats to account for the dependence of the time series on different configurations of the intrinsic frequencies . By computing the sensitivity of the network synchronizability on perturbation in each of these parameters—k , σn , or σd , we show that the observed trend in the data-driven brain network and the power-law approximated network is robust . In Fig 3A , we show the phase difference cos ( θi − θj ) for pairs of nodes ( i , j ) plotted against time and distance between the nodes . Interestingly , for the same amount of change in coupling coefficient Δk , the data-driven brain network switches between an asynchronous state and near-global synchrony , while the power-law governed network fails to make such a drastic change in synchronization state . This difference is manifested in the order parameter . Fig 3B shows the universal order parameter ( Eq 8 ) for subnetworks of different spatial ranges . When k is small , both the data-driven brain network and the power-law approximated network have overall low order parameters . In both cases , however , the order parameter is higher for small spatial scales , indicating that there is some spatially localized coherence in the networks due to the general trend of decreasing connection strength with distance between the connected regions . At a finer scale , we also observe a small amount of initial increase in the order parameter for the shortest-range connections ( 110-346 μm ) followed by a slow decrease in the order parameter as a function of distance in the data-driven brain connectome . Such an initial increase in the order parameter is not seen in the power-law estimated network . However , this initial rise at the very small length-scale should not be over-interpreted , because the experimental data are based on the mesoscopic measurements which are not accurate for distances less than 300-500 μm . In the viral tracing experimental data , the average distance to the closest injection is typically 500 μm at source level , which limits resolution [10 , 22] . In the data-driven brain network , increasing the coupling coefficient k results in a transition from partial coherence to near-global synchrony , manifested by increased order parameters across a range of spatial scales ( Fig 3B , left column , Data ) . However , in the artificially generated , strictly distance-dependent network , the same amount of change in global coupling coefficient does not induce such a leap in the network coherence state as in the real brain network ( Fig 3B , right column , Power law ) . Such trends can be also visualized in the order parameter for the whole network . The overall universal order parameter increases with global coupling coefficient in both the data-driven and the power-law networks ( Fig 3C ) . However , the change in order parameter is significantly larger in the data-driven brain network . This trend appears in both the single hemisphere network with only ipsilateral connections ( Supporting Information S1 Fig ) and the whole brain network with both ipsilateral and contralateral connections ( Fig 3C ) . For comparison , Kuramoto’s original order parameter ( Fig 3C , dotted ) is also plotted . Because the original Kuramoto’s order parameter does not account for different connection strengths among different pairs of nodes nor measure coherence scaled to the overall degree of the network , we see that the Kuramoto order parameter is lower than the universal order parameter for the power-law network . Nevertheless , for either type of the order parameter , we observe that the data-driven brain network spans a larger range of coherence states than the power-law governed network . These trends are more clearly portrayed by plotting the sensitivity of the order parameter ( Δr/Δk ) as the coupling coefficient k is varied ( Fig 3D ) . We observe that the sensitivity remains relatively constant throughout the range of the coupling coefficient in the power-law approximated network . However , the sensitivity of synchronizability is considerably more variable in the data-driven network , peaking around k = 2 . 5 . As the coupling coefficient increases , the sensitivity in synchronizability thus increases and then drops after reaching the maximum in the data-driven network , while the power-law approximated network is marked by relatively invariant , low sensitivity of the order parameter . This result on order parameter can be manifested by a couple of simple measures we use here . To compare the maximum sensitivity of the order parameter to changes in the global coupling coefficient k ( or any other parameters that modulate synchronizability , such as σn and σd ) , we introduce a measure of the maximum sensitivity of synchronizability: Γ k = max k , k + Δ k ( Δ r Δ k ) . ( 3 ) For the power-law network , the averaged maximum sensitivity of the order parameter is Γk = 0 . 1144 ± 0 . 0214 . The maximum sensitivity of the order parameter is higher in the data-driven mouse brain network , at Γk = 0 . 3172 ± 0 . 0829 . The higher value of the sensitivity measure Γk for the data-driven brain network indicates that a small amount of change in the coupling coefficient can induce a significant change in the network’s coherence state , in particular , within the range of k where Δr/Δk is maximum . To evaluate spatial dependence of the order parameter , we use another measure that quantifies the difference between the order parameter for short-range subnetworks and the order parameter for the whole network . This measure is defined as: Γd=〈 r ( d=dshortest ) −r ( d=dlongest ) 〉k , ( 4 ) where dshortest is the distance less than 570 ( μm ) that generates the highest order parameter value r ( d ) , and dlongest is 11955 ( μm ) which is the longest connection length in the mouse whole-brain network . 〈⋅〉k denotes averaging across varied coupling coefficient k , and r ( d ) is computed as defined in Eq 8 . For the data-driven brain network and the power-law-driven network , Γd = 0 . 1851 ± 0 . 0706 and Γd = 0 . 5383 ± 0 . 0234 , respectively . The larger Γd of the power-law network depicts a larger drop in coherence as the region of interest expands from the spatial vicinity to the whole network in the power-law network . In other words , the power-law network exhibits more localized coherence throughout a range of varied coupling coefficients . We also confirmed that such difference between the data-driven brain network and the strictly distance-dependent , power-law network remains unchanged when the natural frequencies of the nodes are moved to 8 Hz and 20 Hz , which are in the ranges of theta ( 6-12 Hz ) and beta ( 10-30 Hz ) oscillations , respectively . Like gamma oscillations , theta and beta oscillations are frequently observed in the large-scale brain dynamics . While gamma oscillations are thought to be linked to cognitive processing and sensing , theta rhythms are observed in hippocampal LFP and thus believed to underlie memory formation . On the other hand , beta rhythms have been associated with movement preparation and motor coordination [23 , 49 , 50] . As with the natural frequencies at 40 Hz , simulations with intrinsic frequencies at 20 Hz and 8 Hz also predict that the synchronizability is more sensitive to changes in global coupling coefficient in the data-driven brain network than in the power-law approximated network ( Supporting Information S3 Fig ) . With the realistic propagation speed 3 . 5 ( m/s ) and the longest connection at 11955 ( μm ) in the mouse whole-brain , the time delays in our model are quite small , and thus different intrinsic frequencies within the biologically realistic range induce qualitatively the same trend in synchronizability . It has been shown in previous studies that when the time delays multiplied by the intrinsic frequencies are sufficiently small compared to the coupling strengths in phase oscillators , the delay enters as a simple phase-lag [51 , 52] . Our results indicate that in the real brain network , a small change in the global coupling coefficient induces a rapid transition between partial network synchrony and a more globally synchronized state , while in the network with connections strictly following a power-law dependence on distance , such a rapid transition to synchronization is not observed . We get qualitatively the same results when we vary parameters other than the coupling coefficient k , namely , σd and σn , to modulate the network synchronizability . The order parameter is more sensitive to changes in the dispersion of intrinsic frequencies ( σd ) and the standard deviation in the additive white noise ( σn ) in the data-driven brain network than in the power-law governed network as well ( Supporting Information S2 Fig ) . Therefore , the residual connection strengths that are not explained by the simple spatial rule may have some computational significance , enabling even small perturbations in cognitive or behavioral states to induce a transition to synchronization . We next examined what aspects of the residual connection strengths confer the network’s ability to span a wide range of coherence states . In previous studies on coupled oscillators , it has been found that even a small fraction of shortcuts in a small-world network significantly improves synchronization of the network [43 , 53] . Motivated by this , we hypothesized that positive residual connections , namely , strong connections between distal brain regions , underlie the rapid transition in network synchronies . We tested this hypothesis by re-introducing small fractions of the positive residuals to the power-law distance-dependent network . As manifested in Fig 4A , adding just a small fraction ( top 20 percentile ) of the strongest positive residuals to the power-law generated network recovers the steep increase in order parameter with growing coupling coefficient ( Fig 4 , purple ) . As the fraction of positive residuals included in addition to the power-law network increases , the sensitivity of the order parameter as a function of the coupling coefficient resembles more of that of the real brain network ( Fig 4B ) . This trend is also depicted by the maximum sensitivity measure which is at Γk = 0 . 1423 ± 0 . 0036 , Γk = 0 . 1617 ± 0 . 1266 , and Γk = 0 . 2785 ± 0 . 0506 , respectively for top 5 , 20 , 40% of the positive residuals added to the power-law network , on the same edges as in the original data-driven whole-brain network . Does the location of these strong connections have any significance in emergence of the rapid phase transition ? To test whether the sensitivity of the network coherence to coupling coefficient can be recovered by simply adding the positive residuals anywhere to increase the overall connection strength of the power-law network , we studied the dynamics of the network constructed by relocating the positive residuals . We generated three networks with positive residuals relocated . In one of them , the positive residuals above the power-law relationship were positioned at random locations on the network ( shuffled ) . In the other two , the positive residuals were preferentially relocated to the shortest 0 . 2% or to the longest 0 . 2% connections of the total edges . For the proximal-relocated network , the positive residual connections were added to connections between spatially close regions , by distributing the total positive residual connection strength among the connections between nodes within 570μm . For the distal-relocated network , the positive residuals were added to the connections between spatially distal regions , by distributing the total positive residual connection strength among the edges longer than 10500μm . The resulting networks thus maintain the total connection strengths of the real brain network , but have altered network structures . When the locations of the positive residuals are randomized and thus there are strong connection weights across multiple spatial scales , the dependence of network synchronization on k remains similar to that of the data-driven network , as portrayed by the order parameter in Fig 5A , in gray and Fig 5B , left . In other words , although the precise network structure is different from that of the data-driven network , the network with shuffled residuals maintains its sensitivity to the global coupling coefficient , rapidly changing network coherence states . However , the spatial structure of the order parameter is dependent on the precise locations of these positive residuals . In the network with positive residuals randomly relocated , there is a steeper decrease in order parameter with distance ( Fig 5B , left ) , compared to the data-driven brain network ( Fig 3B , left ) . This trend is depicted by the higher spatial coherence measure , Γd = 0 . 4091 ± 0 . 00017 for the network with randomized positive residuals , compared to the data-driven brain network ( Γd = 0 . 1851 ± 0 . 0706 ) . When the positive residuals are relocated to proximal connections , the network coherence is no longer as sensitive to small changes in the global coupling coefficient as in the whole-brain network ( Fig 5A , dotted black; B , middle ) , in spite of the unaltered total connection strengths . This trend is robustly maintained when the standard deviation in the natural frequency distribution is varied instead of the global coupling coefficient ( Supporting Information S4 Fig ) . Similarly , when the positive residuals are moved to distal connections , the network coherence loses sensitivity as well ( Fig 5A , solid black; B , right ) . Unlike the network with randomly relocated residuals , the networks with positive residuals relocated only to proximal or distal connections lack strong connections distributed across a range of spatial scales . Thus , connections that are stronger than predicted by the distance-dependence should be spread over varied lengths of edges , for the network to switch between localized and global coherence states with a small change in the global coupling coefficient . In addition , we observe that when positive residuals are relocated to proximal connections , the overall order parameters across the spatial scales are higher ( Fig 5B , middle ) , compared to the network constructed by placing positive residuals to distal connections ( Fig 5B , right ) . This effect arises from the definition of the universal order parameter ( Eqs 7 and 8 ) , where each of the time-averaged phase difference 〈cos ( θi − θj ) 〉t is weighted by the connection strength between the pair of oscillators Aij . When positive residuals are placed on proximal connections , the influence of the phase differences between nearby nodes , which increases the overall network order parameter , is emphasized more by larger connection strengths Aij . On the other hand , when the positive residuals are relocated to distal connections , although distal nodes are now more strongly coupled than before , the phase differences between distal nodes are still quite large . Therefore , in this case , the large phase differences between distal nodes which lower the overall order parameter , are strongly weighted by Aij , and thus , the overall network order parameters are maintained at low values . We also note that the order parameter rapidly increases at large distances in the power-law network with the residuals preferentially added to the longest edges ( Fig 5B , right ) . This rapid increase stems from the relatively high values of the connections strengths of these longest edges ( Aij ) which induce large values of 〈cos ( θi − θj ) 〉t between distal regions i and j . Therefore , the order parameters at the large spatial scales are increased by large values of Aij〈cos ( θi − θj ) 〉t terms . To further examine the relationship between the spatial spread of the strong connections and the sensitivity of synchronizability , we measured the order parameter in networks generated from the power-law approximation by placing the positive residual strengths to different fractions of the longest edges . In Fig 5C , we show the order parameter as a function of the coupling coefficient k when positive residuals are preferentially added back on edges that have lengths greater than various cutoff values . As the spatial scale over which the residuals are added widens , the sensitivity of the order parameter gradually increases . Notably , the sensitivity and the growth of the order parameter become comparable to those of the data-driven brain network when the percentile of the longest edges with added positive residuals reaches 5 − 10% of the total connections . This indicates that while it is important to have a spread of strong connections above the power-law prediction over multiple spatial scales , the spread does not have to extend all the way to the shortest edges of the network in order to generate high sensitivity of the order parameter observed in the data-driven brain network . Our results show that the location of strong connections above the power-law dependence on distance is critical for generating a steep change in the order parameter . While the precise positions of the strong connections do not have to match those of the data-driven network to produce highly sensitive order parameter to the coupling coefficient , there should be a sufficient amount of strong connections across a range of spatial scales . Precise locations of the strong residuals , however , determine the order parameter’s dependence on the spatial scale , modulating spatial coherence patterns . In sum , the spatial structure of the network connectivity plays a key role in maintaining the brain’s ability to change its computational states with small perturbations , and such sensitivity cannot be achieved by simply matching the total network connection strengths . The structure does not have to precisely match that of the real brain network to maintain the high sensitivity . What is critical to maintain , rather , is some connections stronger than the simple distance-dependence distributed over the network . However , the precise connectivity structure is important for generating specific spatial coherence patterns in the network dynamics .
In this paper , we studied synchronization of a spatially constrained model of a weighted whole-brain network at the mesoscale , constructed from viral tracing experiments . The importance of linking connectivity structure and large-scale brain dynamics have been noted in previous studies [54–56] . In particular , the heterogeneity in structural connectivity has been proposed as a key underlying mechanism for certain brain network dynamic features such as functional hubs in resting state dynamics [56] . However , additional complexities in the anatomically precise , weighted and directed whole-brain network that are not captured by spatially-defined connectivity have been often overlooked . In this work , we propose possible computational roles of these additional complexities by studying their effects on network synchronizability . We found that the connectivity has a significant spatial dependence , with the connection strength decreasing with distance between the regions following a power law . However , by studying the network dynamics of phase oscillators , we found that a network generated by the simple spatial constraints alone cannot reproduce the full computational versatility of the mesoscopic whole-brain network . Rather , we need to consider additional complexities of the network structure to capture their possibly significant roles in neural computation . Specifically , we found that residual connections not explained by the power-law dependence on distance have a long positive tail , corresponding to strong connections between distal brain regions . By computing the recently proposed universal order parameter , we showed that these strong distal connections underlie sensitive dependence of network synchrony on perturbations in coupling coefficient ( or intrinsic frequency distribution/noise ) , potentially responsible for the brain’s exceptional ability to change its computational states depending on stimulus and behavioral context . Furthermore , our analyses on networks constructed by adding a small fraction of strong positive residuals to the spatially-constrained connectivity , as well as networks with the positive residuals relocated to random , proximal , or distal connections , reveal the key element underlying the rapid switch between global and partial synchronies—strong connections distributed over varied spatial distances . In other words , the network’s sensitivity to perturbation cannot be reproduced by simply manipulating the overall connection strengths , as locations of positive residual connections should be taken into consideration . A spatially-constrained model plus an idiosyncratic sparse matrix which features strong connections between distal regions provides a parsimonious representation of the measured connectivity . We hypothesize that the sharp transition in synchronization in the data-driven network , which is absent in the spatially-constrained power-law model , may underlie the brain’s ability to rapidly switch computational states [57] . Such a feature is known to be impaired in the brain under pathological conditions such as Alzheimer’s disease , suggested by studies showing more modular structures and decreased global efficiency in brain connectivity constructed from EEG , MEG , fMRI , and diffusion tensor tractography [58–61] . Moreover , there is an experimental evidence for disruption of long-range connections in Alzheimer brain network [60] , in agreement with our model results . Therefore , the strictly distance-dependent power-law network which maintains localized synchronization across a range of coupling coefficients may explain aberrant network dynamics and computational impairments in Alzheimer brains . A more detailed future study on genetically-controlled mouse models of Alzheimer’s disease will shed light on the possible link between changes in structural connectivity and impairment in rapid phase transitions of the whole-brain network . The increased sensitivity of the network synchronizability induced by strong long-range connections further implicates a tradeoff between cost-efficiency and high functional capacity in the brain network . Such tradeoff between wiring cost and computational capacity has been suggested as a network-generating principle in a number of previous studies [18 , 62–67] . The power-law dependence of connection strengths on inter-regional distance reflects spatial and energetic constraints in the brain network . Indeed , if the brain connectivity is designed to exclusively optimize the wiring cost , we will observe strong connections only between proximal regions . Yet , we observe some idiosyncratic , strong long-range connections which are expensive in the mouse brain connectome . By showing that these strong distal connections may serve to promote rapid transitions between network synchronization states and possibly , computational states , our work points to a possible functional role afforded by the presence of the long-range connections despite their high metabolic costs . In this paper , we infer the dynamics of the mesoscopic brain network by constructing a network of phase oscillators with the coupling strengths determined by the structural connectivity obtained by viral tracing experiments . Thus , while the structural connectivity is based on actual data , the dynamics we conferred on the network are arbitrary . Building a more realistic , data-driven dynamic network based on imaging experiments such as calcium-imaging , ECoG , LFP , and MEG will be a crucial future extension of our study of connecting the network structures to the network dynamics . Furthermore , for future studies , more biophysically-motivated neural mass models [68] would be necessary to capture realistic dynamics of the brain network that are not predicted by simple phase oscillator models . However , our simulations with phase oscillators , despite their generality , still make valuable predictions on computational roles of spatial structures of the mesoscopic whole-brain network , underlining the importance of spatially distributed , strong distal connections on the network dynamics .
The mesoscopic mouse whole-brain connectivity was obtained from the Allen Mouse Brain Connectivity Atlas ( http://connectivity . brain-map . org/ ) , constructed based on anterograde viral tracing experiments in wild type C7BL/6 mice [10] . Based on the experimental data , a recently developed interpolative mapping algorithm was used to construct a model of whole brain connectivity at the 100 μm-voxel scale [22] . The voxel-based connection strengths were averaged over each brain region to produce a connectivity matrix with 244 brain regions per hemisphere as nodes , larger than the adjacency matrix of 213 pairs of nodes previously obtained from the linear model in [10] . For elements of the connectivity matrix , we use the normalized projection density , defined as the connection strength between two regions divided by the volume of the source and target regions . In order to account for the size of the source region , we also studied the relationship between the connection strength divided only by the size of the target region and the distance between two regions . In this case , however , the fit to either a power law or an exponential function was not very good which is not surprising given that the connection strengths that are not fully normalized with respect to the size of the source and the target is not an intrinsic quantity . For more details on the viral tracing experiments and the interpolative algorithm used to construct the connectivity matrix , see [10] and [22] . The connectivity matrix was first normalized to have values between 0 and 1 . For the ipsilateral connection matrix , the diagonal entries were set to zeros removing self-connectivity , as done in [4 , 20] . We fitted connection strengths as a function of interregional distance , where the distance between each pair of nodes was determined by computing the Euclidean distance in 3-dimensional coordinates between the centroids of the brain regions . Specifically , power-law functions for relationships between connection strength and interregional distance were fitted by using least squares on the log scale . For each of the ipsilateral and contralateral connectivity matrices , we found α and β by fitting the data to A ˜ i j = α · d i j - β + ∊ i j , where A ˜ i j denotes the connection strength from node j to node i , dij indicates the distance between nodes i and j , and ϵij is the residual error . We obtained α = 6 . 92 × 106 and β = 2 . 886 for ipsilateral connectivity , and α = 6 . 71 × 105 and β = 2 . 685 for contralateral connectivity ( Fig 1B ) . In agreement with previous studies on Allen Mouse Brain Connectivity data [18 , 22] , we found that the power law explains the relationship slightly better than the exponential dependence ( ipsilateral r-square: 0 . 264 vs 0 . 257 , rmse: 1 . 089 vs 1 . 095; contralateral r-square: 0 . 167 vs 0 . 135 , rmse: 1 . 124 vs 1 . 146 ) . We also investigated the power-law constrained network where the relationship between connection strength and interregional distance was found on the real scale , using nonlinear least squares ( Levenberg-Marquardt algorithm ) , which has a poorer explanatory power than linear least squares on the log-scale ( r-square: 0 . 264 vs 0 . 157 ( ipsilateral ) / 0 . 167 vs 0 . 131 ( contralateral ) ) . While this method generated a different power-law function from the one found by least squares on the log-log scale , the dynamics on the power-law network obtained by using nonlinear least squares maintained the same core characteristics , distinct from the data-driven brain network– the order parameter is less sensitive to changes in the global coupling coefficient . In this section , we describe order parameters that were proposed previously [41 , 42 , 45 , 46] , demonstrating advantages of the recently developed universal order parameter [40] in our analysis . In order to quantify network coherence in the original model of phase oscillators with all-to-all connectivity , Kuramoto introduced the complex order parameter [41 , 42] , r ( t ) e i ψ ( t ) ≡ 1 N ∑ i = 1 N e i θ i , ( 5 ) where ψ ( t ) gives the average phase of all oscillators and r ( t ) describes the degree of phase coherence at time t . The overall phase coherence is measured by the absolute value of the complex order parameter averaged over time . We denote this value rKuramoto , as the measure of the averaged phase differences of all pairs of oscillators: r Kuramoto 2 ≡⟨ | r ( t ) e i ψ ( t ) | 2⟩ t = ⟨ 1 N 2 ∑ i , j = 1 N e i ( θ i - θ j ) ⟩ t = 1 N 2 ∑ i , j = 1 N ⟨ cos ( θ i - θ j ) ⟩ t . ( 6 ) < … >t denotes the average over time . However , this unweighted order parameter is not a good measure when comparing collective synchronizations in two networks described by different connectivity matrices , as it does not capture the topology of the networks . To extend the use of order parameter to more general , weighted networks of oscillators , Restrepo et al [45 , 46] proposed an order parameter which is defined as the average of local order parameters which measure the coherence of the inputs to each node . This parameter , however , does not capture partially phase-locked states well . Recently , Schroeder et al [40] proposed a new , “universal order parameter” to accurately measure phase coherence in weighted and directed networks of arbitrary topology , which overcomes the shortcomings of the previous order parameters . This newly proposed universal order parameter is defined as: r ≡ 1 ∑ i = 1 N k i ∑ i , j = 1 N A i j ⟨ R e ( e i ( θ i - θ j ) ) ⟩ t = 1 ∑ i = 1 N k i ∑ i , j = 1 N A i j⟨ cos ( θ i - θ j ) ⟩ t ( 7 ) where k i = ∑ j = 1 N A i j is the input strength of node i . Note that in unweighted binary networks , this measure represents in-degree [4] . This order parameter accounts for the network topology and its influence on the phase coherence , enabling a fair comparison between two topologically different weighted networks even when their total connection strengths are not matched . As this universal order parameter accurately captures partial synchrony within the network , different degrees of synchronization can be measured by order parameter of the whole network . Furthermore , degree of coherence as a function of spatial extent can be obtained by computing the order parameter for subnetworks of different spatial scales . The order parameter r can be described as a function of distance d: r ( d ) ≡ 1 ∑ i = 1 N k i ∑ i = 1 N ∑ j ∈ γ ( i , d ) A i j ⟨ R e ( e i ( θ i - θ j ) ) ⟩ t = 1 ∑ i = 1 N k i ∑ i = 1 N ∑ j ∈ γ ( i , d ) A i j ⟨ cos ( θ i - θ j ) ⟩ t ( 8 ) where γ ( i , d ) indicates the set of nodes within spatial distance d from node i . ki = ∑j∈γ ( i , d ) Aij is the total connection strength of node i when the subnetwork composed of nodes within distance d from node i is considered . The order parameter of the whole network is obtained when d = size of the network ( 11752μm for ipsilateral and 11955μm for contralateral connectivity ) . All of the MATLAB code used to numerically compute time-series data of coupled oscillators and the order parameters on the mouse whole-brain network from [10 , 22] and the power-law approximated network are available at https://github . com/AllenInstitute/Choi2019_ConnectomeSynchrony . | In a previous study , a data-driven large-scale model of mouse brain connectivity was constructed . This mouse brain connectivity model is estimated by a simplified model which only takes in account anatomy and distance dependence of connection strength which is best fit by a power law . The distance dependence model captures the connection strengths of the mouse whole-brain network well . But can it capture the dynamics ? In this study , we show that a small number of connections which are missed by the simple spatial model lead to significant differences in dynamics . The presence of a small number of strong connections over longer distances increases sensitivity of synchronization to perturbations . Unlike the data-driven network , the network without these long-range connections , as well as the network in which these long range connections are shuffled , lose global synchronization while maintaining localized synchrony , underlining the significance of the exact topology of these distal connections in the data-driven brain network . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"built",
"structures",
"neural",
"networks",
"engineering",
"and",
"technology",
"signal",
"processing",
"condensed",
"matter",
"physics",
"neuroscience",
"artificial",
"neural",
"networks",
"artificial",
"intelligence",
"white",
"noise",
"brain",
"mapping",
"network",
... | 2019 | Synchronization dependent on spatial structures of a mesoscopic whole-brain network |
The growth of plant organs is a complex process powered by osmosis that attracts water inside the cells; this influx induces simultaneously an elastic extension of the walls and pressure in the cells , called turgor pressure; above a threshold , the walls yield and the cells grow . Based on Lockhart’s seminal work , various models of plant morphogenesis have been proposed , either for single cells , or focusing on the wall mechanical properties . However , the synergistic coupling of fluxes and wall mechanics has not yet been fully addressed in a multicellular model . This work lays the foundations of such a model , by simplifying as much as possible each process and putting emphasis on the coupling itself . Its emergent properties are rich and can help to understand plant morphogenesis . In particular , we show that the model can display a new type of lateral inhibitory mechanism that amplifies growth heterogeneities due e . g to cell wall loosening .
Plants grow throughout their lifetime at the level of small regions containing undifferentiated cells , the meristems , located at the extremities of their axes . Growth is powered by osmosis that tends to attract water inside the cells . The corresponding increase in volume leads to simultaneous tension in the walls and hydrostatic pressure ( so-called turgor pressure ) in the cells . Continuous growth occurs thanks to the yielding of the walls to these stretching forces [1–3] . This interplay between growth , water fluxes , wall stress and turgor was first modelled by Lockhart in 1965 [4] , in the context of a single elongating cell . Recent models focused on how genes regulate growth at more integrated levels [5–9] . To accompany genetic , molecular , and biophysical analyses of growing tissues , various extensions of Lockhart’s model to multicellular tissues have been developed . The resulting models are intrinsically complex as they represent collections from tens to thousands of cells in 2- or 3-dimensions interacting with each other . To cut down the complexity , several approaches abstract organ multicellular structures as polygonal networks of 1D visco-elastic springs either in 2D [7 , 10–12] or in 3D [6 , 13] submitted to a steady turgor pressure . Other approaches try to represent more realistically the structure of the plant walls by 2D deformable wall elements able to respond locally to turgor pressure by anisotropic growth [8 , 14 , 15] . Most of these approaches consider turgor as a constant driving force for growth , explicitely or implicitly assuming that fluxes occur much faster than wall synthesis . Cells then regulate the tissue deformations by locally modulating the material structure of their walls ( stiffness and anisotropy ) [6 , 16–20] . However , the situation in real plants is more complex: turgor heterogeneity has been observed at cellular level [21 , 22] , which challenges the assumption of very fast fluxes . As a matter of fact , the relative importance of fluxes or wall mechanics as limiting factors to growth has fuelled a long standing debate [3 , 23] and is still an open question . Moreover , from a physical point of view , pressure is a dynamic quantity that permanently adjusts to both mechanical and hydraulic constraints , which implies that a consistent representation of turgor requires to model both wall mechanics and hydraulic fluxes . The aim of this article is to explore the potential effect of coupling mechanical and hydraulic processes on the properties of the “living material” that corresponds to multicellular populations of plant cells . To this end , we build a model that describes in a simple manner wall mechanics and cell structure , but do not compromise on the inherent complexity of considering a collection of deformable object hydraulically and mechanically connected . The article is organized as follows ( see Fig 1 ) : we first recall the Lockhart-Ortega model and its main properties . Then we explore two simple extensions of this model: first we relax the constraint of uniaxial growth in the case of a single polygonal cell; then we study how two cells hydraulically connected interact with each other . Finally we describe our multicellular and multidimensional model and numerically explore its properties . A table of notations is provided in Supplementary Information ( S1 Table ) .
In 1965 , Lockhart [4] derived the elongation of a plant cell by coupling osmosis-based fluxes and visco-plastic wall mechanics . Ortega [24] extended this seminal model to include the elastics properties of the cell walls . We recall here the main properties of this model , see Fig 1a for the geometrical configuration . A multicellular extension of the Lockhart-Ortega model adapted to the study of morphogenesis requires first to relax the constraint of uniaxial growth and allow multidimensional geometries , and second is complexified by the possibility of fluxes between cells . We study separately the effect of each of these extensions before presenting the complete model . In the Lockhart-Ortega model , the compatibility between wall enlargement and cell volume variation is automatically enforced through the geometrical constraint of uni-directional growth that leads to the identity between the relative growth rate of the cell and the strain rate of the walls . In contrast , in the multicellular model , this identity is no longer true . One has to solve the closed set of Eqs ( 7 ) – ( 11 ) and ( 12 ) with respect to the unknowns X , P , and εe . Despite its apparent simplicity , the problem to be solved is not straightforward as water fluxes induce potentially long range interactions . In this respect , it differs from most vertex-based models ( e . g [11 , 26] ) where turgor is an input of the model . The numerical resolution required the development of an original algorithm ( see S5 Text ) implemented in an in-house code . The properties of this model cannot be as thoroughly studied as those of the simpler models presented above , first because of the numerical cost of the resolution , but above all because it allows an infinite variety of geometries and spatial distribution of its parameters . We present here a numerical experiment that illustrates on the one hand how the properties of the simple multidimensional and multicellular submodels are combined in the generalized model; in turn the study of these models helps us to anticipate the properties of the generalized model . And on the other hand , we show that this model is readily applicable to the study of systems of biological interest . Growth heterogeneities can be triggered by the local modulation of the mechanical properties of the cell walls [27] . In SAMs , new organs are initiated by a local increase in growth rate that leads to the appearance of small bumps . Measurements show that physico-chemical properties of walls are modified so that mechanical anisotropy and elastic modulus are decreased . Our 2D model is used to represent a cross section of a SAM and we explore what effect a local softening of the walls has on growth rate and turgor heterogeneities; based on our previous analysis of the model in simple configurations , we expect that the growth heterogeneities will be maximal for parameters such that the growth is restricted by fluxes rather than wall synthesis ( αa close to 0 ) , cell-cell conductivity is large ( αs close to 1 ) , and the walls deformations are just above the growth threshold , which can be enforced by a low value of the osmotic pressure ( yet large enough to ensure growth ) . The set of parameters ( REF ) is chosen according to these criteria; then we explore the effect of a higher αa ( ( ALPHA+ ) set ) and lower cell-cell conductivity ( ( CC- ) set ) that should both decrease the growth heterogeneities , and also test the effect of a lower osmotic pressure ( ( PM- ) set ) that should conversely increase the growth heterogeneity . See S6 Text for detailed explanations on the values of the parameters corresponding to these sets . We build a mesh made primarily of hexagons ( see Fig 3a ) and first let it grow with homogeneous parameters until the elastic regime ends and plastic growth occurs . Then we divide by two the elastic modulus of a small group of cells ( marked with a white star in Fig 3a ) that will be referred to as “bump cells” thereafter . First , Fig 3b shows that the multicellular system grows globally in the same way as the single hexagonal cell studied above; it diverges from the Lockhart predictions because the ratio A/V of the cells is not constant: the ( ALPHA+ ) simulations exhibit a very large initial growth rate that decreases only when the cells are so large that water fluxes become limiting . The ( PM- ) set leads to a roughly twice lower growth rate than ( REF ) . The set ( CC- ) leads to the same dynamics at the tissue level as ( REF ) , because the total influx of water is not affected by fluxes between cells in this setup . Then we turn to the observation of heterogeneities: we focus on the differences between the bump region and the rest of the tissue . For all the parameters sets , Fig 3c shows that turgor is in general lower in bump cells , but the gap varies depending on the parameters , as it has been predicted by the study of the two-cells model: compared to ( REF ) , the heterogeneity in turgor is increased by a lower cell-cell conductivity ( set CC- ) , and decreased by a larger value of αa ( set ALPHA+ ) . Decreasing the value of PM ( set PM- ) does not alter much the turgor heterogeneity compared to ( REF ) . The maps of turgor ( Fig 3e , 3g , 3i and 3k ) confirm visually these observations . Fig 3d shows the time evolution of γ ˙ / γ ˙ * where γ ˙ * is the relative growth rate predicted by the Lockhart model ( see ( 6 ) ) ; its value is 2% h−1 for ( REF ) , ( CC- ) and ( ALPHA+ ) , and 0 . 5% h−1 for ( PM- ) . In the considered time frame , the relative growth rate of bump cells is always higher except for ( ALPHA+ ) : after an initial fast increase where bump cells grow faster , the tendency is inversed at t ≈ 20h because the bump cells have grown so much that fluxes become limiting . In the ( REF ) simulation , while the growth rate of non bump cells is almost constant and close to γ ˙ * , the growth rate of the bump cells is up to 6 times γ ˙ * at the beginning of the simulation and progressively decreases toward γ ˙ * . As a result of this large discrepancy , the bump region can be clearly distinguished from the rest of the tissue ( Fig 3e and 3f ) . In ( CC- ) , the growth rate of the non bump cells is close to that of ( REF ) , but the growth rate of the bump cells is much lower ( Fig 3d ) . As a result , the global shape remains convex and the bump is not clearly detached from the rest of the tissue ( Fig 3i and 3j ) . Note that ( CC- ) corresponds to a lower value of αs compared to ( REF ) , which corresponded to a lower growth heterogeneity with the two-cells model studied above; this is also confirmed by the lower cell-cell fluxes towards the bump cells for ( CC- ) , see the arrows in Fig 3e and 3i . The ( ALPHA+ ) simulation exhibits also a convex shape ( Fig 3k and 3l ) ; it corresponds to a larger value of αa than ( REF ) , and similarly to the two-cells model studied above , the growth rate heterogeneity is lower than ( REF ) . Finally , the set ( PM- ) corresponds to an increase of the dimensionless parameter ρ ( see ( 10 ) ) , and accordingly to an increase in growth rate heterogeneity as can be seen with Fig 3d . Consequently , the bump region can be clearly distinguished from the rest of the tissue , even better than ( REF ) ( Fig 3g and 3h ) ; moreover , the growth of the cells close to the bump seems to be inhibited by fluxes as explained in the two-cells model described above and further explored below .
The model proposed in this article is a minimal multicellular and multidimensional extension of the Lockhart 1-D single cell model; it can be regarded as a conceptual tool to study the interplay between fluxes and wall mechanics in a multicellular tissue . Wall expansion is modeled with a visco-elasto-plastic rheological law , while fluxes derive from water potential gradients . These two contributions are integrated into the mechanical equilibrium and interact through the pressure term . Contrary to most previous approaches , turgor is not an input of the model but a variable that adjusts simultaneously to mechanical , hydraulic , and geometrical constraints . First of all , this leads to a physically consistent representation of turgor: for instance , the model predicts that cells with softer walls have a lower turgor . Moreover , this has deep implications at tissue level: in the previous example , lower turgor is associated with a faster growth which can be itself amplified by fluxes that follow decreasing pressure gradients . Thanks to the simplicity of the model , the predicted behavior can be analyzed and interpreted with two submodels built from the Lockhart model: first , in a 1-D system , cells are only elongating and their surface-to-volume ratio is constant . We thus extended the Lockhart model in two dimensions , where cells have more degree of freedom to change their shape . In particular their allometric surface-to-volume ratio may then vary . This new possibility induces additional complexity in the tissue development as the rate of growth of cell surfaces may become a limiting factor for growing cells . Second , a 1-D multicellular submodel was build with two or more side-by-side cells; it was used to study the growth of competing cells with heterogeneous properties . Key ingredients here are the wall synthesis threshold , the fact that fluxes and growth can relax turgor , and cell to cell fluxes that allow long range interactions . Depending on mechanical and hydraulic parameters of tissue regions , the model exhibits different growth regimes corresponding to either uniform or differential growth . One unexpected consequence of such an hydraulic-mechanical coupling at the tissue level is the observation that in certain regions of the parameters space where cell-to-cell hydraulic exchanges are non-limiting , growing tissue may exert an inhibiting influence on the growth of neighboring regions . This may be interpreted as a lateral inhibition mechanism . It has for long been recognized that lateral inhibitory mechanisms play a key role in setting some morphogenetic patterns in procaryotes ( e . g . [28] ) , animals ( e . g . [29 , 30] ) or plants ( e . g . [31 , 32] ) . Lateral inhibition operates in these systems via chemical signals , such as delta-notch in animals or auxin in plants . Our model predicts the existence of a novel type of lateral inhibition mechanism based on the coupling between mechanics and water fluxes . Previous observations of tissue growth suggest that such a phenomenon may occur in real tissues . In the shoot apical meristem for instance , detailed quantification of growth with cellular resolution indicates that the region surrounding primordia growth may have a negative growth rate ( [33] , Figs 2G and 3K ) . According to our model , this decrease of volume in boundary regions might be due to the primordium growth attracting locally most of the water supply and depriving lateral regions from water , and thus conforts the hypothesis of a new hydraulic-mechanical component of primordium lateral inhibition , beyond already identified auxin and cytokinin signals [34] . Throughout the development of the model , we made several key choices concerning the abstraction of a multicellular plant tissue . First , our model was developed in 2-D for reasons of computational efficiency . In principle , it can be extended in 3-D , though at the expense of more complex formalism and implementation . Second , the current model considers that water transport is performed in the plant tissue through two conceptually different pathways [1] . Water can first move within the apoplasmic compartment between the cells and finally enter a cell . Water can also move locally from cell to cell . This movement includes itself conceptually both symplasmic movements ( water circulates between cells through plasmodesmata without crossing membranes ) and movements from cell to cell with intermediate steps in the wall ( water is for example exported locally out of the cell by water transporters like aquaporins into the wall and immediately re-imported by water transporters into neighboring cells ) . For the sake of simplicity in this first analysis , we represented the apoplasm as a single abstract compartment able to exchange water with every cell . To analyze precisely the effect of water transporters and their genetic regulation or to assess the impact of wall resistance to water movement in the processes , explicit spatial representation of the apoplasm , of plasmodesmata and of membrane water transporters could be integrated into the model in the future . Finally , we considered a simplified situation here by imposing constant cell osmolarity . Allowing osmolarity variations ( for instance higher values in faster growing regions ) may impact turgor distribution ( e . g [35] ) . However , this should not affect the ability of the system to build up growth heterogeneities . Similarly , we further simplified our model by keeping constant the apoplasmic water potential . Relaxing this hypothesis would increase cell-cell water fluxes ( via the apoplasm ) and could also shift the model in the direction of the flux-limiting regime . This would therefore favor regimes where growth heterogeneities are amplified by fluxes . This model may impact our understanding of various biological questions at the interface between mechanics and hydraulics in plants thanks to its emergent properties that are far more complex and rich than the Lockhart model it is based on . We showed here the impact of fluxes on turgor and growth rate heterogeneities at tissue level and how they can impact morphogenesis . In a recent study [22] , we focused on heterogeneities at cell level and compared the model to experimental measurements; in particular , we correctly predicted that the number of cell neighbors is negatively correlated with cell turgor . Finally , based on its ability to provide quantitative insights in growing multicellular systems , this model could contribute to revisit the long-standing debate initiated by Boyer and Cosgrove regarding the relative importance of fluxes and wall softening in the limitation of growth in plants . | Plant cells are surrounded by a rigid wall that prevents cell displacements and rearrangements as in animal tissues . Therefore , plant morphogenesis relies only on cell divisions , shape changes , and local modulation of growth rate . It has long been recognized that cell growth relies on the competition between osmosis that tends to attract water into the cells and wall mechanics that resists to it , but this interplay has never been fully explored in a multicellular model . The goal of this work is to analyze the theoretical consequences of this coupling . We show that the emergent behavior is rich and complex: among other findings , pressure and growth rate heterogeneities are predicted without any ad-hoc assumption; furthermore the model can display a new type of lateral inhibition based on fluxes that could complement and strengthen the efficiency of already known mechanisms such as cell wall loosening . | [
"Abstract",
"Introduction",
"Models",
"and",
"results",
"Discussion"
] | [
"cell",
"walls",
"plant",
"anatomy",
"stem",
"anatomy",
"classical",
"mechanics",
"plant",
"cell",
"biology",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"plant",
"physiology",
"osmotic",
"pressure",
"shoot",
"apical",
"meristem",
"plant",
"ce... | 2019 | Coupling water fluxes with cell wall mechanics in a multicellular model of plant development |
Expansions of trinucleotide GAA•TTC tracts are associated with the human disease Friedreich's ataxia , and long GAA•TTC tracts elevate genome instability in yeast . We show that tracts of ( GAA ) 230• ( TTC ) 230 stimulate mitotic crossovers in yeast about 10 , 000-fold relative to a “normal” DNA sequence; ( GAA ) n• ( TTC ) n tracts , however , do not significantly elevate meiotic recombination . Most of the mitotic crossovers are associated with a region of non-reciprocal transfer of information ( gene conversion ) . The major class of recombination events stimulated by ( GAA ) n• ( TTC ) n tracts is a tract-associated double-strand break ( DSB ) that occurs in unreplicated chromosomes , likely in G1 of the cell cycle . These findings indicate that ( GAA ) n• ( TTC ) n tracts can be a potent source of loss of heterozygosity in yeast .
Several inherited human diseases are a consequence of the expansion of trinucleotide tracts [1] , [2] . Although the mechanism by which tract expansions are generated is not yet understood , most of the trinucleotide tracts prone to expansion can form secondary structures such as “hairpin-like” DNA ( intrastrand pairing ) or triplexes ( intramolecular pairing events involving complexes with three paired strands ) . Friedreich's ataxia is caused by expansion of tracts of the trinucleotide GAA•TTC , a sequence that is associated with triplex formation [3] . In the yeast Saccharomyces cerevisiae , ( GAA ) n• ( TTC ) n tracts greater than 40 repeats in length result in an orientation-dependent stall of the replication fork [4] , [5] . The stall of the replication fork is observed when the ( GAA ) n sequence is located on the lagging strand template . Long ( GAA ) n• ( TTC ) n tracts have high frequencies of contractions and expansions ( primarily contractions ) in both orientations , although these alterations are somewhat more frequent when the ( GAA ) n sequences are on the lagging strand template; in our subsequent discussion , we will refer to tracts in this orientation as ( GAA ) n tracts and the same sequence in the opposite orientation as ( TTC ) n tracts . The poly ( GAA ) tracts are also associated with a high rate of double-stranded DNA breaks ( DSBs ) and a high rate of terminal chromosome deletions [5] . In addition , ( GAA ) 230 tracts stimulate ectopic recombination between lys2 heteroalles 200-fold more than ( TTC ) 230 tracts [5] . In contrast to the strong orientation-dependence observed in studies of replication fork stalling , DSB formation , and ectopic recombination , the frequency of large-scale expansions of the long ( GAA ) n• ( TTC ) n tracts is affected only slightly by tract orientation [6] . In addition to studies done in yeast , the properties of ( GAA ) n• ( TTC ) n repeats were also examined in bacterial and mammalian systems . In E . coli , ( GAA ) n• ( TTC ) n tracts stimulate plasmid-plasmid recombination by a mechanism that is dependent on both the orientation and length of the repetitive tract [7] . In mammalian cells , length-dependent expansions of ( GAA ) n• ( TTC ) n and ( CTG ) n• ( CAG ) n tracts are observed; these expansions are stimulated by transcription , and are observed in non-dividing cells , indicating that they are not initiated by stalled replication forks [8]–[10] . The yeast studies of ( GAA ) n• ( TTC ) n tracts described above were done in haploid strains . In the analysis described below , we examined the properties of long ( 230 repeats ) and short ( 20 repeats ) tracts on reciprocal mitotic crossovers ( RCOs ) between homologous chromosomes in diploids . The diploid strains described in the Results section allow the selection and mapping of mitotic crossovers . In addition , crossovers are often associated with gene conversion events , the local non-reciprocal transfer of information near the site of the crossover [11] , [12] . Most meiotic gene conversion events reflect heteroduplex formation between allelic sequences , followed by repair of the resulting mismatch [11] , [13] . During meiotic recombination in yeast , the length of a gene conversion tract is usually about 1–2 kb [14] , although mitotic conversion tracts are often much longer with a median length of 7 kb [15] . In our study , both crossovers and conversion events were mapped . We find a strong stimulation of RCOs for long ( 230-repeat ) , but not short ( 20-repeat ) tracts . This hotspot activity is observed in strains heterozygous , as well as homozygous , for the long tracts , and this stimulation is not substantially affected by the orientation of the tract relative to the replication origin . Analysis of the recombination events suggests that the recombinogenic property of the long tracts is a consequence of a double-strand DNA break ( DSB ) formed within an unreplicated chromosome .
The method allowing the selection and mapping of crossovers and associated gene conversion events is shown in Figure 1 [15]–[17] . A G2-associated RCO can generate two daughter cells that are homozygous for markers that were heterozygous in the starting diploid strain . On one copy of chromosome V , the diploid has the can1-100 allele , an ochre-suppressible mutation in a gene regulating sensitivity to canavanine; yeast strains with the wild-type CAN1 allele are killed by this drug . On the other copy of chromosome V , the CAN1 gene has been deleted and replaced by SUP4-o , a tRNA gene encoding an ochre suppressor . In addition , the diploid is homozygous for ade2-1 , also an ochre mutation . In the absence of an ochre suppressor , ade2-1 strains are adenine auxotrophs and form red colonies as a consequence of accumulation of a pigmented precursor to adenine [18] . The starting diploid strain is canavanine-sensitive ( CanS ) , and forms white colonies . A RCO can be selected as a red/white sectored canavine-resistant colony . In Figure 1 , we show only one of the two possible segregation patterns , the one in which the recombined chromosomes segregate with the unrecombined chromosomes . If the two recombined chromosomes segregate into one daughter cell and the two unrecombined chromosomes segregate into the other , no canavanine-resistant sectored colony will be observed . In S . cerevisiae , these two segregation patterns are equally frequent [19] . Thus , the rate of RCOs is equivalent to twice the frequency of CanR sectored colonies in the 120 kb CEN5-can1-100/SUP4-o interval [16] . By constructing diploid strains from haploids with diverged sequences , Lee et al . [15] used single-nucleotide polymorphisms ( SNPs ) located on chromosome V to map recombination events . Thirty-four polymorphisms that altered restriction enzyme recognition sites were used . Genomic DNA from each sector of a red/white CanR colony was purified and used as a template to generate PCR products containing the SNPs . By treating these fragments with diagnostic restriction enzymes , followed by gel electrophoresis , Lee et al . [15] , [17] could determine whether the sector was homozygous or heterozygous for the polymorphism . As described in the Introduction , crossovers are frequently associated with gene conversion events . For example , in Figure 2A , we show conversion of one of the polymorphic sites adjacent to the RCO , resulting in the converted allele being found in three of the four chromosomes involved in the initial exchange; this type of event is termed a “3∶1” conversion . These events can be detected by examining the markers in both sectors of a sectored colony . In addition to 3∶1 conversion tracts ( Figure 2A ) , in analyzing spontaneous mitotic crossovers , Lee et al . [15] also found two other types of conversion tracts: 4∶0 tracts ( Figure 2B ) and 3∶1/4∶0 hybrid tracts ( Figure 2C ) . These events are likely to reflect a DSB in one homologue in G1 of the cell cycle , followed by replication of the broken chromosome , and repair of two broken chromatids in G2 . Replication of a chromosome broken in G1 is an expected outcome , since single DSBs formed in G1 do not activate the DNA damage checkpoint machinery [20] and are inefficiently processed to recombination intermediates [21] , [22] . If the conversion tracts associated with repair of both DSBs include the same markers , a 4∶0 event is generated . If one conversion tract is more extensive than the other , a hybrid 3∶1/4∶0 event would be observed . This explanation of the spontaneous mitotic RCOs and associated conversions is supported by the observation that the RCOs resulting from gamma-radiation of G1-synchronized yeast cells have 4∶0 and 3∶1/4∶0 hybrid tracts , whereas cells irradiated in G2 do not [17] . An alternative explanation of the 4∶0 and 3∶1/4∶0 hybrid tracts is that they represent two independent repair events of DSBs generated in G2 . The rate of RCOs in WXT46 is 8 . 5×10−5/division ( Table 1 ) . Of the 29 conversion events associated with the RCOs , 8 were 3∶1 events and 21 were 4∶0 or 3∶1/4∶0 hybrid tracts . If the 3∶1 events are interpreted as the frequency of single repair events in G2 , we calculate that the frequency of single events is about 2 . 3×10−5 ( [8/29] × [8 . 5×10−5] ) . The expected frequency of independent double events would be ( 2 . 3×10−5 ) 2 or about 5 . 3×10−10 . The observed frequency of “double events” ( conversion events of the 4∶0 or 3∶1/4∶0 classes ) was 5 . 3×10−5 . We conclude , therefore , that the 4∶0 and 3∶1/4∶0 hybrid tracts do not reflect two independent cycles of DSB formation and DSB repair . Previously , we used the system shown in Figure 1 and Figure 2 to measure the frequency and location of spontaneous or gamma-ray-induced recombination events in the 120 kb interval between CEN5 and the can1-100/SUP4-o markers on chromosome V . In the current study , we constructed yeast strains with insertions of ( GAA ) n• ( TTC ) n tracts of two different sizes ( 230 and 20 repeats ) in two different orientations near the URA3 gene on chromosome V ( details of the constructions in Text S1 ) . In the strains used in our study , the ( GAA ) n• ( TTC ) n tracts are embedded within lys2 sequences inserted in the intergenic region between GEA2 and URA3 . This position is about 22 kb centromere-proximal to ARS508 and about 31 kb centromere-distal to ARS510; both of these ARS elements are active origins [23] . In previous studies [4] , [5] , it was shown that long ( >100-repeat ) ( GAA ) n tracts on the lagging strand template result in a replication fork block whereas long TTC tracts on the lagging strand template do not . To determine how replication forks were blocked for strains with the ( GAA ) n• ( TTC ) n tracts inserted on chromosome V , we constructed two isogenic haploid strains in which a ( GAA ) 230• ( TTC ) 230 tract was inserted in two orientations . In the haploid MD512 , the tract was oriented such that the GAA sequence was on the “Watson” strand as designated in Saccharomyces Genome Database , and the haploid MD510 had the tract in the opposite orientation . By two-dimensional gel electrophoresis , we found a blocked replication fork in MD510 but not in MD512 ( Figure 3 ) . Since a replication fork initiated at ARS510 would encounter the GAA tract on the lagging strand in MD510 , this result suggests that tracts are replicated primarily by a replication fork initiated at ARS510 rather than ARS508 , although we have not directly examined fork movement . In our subsequent discussion of yeast strains , tracts oriented in the same direction as MD510 will be termed “ ( GAA ) n” tracts and those with the opposite orientation will be termed “ ( TTC ) n” tracts; this nomenclature is consistent with previous studies [5] . It should be noted that , in other genetic backgrounds , the chromosomal region in which we inserted the ( GAA ) n• ( TTC ) n tracts is replicated using forks that move in the opposite direction from the one observed in our genetic background [24] . We first performed a pilot experiment to examine the recombinogenic effects of ( GAA ) 230 and ( TTC ) 230 tracts in diploids heterozygous for insertion near URA3 . As described above , the rate of RCOs in the CEN5-can1-100/SUP4-o interval can be calculated from the frequency of CanR red/white sectored colonies . The rates of RCOs in MD506 ( heterozygous for the [GAA]230 tract ) and MD508 ( heterozygous for the [TTC]230 tract ) were 13×10−5/division ( ±3×10−5 ) and 6 . 2×10−5/division ( ±2×10−5 ) , respectively; 95% confidence limits are shown in parentheses . The rate of RCOs in an isogenic diploid without the tract insertion is 5 . 8×10−6/division [15] . Thus , the heterozygous tract insertions stimulated RCOs in the CEN5 to can1-100/SUP4-o interval by about 10- to 20-fold and tracts in both orientations were recombinogenic . The diploids MD506 and MD508 did not have the polymorphisms required to map the recombination events ( details of their genotypes in Text S1 and Table S1 ) . Consequently , we constructed six other diploids that were heterozygous for polymorphisms that allowed mapping of RCOs and associated conversions . The strain names , and their tract sizes and orientations are: WXTMD42 , ( GAA ) 20/ ( GAA ) 20; WXTMD46 , ( GAA ) 230/ ( GAA ) 20; WXTMD43 , ( GAA ) 230/ ( GAA ) 230; WXTMD40 , ( TTC ) 20/ ( TTC ) 20; WXTMD45 , ( TTC ) 20/ ( TTC ) 230; WXTMD41 , ( TTC ) 230/ ( TTC ) 230 . In all strains , the ( GAA ) n• ( TTC ) n tracts were inserted at the same position on chromosome V near the URA3 gene ( green rectangles in Figure 4 ) . These diploid strains were constructed from two haploid parents ( PSL2 and PSL5 ) with numerous sequence polymorphisms allowing mapping of the positions of the crossovers as described further below . The rates of RCOs with 95% confidence limits , based on an average of the number of sectored colonies in at least 20 cultures , are shown in Table 1 . Strains homozygous for ( GAA ) 20 or ( TTC ) 20 tracts ( WXTMD42 and WXTMD40 ) had rates of RCOs of about 4×10−6/division . These rates are very similar to that observed in the isogenic PSL101 strain ( 6×10−6 ) that had no GAA•TTC tracts [15] . The strains homozygous for either the ( GAA ) 230 or ( TTC ) 230 tracts ( WXTMD43 and WXTMD41 , respectively ) had RCO rates of about 2×10−4/division . Thus , the addition of a GAA•TTC tract that is only 690 base pairs in length elevated the rate of RCOs in a 120 kb interval by more than 30-fold . The strains heterozygous for the long tracts ( WXTMD46 and WXTMD45 ) also had substantially ( 20-fold ) elevated rates of RCOs; the rates of RCOs in the heterozygous strain were about half those observed in the homozygous strains , indicating the GAA•TTC sequences on the two homologues functioned independently . As found previously for the MD506 and MD508 strains , the orientation of the GAA•TTC tract has no strong effect on its recombinogenic properties . It should be noted that Break-Induced Replication ( BIR ) [12] and local gene conversion events can generate unsectored canavanine-resistant colonies; however , these colonies cannot be unambiguously distinguished from RCOs that occur prior to plating cells on canavanine-containing medium [16] . From the results described above , one obvious possibility is that ( GAA ) 230• ( TTC ) 230 tracts are preferred sites for formation of a DSB or some other type of recombinogenic DNA lesion . By this model , one would expect most of the tract-stimulated recombination events to map at or near the position of the tract . In addition , in meiotic and mitotic recombination events in yeast analyzed previously , if a diploid is heterozygous for a preferred site of DSB formation , the chromosome with the preferred site is the recipient of genetic information in a gene conversion event [12] . We examined the positions of crossovers and associated gene conversion events in two strains: WXTMD46 ( a diploid heterozygous for a ( GAA ) 230 tract ) and WXTMD42 ( a diploid homozygous for [GAA]20 tracts ) . The positions of the crossovers and gene conversion events were mapped by the methods described previously [15] . In brief , using PCR and restriction analysis , for both sectors of a CanR red/white sectored colony , we determined whether polymorphic sites on chromosome V were homozygous for the PSL2 form of the polymorphism ( shown in red in Figure 4 ) , the PSL5 form of the polymorphism ( shown in black in Figure 4 ) , or were heterozygous . In the previous studies of spontaneous or gamma-ray-induced mitotic crossovers , four types of sectored colonies were commonly observed: 1 ) RCOs unassociated with an adjacent gene conversion tract , 2 ) RCOs associated with an adjacent 3∶1 tract ( as defined in Figure 2 ) , 3 ) RCOs associated with an adjacent 4∶0 tract , and 4 ) RCOs associated with a hybrid 3∶1/4∶0 tract . Spontaneous recombination events are distributed throughout the 120 kb interval with a minor “hotspot” located near the can1-100/SUP4-o marker and a minor “coldspot” near CEN5 [15] . A summary of the mapping of crossovers and associated conversions in WXTMD46 is shown in Figure 4A . All markers proximal to the crossover are heterozygous in both red and white sectors , and homozygous distal to the crossover in both sectors ( as illustrated in Figure 2 ) . As observed for spontaneous events previously , most of the crossovers ( 29 of 33 ) were associated with conversion tracts of various sizes . 3∶1 and 4∶0 conversion tracts ( as defined in the Introduction ) are indicated by thin and thick vertical lines in Figure 4 , respectively . 3∶1/4∶0 hybrid tracts are shown by adjacent thick and thin lines . Conversion tracts shown in black indicate that genetic information was transferred from the PSL5-related homologue and red tracts show transfer of information from the PSL2-related homologue . Almost all of the conversion events in WXTMD46 included one or both of the markers flanking the GAA•TTC tract , as expected if the recombination event initiated within the tract . All four of the crossovers unassociated with conversion ( shown as green Xs ) occurred in the region containing the tract . In Figure 5 , we compare the distribution of conversion events in WXTMD46 and PSL101 ( an isogenic diploid without a GAA•TTC tract; data from Lee et al . [15] ) . The difference in the distributions of conversion events in the two strains is evident . In addition , in WXTMD46 , the conversion tracts were strongly biased in the direction that represents transfer of information from the PSL5-related homologue . This result is consistent with the recombinogenic lesion occurring on the PSL2-related homologue that contains the ( GAA ) 230 tract rather than the chromosome with the ( GAA ) 20 tract . Several other features of the conversion events are important . First , most of the conversion tracts were either 4∶0 tracts or hybrid 3∶1/4∶0 tracts . As discussed previously , such tracts are most simply interpreted as representing repair in G2 of a DSB formed in G1 ( Figure 2B and 2C ) . This issue will be discussed in more detail below . Second , although some of the observed conversion events extended symmetrically to both sides of the tract , others were asymmetric . Thus , conversion events can extend either unidirectionally or bidirectionally from the initiating DNA lesion . Third , as observed with spontaneous recombination events and events induced by gamma rays in G1 [15] , [17] , the conversion tracts were long compared to those observed in meiosis . We estimated tract length by averaging the minimal tract length ( the distance between the markers included in the tract ) and the maximal tract length ( the distance between the closest flanking markers not included in the tract ) . The median length of the tracts was 20 . 3 kb ( 95% confidence limits of 12 . 5–23 . 4 kb ) , somewhat larger than the length observed in spontaneous events without the ( GAA ) n• ( TTC ) n tracts ( 6 . 5 kb; [15] ) . The median size of meiotic conversion tract lengths is about 2 kb [14] . Fourth , as in previous studies , we found a number of examples of crossovers within a conversion tract; these events are indicated by asterisks in Figure 4 . As discussed in Lee et al . [15] , most of these events are explicable as representing the independent repair of two broken chromatids . An example of this class of conversion event is shown in Figure S1 . We also mapped a small number of RCOs in WXTMD42 , the strain homozygous for the ( GAA ) 20 tracts ( Figure 4B ) . As expected , these events were distributed throughout the CEN5 to can1-100/SUP4-o interval . In addition , the conversion events involved transfer of information from both homologues with approximately the same frequency . The median conversion tract length in WXTMD42 is 11 . 6 kb ( 95% confidence limits of 3 . 7–22 . 3 kb ) . The genetic evidence predicts the existence of a tract-associated DSB in G1 diploid cells . To look for such DSBs directly , we prepared DNA samples from stationary phase cells ( >95% unbudded cells ) of two isogenic haploid strains , WXT10 with a ( TTC ) 20 tract and WXT11 with a ( TTC ) 230 tract . Intact chromosomal DNA was isolated from cells suspended in agarose plugs to prevent shearing and the resulting samples were analyzed by contour-clamped homogeneous electric field gel electrophoresis ( CHEF gels; [25] ) . The separated chromosomal DNA molecules were transferred to nylon membranes and hybridized to URA3-specific probe . We observed a chromosomal fragment at the position expected for a DSB within the tract ( Figure 6A ) in WXT11 , but not in WXT10 ( Figure 6B ) . The fraction of broken chromosomal molecules observed in three independent experiments was about 0 . 013 ( average of 0 . 013 , 0 . 017 , and 0 . 01 ) . Although this frequency of DSBs is considerably higher than the observed frequency of RCOs ( about 10−4 ) , it is likely that many of the DSBs are repaired by pathways , such as BIR and gene conversion unassociated with RCOs , that do not generate RCOs [12] . In yeast , long ( CTG ) n• ( CAG ) n tracts are preferred sites for DSB formation in mitosis [26] . In meiosis , long ( greater than 75 repeats ) ( CTG ) n• ( CAG ) n tracts were hotspots of recombination in one study [27] , but were not in another [28] . Short ( 10-repeat ) ( CTG ) n• ( CAG ) n tracts were not meiotic recombination hotspots [29] . As shown above long ( GAA ) n• ( TTC ) n promote DSB formation in mitosis . It was reasonable to ask , therefore , whether long GAA•TTC tracts stimulate meiotic recombination , as well . To address this question , we performed tetrad analysis , measuring meiotic recombination distances in three intervals on chromosome V: CEN5-ura3; ura3-can1-100/SUP4-o ( the interval containing the tracts ) , and can1-100/SUP4-o to V9229::HYG . The heterozygous HYG gene ( encoding a protein that results in resistance to hygromycin ) was inserted approximately 20 kb centromere distal to the can1-100 gene . This analysis was done in WXTMD46 ( which contains ( GAA ) 230 on one homologue and ( GAA ) 20 on the other ) and PSL101 ( which lacks ( GAA ) n• ( TTC ) n tract insertions ) . No significant differences were observed in map distances for any of the intervals ( details of the analysis in Table S4 ) . The map distance for the interval containing the insertion was 36 cM in WXTMD46 and 37 cM in PSL101 ( total of about 100 tetrads examined in each strain ) . Strong meiotic recombination hotspots are associated with high rates of gene conversion and crossovers [30] . The ( GAA ) 230 tract in WXTMD46 is located about 1 kb from the mutant ura3 allele and the ( GAA ) 20 tract is located the same distance from the wild-type URA3 allele . If the ( GAA ) 230 tract is a preferred site for meiotic DSB formation , we would expect an elevation in gene conversion events of the 3 Ura+:1 Ura− class , since the chromosome that receives the DSB acts as a recipient for information derived from the uncut chromosome [12] . This effect should be detectable since the strong meiotic recombination HIS4 hotspot stimulates meiotic conversion events at sites located 2 . 7 kb from the hotspot [31] , a distance longer than that between the ( GAA ) 230 tract and URA3 . In PSL101 , we observed two conversion events , both 1 Ura+: 3 Ura− tetrads , in a total of 118 tetrads . In 105 tetrads derived from WXTMD46 , we found no gene conversions of the 3+:1− or 1+:3− classes , but four tetrads that had 4 Ura+: 0 Ura− spores . This 4∶0 type of conversion is consistent with a mitotic gene conversion occurring within a sub-population of the WXTMD46 cells prior to sporulation [11] . Consistent with this hypothesis , in two of the tetrads with 4 Ura+ spores , all four spores had the SUP4-o marker and were HygS . These segregation patterns are consistent with a mitotic gene conversion at the ura3 locus associated with a mitotic crossover . We also examined the meiotic stability of the ( GAA ) n• ( TTC ) n tracts by PCR analysis of spore DNA in 20 tetrads . Three patterns were observed . In 10 tetrads , two of the tracts were about 20 repeats in length and two were about 230 repeats in length . In 5 tetrads , two of the tracts were 20 repeats in length and two were of equal size but shorter than 230 repeats; this class is consistent with a sub-population of WXTMD46 cells in which the 230-repeat tract had undergone a mitotic deletion . In the third class ( 5 tetrads ) , two spores had 20-repeat tracts , one had a 230-repeat tract , and one had a tract of intermediate size; this class is consistent with a meiotic deletion event in one of the two 230-repeat tracts . Taken together with the mapping and gene conversion data , these results argue that the long ( GAA ) n• ( TTC ) n tracts are somewhat meiotically unstable , but the DSBs formed within the tract do not strongly stimulate meiotic recombination between the homologous chromosomes . This issue will be discussed further below .
Although the tendency of certain trinucleotide tracts to expand in size was first demonstrated in humans , much of the experimental research concerning the effects of genome-destabilizing effects of these sequences has been done in bacteria and the yeast Saccharomyces cerevisiae [1] , [2] , [32] . In yeast , three types of repetitive trinucleotide tracts , ( CTG ) n• ( CAG ) n , ( CGG ) n• ( CCG ) n , and ( GAA ) n• ( TTC ) n , have been examined in detail . All three types of tracts undergo frequent size alterations with the frequencies of alterations increasing as a function of the number of repeats [32] . The frequency of these alterations is also affected by the orientation of the repetitive tract with respect to the replication origin . All three tracts are capable of forming secondary structures in vitro with one strand forming a more stable secondary structure than the other [1] . The orientation in which the strand with the most stable secondary structure is on the lagging strand for replication has the highest frequency of tract alterations . This orientation is also associated with replication fork pausing [1] . For the ( GAA ) n• ( TTC ) n repeats , as discussed above , replication fork pausing is observed when the ( GAA ) n repeats are on the lagging strand [4] . Somewhat unexpectedly , large-scale expansions of ( GAA ) n• ( TTC ) n tracts occur with approximately the same frequency regardless of the orientation of the tract [6] . Since DSBs are recombinogenic [12] and since DSBs are observed at the sites of long ( CTG ) n• ( CAG ) n and ( GAA ) n• ( TTC ) n tracts [5] , [26] , one would expect that such tracts would be hotspots for recombination . Long ( CTG ) n• ( CAG ) n tracts stimulate intrachromosomal recombination between repeats and sister-chromatid exchanges [26] , [33]; long ( GAA ) n• ( TTC ) n tracts elevate the frequency of recombination between repeats on non-homologous chromosomes in yeast [5] and plasmid-plasmid recombination in E . coli [7] . In these assays , it was unclear whether the recombination events were reciprocal ( producing two recombined DNA molecules ) or non-reciprocal . The assay used in our current study selects for reciprocal events . We found that the 230-repeat tract elevates the rate of RCOs in 120 kb interval from about 5×10−6/division ( strains with no tract or a 20-repeat tract ) to about 2×10−4/division . We calculate that the rate of RCOs/kb in the strains without the tract is about 4×10−8/kb/division . The 690 bp tract has a rate of RCOs of about 3×10−4/kb/division . Consequently , the ( GAA ) 230• ( TTC ) 230 tract is about 104-fold more recombinogenic than an average yeast sequence . In contrast to the strong recombinogenic effects of the tract on mitotic recombination , no strong stimulation was observed for meiotic exchange . Since we observed meiosis-specific alterations in tract length in about 25% of the tetrads that were analyzed , it is likely that the long ( GAA ) n• ( TTC ) n tracts are substrates for DSB formation in meiosis . The lack of a detectable effect of the tracts on meiotic recombination can be explained in two ways . First , it is possible that tract-associated DSBs are repaired by intrachromosomal interactions ( Synthesis-Dependent Strand Annealing , SDSA ) or sister-chromatid exchanges [12]; neither of these events would be detected by standard tetrad analysis . Meiosis-specific intra-allelic changes in the lengths of minisatellites consistent with SDSA events have been observed previously in humans [34] and yeast [35] , [36] . Second , it is possible that the effects of a weak tract-associated hotspot would be obscured by the very high frequency of meiosis-specific DSBs catalyzed by Spo11p . We note , however , that a strong tract-associated hotspot would have been detected by our analysis . The strong HIS4 recombination hotspot , for example , increases the map length in the LEU2-HIS4 interval from 20 cM to 36 cM [37] . Previously , we showed that about 40% of spontaneous RCOs were associated with 4∶0 or 3∶1/4∶0 hybrid conversion tracts [15] . We suggested that such events were a consequence of DSB formation on an unreplicated chromosome , followed by replication of the broken chromosome , and repair of the two resulting broken chromatids ( Figure 2B and 2C ) . Since most of the RCOs stimulated by the ( GAA ) n• ( TTC ) n repeats are associated with 3∶1/4∶0 tracts , it is likely that the recombinogenic DSBs are formed in G1 . This conclusion , based on genetic analysis , is also supported by the physical analysis demonstrating tract-associated DSBs in stationary phase cells ( Figure 6 ) . Since the DSBs occur in G1/G0 , the observation that the tract-associated stimulation of RCOs is independent of the orientation of the tract is expected . It should be emphasized that our results do not show that tract-associated DSBs occur only in G1/G0 . We observed previously that ( GAA ) n• ( TTC ) n tracts stimulate ectopic recombination between repeats on non-homologous chromosomes in an orientation-dependent mechanism [5] . We suggest that these events are likely to be non-reciprocal and , therefore , regulated differently than the RCOs that are the subject of the present study . In summary , our studies of the properties of ( GAA ) n• ( TTC ) n tracts indicate that they promote genetic instability by several different mechanisms . One mechanism is dependent on the orientation of the repeats and is likely to reflect breakage of replication forks [5]; this mechanism is also associated with small tract contractions/expansion and ectopic recombination events [4] , [5] . A second mechanism is the orientation-independent large expansion of ( GAA ) n• ( TTC ) n tracts that may involve strand-switching events in which the leading strand copies an Okazaki fragment [6] . The third mechanism is also independent of the orientation and likely reflects DSB formation in G1 to yield RCOs . Although we have not determined the source of the G1-induced DSBs , they may reflect the action of DNA repair enzymes and/or topoisomerases interacting with secondary structures formed by the tracts . Replication-independent instability has been observed in mammalian cells for both ( GAA ) n• ( TTC ) n and ( CTG ) n• ( CAG ) n tracts [8] , . This instability appears to be related to DNA repair events associated with transcription [9] , [10] . In most of the strains examined in our study , the ( GAA ) n• ( TTC ) n tracts were embedded in a promoter-less fragment of the LYS2 gene . The most obvious difference in the patterns of spontaneous RCOs observed previously [15] and those seen in the current study is the location of the events . All of the events observed in the current study are at or near the site of the ( GAA ) n• ( TTC ) n tracts , presumably because all events are initiated at or near the tracts . Although the distribution of spontaneous events observed by Lee et al . is not completely random , it is clear that the events can be initiated at many sites within the 120 kb interval ( Figure 5 ) . Although the properties of DNA sequences that regulate the probability of initiating a mitotic recombination events have not yet been completely established , mitotic recombination is promoted by closely-spaced inverted repeats [38] and by high rates of transcription [39] , [40] . The median length of the tract-stimulated conversion events in WXTMD46 ( 20 . 3 kb ) is longer than those observed for spontaneous events in the absence of the repetitive sequence ( 6 . 5 kb ) and conversion events generated in G1-arrested cells by gamma radiation ( 7 . 3 kb; [17] ) . The median tract length is much longer than the median length observed associated with RCOs induced by gamma radiation in G2-arrested cells ( 2 . 7 kb; [17] ) . Most of the conversion tracts are 3∶1/4∶0 hybrid tracts ( Figure 4A ) . As discussed in the Introduction , such tracts can be explained by independent repair of two DSBs . If the DSBs occur within the GAA•TTC insertions , we expect that the 4∶0 region of the hybrid tract should include one or both of the markers flanking the tract , and this expectation is met ( Figure 4A ) . If processing of the broken DNA ends is bidirectional and symmetric from the site of the DSB , most tracts should have a 4∶0 region flanked by 3∶1 regions . Although we observe this pattern for some of the conversion events , for other events , the 4∶0 region is at one end of the hybrid tract . Thus , we infer that the mechanism that generates the gene conversion in mitosis can be asymmetric . In addition , single conversion events can be propagated from the initiation site either toward the centromere or toward the telomere . Meiotic gene conversion tracts share these properties [41] , [42] . Two different mechanisms can result in a gene conversion event . During meiotic recombination , most conversion events reflect heteroduplex formation followed by repair of any resulting mismatches . One key early intermediate in this process is a broken end that has been “processed” by 5′ to 3′ degradation on one of the two strands [13] . It is possible that mitotic conversion events involve much more extensive processing than meiotic events or extensive branch migration of the Holliday junction ( s ) associated with the strand invasion . An alternative possibility is that the conversion events involve the repair of a double-stranded gap [43] . Although there is strong evidence that mitotic events that generate relatively short conversion tracts are a consequence of heteroduplex formation followed by mismatch repair [44] , [45] , it is currently unclear whether the very long tracts are a consequence of mismatch repair or gap repair [15] . In summary , we have demonstrated that ( GAA ) 230• ( TTC ) 230 tracts strongly stimulate RCOs and our analysis indicates that these events are initiated by a DSB in unreplicated DNA . These results have several implications relevant to the genetic instability observed in patients with Friedreich's ataxia . First , a G1-associated DSB may be an intermediate in the expansion process in at least a sub-set of the expansion events . Second , since we find that the ( GAA ) 230• ( TTC ) 230 tracts are highly recombinogenic by a mechanism that is independent of DNA replication , our findings may be relevant to the observation that the FRDA-associated tracts are unstable in post-mitotic ( non-dividing ) cells and these expansions contribute to pathogenesis . For example , the highest rate of somatic instability is observed in dorsal root ganglia , which is the most damaged tissue in FRDA patients [46] . In addition , expanded ( GAA ) n• ( TTC ) n tracts may elevate the frequency of loss of heterozygosity ( LOH ) on the chromosome containing the expanded tract , allowing heterozygous mutations to become homozygous . Since there are other ( GAA ) n• ( TTC ) n runs within mammalian genomes that are prone to expansions [47] , such tracts may also promote LOH on other chromosomes . It would be of interest to examine tissues of FRDA patients or cell lines derived from patients for tract-associated DSBs ( using ligation-mediated PCR ) or LOH of single-nucleotide polymorphisms located centromere-distal to the expanded tracts .
Most of the experiments involve diploids generated by crosses of haploids with diverged DNA sequences . The haploid strain PSL5 [15] is derived from the YJM789 genetic background whereas PSL2 [15] is derived from W303a [39] . The details of the constructions and genotypes of the haploid and diploid strains are given in Text S1 and Tables S1 , S2 , S3 . The diploids strains used to measure the effect of GAA•TTC tracts on RCOs were homozygous for the ade2-1 mutation , and heterozygous on chromosome V for can1-100 and an allelically-placed copy of SUP4-o . As described in the text , this system allows the selection of RCOs as CanR red/white sectored colonies . Standard yeast procedures were used for transformations , mating , sporulation , and tetrad dissection [48] . Media were prepared as described previously [15] , [16] . The two-dimensional gel analysis of replication forks was done as described previously [5] . DNA samples for the gel analysis were treated with the AflII restriction enzyme , and the Southern blot was hybridized to the 3 . 9 kb LYS2-specific AflII fragment isolated from pFL39LYS2 ( described in Text S1 ) . To analyze tract-associated DSBs , we grew haploid strains to stationary phase ( three days of growth in rich growth medium [YPD] at 30°C ) , and then prepared DNA by methods described previously [25]; in the stationary-phase cultures , >95% of the cells were unbudded as expected for cells in G1/G0 . Chromosomal DNA molecules were separated using the Bio-Rad CHEF Mapper XA . The Southern analysis was done using a URA3-specific probe that was prepared by PCR amplification of genomic DNA with the primers: URA3-f ( 5′ GGTTCTGGCGAGGTATTGGATAGTTCC ) and URA3-r ( 5′ GCCCAGTATTCTTAACCCAACTGCAC ) . The hybridization signals were detected and quantitated using a PhosphorImager . The methods used to quantitate RCOs in various strains were identical to those described previously [15] . In brief , individual colonies formed on rich growth medium were suspended in water , and plated on non-selective medium ( omission medium lacking arginine [SD-arg] ) or on medium containing canavanine ( SD-arg with 120 micrograms/ml canavanine ) . Plates were incubated at room temperature for four days , followed by storage for one day at 4°C ( which accentuates the red color of sectors ) . The rate of RCOs for each strain was determined by averaging the frequency of crossovers observed in at least 20 independent cultures ( colonies ) . Red and white CanR strains were purified from each half of the sectored colonies . DNA was isolated by standard procedures [48] . As we have done previously , we mapped crossovers by examining 34 single-nucleotide polymorphisms ( SNPs ) located in the 120 kb interval between CEN5 and the can1-100/SUP4-o markers . For each SNP , the DNA from one of the haploid parents contained a diagnostic restriction enzyme recognition that was altered for the other parent . For each SNP , we amplified genomic DNA using primers flanking the heterozygous marker , treated the fragment with the diagnostic restriction enzyme , and examined the products by gel electrophoresis . From this analysis , we could determine whether the sectored colony was homozygous for the YJM789 form of the SNP , homozygous for the W303a form of the SNP , or heterozygous for the polymorphism . The sequence of the primers and restriction enzymes used in the analysis are given in Lee et al . [15] . Statistical analyses were done using the VassarStats Website ( http://faculty . vassar . edu/lowry/VassarStats . html ) . Most of the comparisons involved the Fisher exact test . 95% confidence limits on the rates of RCOs were calculated by determining the 95% confidence limits on the proportions ( number of sectored colonies/number of colonies on non-selective plates ) using the Wilson procedure with a correction for continuity . Calculations of median conversion tract lengths and 95% confidence limits on the median were done as described previously [17] . | Although meiotic recombination has been much more studied than mitotic recombination , mitotic recombination is a universal property . Meiotic recombination rates are quite variable within the genome , with some chromosomal regions ( hotspots ) having much higher levels of exchange than other regions ( coldspots ) . For mitotic recombination , although some types of DNA sequences are known to be associated with elevated recombination rates ( highly-transcribed genes , inverted repeated sequences ) , relatively few hotspots have been described . In this report , we show that a 690 base pair region consisting of 230 copies of the ( GAA ) n• ( TTC ) n trinucleotide repeat stimulates mitotic crossovers in yeast 10 , 000-fold more strongly than an “average” yeast sequence . This sequence is a preferred site for chromosome breakage in stationary phase yeast cells . Our findings may be relevant to understanding the expansions of the ( GAA ) n• ( TTC ) n trinucleotide repeat tracts that are associated with the human disease Friedreich's ataxia . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/chromosome",
"biology"
] | 2011 | Friedreich's Ataxia (GAA)n•(TTC)n Repeats Strongly Stimulate Mitotic Crossovers in Saccharomyces cerevisae |
During development , axons must integrate directional information encoded by multiple guidance cues and their receptors . Axon guidance receptors , such as UNC-40 ( DCC ) and SAX-3 ( Robo ) , can function individually or combinatorially with other guidance receptors to regulate downstream effectors . However , little is known about the molecular mechanisms that mediate combinatorial guidance receptor signaling . Here , we show that UNC-40 , SAX-3 and the SYD-1 RhoGAP-like protein function interdependently to regulate the MIG-2 ( Rac ) GTPase in the HSN axon of C . elegans . We find that SYD-1 mediates an UNC-6 ( netrin ) independent UNC-40 activity to promote ventral axon guidance . Genetic analysis suggests that SYD-1 function in axon guidance requires both UNC-40 and SAX-3 activity . Moreover , the cytoplasmic domains of UNC-40 and SAX-3 bind to SYD-1 and SYD-1 binds to and negatively regulates the MIG-2 ( Rac ) GTPase . We also find that the function of SYD-1 in axon guidance is mediated by its phylogenetically conserved C isoform , indicating that the role of SYD-1 in guidance is distinct from its previously described roles in synaptogenesis and axonal specification . Our observations reveal a molecular mechanism that can allow two guidance receptors to function interdependently to regulate a common downstream effector , providing a potential means for the integration of guidance signals .
Axonal migrations are guided through the developing nervous system by multiple extracellular guidance cues that activate receptors on the surface of the axonal growth cone . The axonal growth cone must integrate these multiple signals to arrive at a single directional response [1–7] . Progress has been made in characterizing individual guidance cues , receptors and their signaling pathways . However , the signaling mechanisms that allow for integration between receptor signaling pathways are largely unknown [8] . Integration of multiple guidance signals could be mediated by combinatorial receptor signaling [8] . In fact , several examples of combinatorial axon guidance receptor signaling have been identified in various systems . For example , commissural axons in the mammalian spinal cord exhibit a hierarchical interaction between the DCC ( UNC-40 ) and Robo ( SAX-3 ) receptors [9] . Upon reaching the midline , spinal axons are exposed to Slit , which promotes formation of a DCC/Robo heterodimer that silences the attractive response to Netrin , thereby allowing the axons to exit the midline . Variations of hierarchical interactions between the DCC and Robo receptors have also been observed in several other systems [10–13] . Moreover , hierarchical interactions between semaphorin receptors have also been observed [14 , 15] . Guidance signaling pathways can also exhibit synergistic interactions . For example , in C . elegans , the AVM neuron is simultaneously exposed to both the UNC-6 ( Netrin ) and SLT-1 ( Slit ) guidance cues . Both of these cues guide the AVM axon towards the ventral nerve cord . UNC-40 ( DCC ) can function individually as a receptor for UNC-6 , whereas SAX-3 ( Robo ) can function individually as a receptor for SLT-1 . However , UNC-40 can also function independently of UNC-6 , when it binds to the SAX-3 receptor to promote the response to SLT-1 [16] . Despite these observations , the signaling mechanisms that mediate combinatorial guidance signaling remain mostly unexplored . SYD-1 is an intracellular protein that contains a RhoGAP-like domain that has been implicated in synaptogenesis and axonal specification [17–21] . Here , we present evidence that SYD-1 can also function in axon guidance to promote UNC-40 signaling independently of its canonical ligand UNC-6 . Our data suggest that UNC-40 , SAX-3 and SYD-1 can function interdependently to regulate the MIG-2 GTPase . Our observations provide an example of a signaling mechanism that can mediate combinatorial guidance signaling .
The cell body of the HSN neuron is located on the lateral body wall of C . elegans and extends an axon ventrally to the ventral nerve cord . To determine the relative contributions of guidance cues in HSN ventral axon guidance , we examined HSN axon guidance in slt-1 ( eh15 ) and unc-6 ( ev400 ) null mutants ( Fig 1A–1C ) . We found that the unc-6 null mutants exhibited 59% guidance defects , whereas the slt-1 null mutants had 2% guidance defects . However , the unc-6 slt-1 double null mutants had 89% guidance defects . These observations suggest that the HSN axon is guided primarily by the UNC-6 and SLT-1 guidance cues . Moreover , the synergistic nature of these interactions suggest interactions between the UNC-6 and SLT-1 signaling pathways in the HSN neuron , providing a system for studying the factors that mediate interactions between these pathways . To investigate interactions between the signaling pathways that transduce the UNC-6 and SLT-1 cues , we examined guidance defects in mutants that disrupt their receptors , UNC-40 and SAX-3 , respectively . We attempted to construct an unc-40; sax-3 double null mutant . However , this double null mutant was completely embryonic lethal , precluding analysis of axon guidance . As an alternative we used the sax-3 ( ky200 ) temperature sensitive mutation . The single sax-3 ( ky200 ) mutant at 24°C has guidance defects with a penetrance of 19% , suggesting that this allele behaves as a strong loss of function at 24°C ( Fig 1D ) . We constructed an unc-40 ( e1430 ) ; sax-3 ( ky200 ) double mutant . Although these animals were mostly embryonic lethal at 24°C , we were able to analyze axon guidance in the survivors and found that guidance defects in unc-40 ( e1430 ) ; sax-3 ( ky200 ) double mutants were significantly enhanced relative to either single mutant ( Fig 1D ) . Likewise , we also found that guidance defects in unc-6 ( ev400 ) sax-3 ( ky200 ) double mutants were significantly enhanced relative to either single mutant . These results suggest that UNC-40 and SAX-3 can function in parallel to each other and that these two receptors mediate most , if not all , of the ventral guidance signals in the HSN axon . SAX-3 can function independently of its canonical ligand SLT-1 . To investigate the relationship between SLT-1 and SAX-3 , we analyzed mutants that disrupt these proteins . Although SAX-3 can function as a receptor for the SLT-1 guidance cue [22–24] , we found that in the HSN , sax-3 ( ky123 ) null mutants had 20% defects , whereas slt-1 ( eh15 ) null mutants had only 2% defects ( Fig 1E ) . The sax-3 slt-1 double mutants showed no enhancement relative to the sax-3 single mutants . These observations suggest that in addition to being a receptor for SLT-1 , SAX-3 can also function independently of SLT-1 . These results are consistent with several previous findings of SLT-1 independent SAX-3 function [12 , 23 , 25] . UNC-40 can function independently of its canonical ligand UNC-6 . UNC-40 primarily functions as a receptor for the UNC-6 guidance cue [26–27] . However , the penetrance of HSN axon guidance defects was enhanced in unc-40 ( e1430 ) ; unc-6 ( ev400 ) double null mutants relative to unc-40 ( e1430 ) and unc-6 ( ev400 ) single null mutants ( Fig 1F ) . Similar results were obtained using the unc-40 ( n324 ) null allele ( S1 Fig ) . These observations suggest that UNC-40 can function independently of UNC-6 , in the HSN neuron , consistent with previous reports of an UNC-6 independent function for UNC-40 in several different processes [16 , 26 , 28–30] . Moreover , the identification of an UNC-6 independent UNC-40 signaling activity in the HSN neuron provides a means to identify the signaling mechanisms that mediate interactions between the UNC-40 and the SAX-3 signaling pathways . To identify a signaling mechanism that can mediate UNC-6 independent UNC-40 function , we searched for candidate mutations that can enhance guidance phenotypes in unc-6 null mutants , but not in unc-40 null mutants . We found that the syd-1 ( ju2 ) null mutation could enhance guidance defects in unc-6 ( ev400 ) null mutants , but not in unc-40 ( e1430 ) null mutants ( Fig 2A ) . We also obtained similar results using different null alleles for syd-1 and unc-40 ( Figs 3B and S2 ) . These observations suggest that SYD-1 might mediate UNC-6 independent UNC-40 signaling . To determine if SYD-1 and UNC-40 function together to mediate UNC-6 independent UNC-40 function , we analyzed guidance defects in unc-40; syd-1; unc-6 triple null mutants and compared these to defects in the syd-1; unc-6 and unc-40; unc-6 double null mutants ( Fig 2B ) . If unc-40 and syd-1 function in a genetic pathway , the triple mutant should have a penetrance of guidance defects similar to the double mutants . Indeed , we found that the penetrance of guidance defects in the unc-40; syd-1; unc-6 triple mutant was similar to the syd-1; unc-6 or the unc-40; unc-6 double null mutants . These observations are consistent with a model where UNC-40 and SYD-1 function together to mediate UNC-6 independent UNC-40 function . To further test the idea that UNC-40 and SYD-1 function together , we analyzed double mutants between the syd-1 ( ju2 ) null allele and two different hypomorphic unc-40 alleles ( Fig 2C ) . When two genes function in a pathway , a loss of function mutation in one gene can enhance defects associated with a hypomorphic mutation in the second gene . Indeed , we found that guidance defects associated with either of two unc-40 hypomorphic alleles could be enhanced by the syd-1 ( ju2 ) null mutation , consistent with the idea that unc-40 and syd-1 function in a genetic pathway . SYD-1 is expressed as three different isoforms , known as SYD-1A , SYD-1B , and SYD-1C . SYD-1A and SYD-1B contain a PDZ domain , whereas SYD-1C does not contain the PDZ domain ( Fig 3A ) . Loss of SYD-1A and SYD-1B disrupts axonal identity in DD and VD motor neurons [17] and also disrupts presynaptic development in the HSN neuron [20 , 31] . For these phenotypes , the syd-1 ( ju82 ) mutation that disrupts SYD-1A and SYD-1B , but spares SYD-1C ( see Fig 3A ) , behaves as a null allele . These observations indicate that SYD-1C is not sufficient for axonal specification or presynaptic development . To examine the role of each isoform of SYD-1 in axon guidance , we analyzed guidance defects in different mutant alleles of syd-1 ( Fig 3B ) . The syd-1 ( ju82 ) allele disrupts the SYD-1A and SYD-1B isoforms , but not the SYD-1C isoform . The syd-1 ( ju2 ) and syd-1 ( tm6234 ) alleles disrupt all three isoforms . We found that the syd-1 ( ju82 ) allele , did not enhance guidance defects in unc-6 null mutants . However , both syd-1 ( ju2 ) and syd-1 ( tm6234 ) did enhance guidance defects in unc-6 null mutants . These genetic interactions suggest that the role of the syd-1 gene in guidance may be mediated by the SYD-1C isoform . To further test the role of the SYD-1C isoform , we used the unc-86 promoter to drive expression of SYD-1C in the HSN neuron . We found that expression of SYD-1C in the HSN neuron can rescue guidance defects in syd-1 ( ju2 ) ; unc-6 ( ev400 ) double null mutants ( Fig 3B ) , indicating that the SYD-1C isoform is sufficient to mediate the function of the syd-1 gene in guidance . Moreover , these results also indicate that SYD-1C functions cell-autonomously in the HSN neuron . Together , these results suggest that the protein domains contained within SYD-1C are sufficient to mediate axon guidance and that the PDZ domain contained within SYD-1A and SYD-1B is not required for axon guidance . By contrast , loss of SYD-1A and SYD-1B disrupts synaptogenesis and axon specification , suggesting that the PDZ domains are required for these functions and implying that the function of SYD-1 in guidance is distinct from its role in synaptogenesis and axon specification . Although our data suggest that the SYD-1C isoform is necessary and sufficient to mediate the role of syd-1 in axon guidance , other interpretations are possible and we cannot rule out potential roles for the other isoforms of SYD-1 . Since genetic data suggest that UNC-40 and SYD-1 function together , we tested for a physical interaction between SYD-1C and the cytoplasmic domain of UNC-40 ( Fig 3C ) . We found that SYD-1C binds to the cytoplasmic domain of UNC-40 fused to GST ( GST::UNC-40 ) , but not to GST alone . In addition , we found that SYD-1C can also bind to the cytoplasmic domain of SAX-3 ( GST::SAX-3 ) , but not to GST alone . By contrast , we observed no binding between an unrelated protein , luciferase , and either GST::UNC-40 or GST::SAX-3 . These observations suggest that SYD-1C can bind to the cytoplasmic domains of both UNC-40 and SAX-3 . SYD-1 contains a RhoGAP-like domain , but has not been associated with any Rho GTPases . By testing candidate GTPases , we found that the GAP-like domain of SYD-1 binds to MIG-2 . To confirm this interaction we performed binding assays with recombinant MIG-2 and the C-terminus of SYD-1 , which contains the RhoGAP-like domain ( Fig 4A ) . We found that MIG-2 fused to GST ( GST-MIG-2 ) was pulled-down with the His6-tagged C-terminus of SYD-1 ( His-SYD-1 ) , while GST alone was not . Moreover , binding between MIG-2 and SYD-1 was markedly enhanced when MIG-2 was bound to GTPγS ( a non-hydrolyzable analog of GTP ) relative to when MIG-2 was bound to GDP . These observations indicate that the RhoGAP-like domain of SYD-1 preferentially binds to the GTP-bound active form of MIG-2 . To determine if SYD-1 can regulate MIG-2 activity , we tested for genetic interactions between the syd-1 ( ju2 ) null allele and mig-2 gain of function mutations . The mig-2 ( gm103 ) mutation encodes a fully activated MIG-2 , whereas mig-2 ( gm38 ) encodes a partially activated MIG-2 [32–33] . We found that loss of syd-1 can enhance the defects associated with the partially active mig-2 ( gm38 ) allele ( Fig 4B ) , suggesting that SYD-1 can negatively regulate MIG-2 . By contrast , we observed no enhancement of the fully active mig-2 ( gm103 ) allele , consistent with the expectation that fully activated MIG-2 cannot be further activated . To further test the functional relationship between SYD-1 and MIG-2 , we transgenically expressed SYD-1C in the mig-2 ( gm38 ) partially activated gain of function mutants ( Fig 4C ) . Consistent with the idea that SYD-1 can negatively regulate MIG-2 , we found that transgenic expression of SYD-1C suppresses guidance defects caused by the mig-2 ( gm38 ) gain of function mutation . To further test the interaction between SYD-1 and MIG-2 , we constructed a syd-1 ( ju2 ) ; mig-2 ( ok2273 ) double null mutant . We found that penetrance of HSN guidance defects in these double null mutants was not enhanced relative to mig-2 ( ok2273 ) single mutants ( Fig 5 ) , further supporting the idea that SYD-1 functions with MIG-2 . Together , these data are consistent with a model where SYD-1 can negatively regulate the activation state of MIG-2 . To consider the role of mig-2 in HSN axon guidance , we examined the HSN axon in mig-2 ( ok2273 ) null mutants ( Fig 5 ) . We found that 14% of HSN axon migrations were defective in the mig-2 null mutants , indicating that MIG-2 is required for HSN axon guidance . Likewise , 20% of HSN axon migrations were defective in sax-3 ( ky123 ) null mutants . To determine if mig-2 functions in a genetic pathway with sax-3 , we examined sax-3 mig-2 double null mutants . If two genes function in a genetic pathway , a double null mutant should have defects similar to the greatest of the single mutants . We found that the sax-3 mig-2 double mutants had a penetrance of guidance defects similar to that of sax-3 single mutants , indicating that mig-2 and sax-3 do function in a genetic pathway . Consistent with a function for mig-2 in the sax-3 pathway , we also found that the mig-2 null mutation can enhance defects in the unc-40 null mutants . Together , these genetic interactions support the conclusion that MIG-2 functions in the SAX-3 pathway . Since SYD-1 regulates MIG-2 , which functions with SAX-3 , we asked if SYD-1 function requires SAX-3 . To address this question , we examined genetic interactions between mutations in syd-1 and sax-3 . We found that the syd-1 ( ju2 ) null mutation fails to enhance axon guidance defects in sax-3 ( ky123 ) null mutants , suggesting that SYD-1 might function with SAX-3 ( Fig 6A ) . To further determine if SYD-1 can function with SAX-3 , we used the sax-3 ( ky200 ) hypomorphic allele and found that the syd-1 ( ju2 ) null mutation enhances guidance defects in sax-3 ( ky200 ) hypomorphic mutants ( Fig 6B ) . Since , the syd-1 null mutant can enhance guidance defects in the sax-3 hypomorphic mutants , but not in sax-3 null mutants , we conclude that SYD-1 functions with SAX-3 , implying that SYD-1 function requires SAX-3 .
We have found that single null mutants of unc-40 and sax-3 exhibit partially penetrant guidance defects . We have also found that unc-40; sax-3 double strong loss of function mutants exhibit guidance defects that are nearly fully penetrant . Together , these results suggest that UNC-40 and SAX-3 can function in parallel to each other . Although UNC-40 and SAX-3 can function in parallel , our genetic analysis suggests that UNC-40 and SAX-3 can also function together . This idea is consistent with previous work in the AVM neuron showing that UNC-40 and SAX-3 can function in combination with each other , whereby UNC-40 functions independently of UNC-6 to potentiate SAX-3 signaling [16] . Consistent with this genetic analysis , biochemical data indicate that both UNC-40 and SAX-3 can bind to each other [16] . Likewise , DCC and Robo , vertebrate homologs of UNC-40 and SAX-3 , respectively , also exhibit functional and biochemical interactions [9–11] . Despite these findings , little is known about the signaling events that mediate combinatorial guidance receptor function . Our results suggest that SYD-1 function requires both UNC-40 and SAX-3 , implying that UNC-40 , SAX-3 and SYD-1 function interdependently . Loss of syd-1 function fails to enhance guidance defects in null mutants of either unc-40 or sax-3 . However , loss of syd-1 function does enhance guidance defects in hypomorphic mutants of either unc-40 or sax-3 . These observations suggest that SYD-1 function does not occur in the complete absence of UNC-40 or SAX-3 , but becomes apparent when UNC-40 or SAX-3 function is reduced . Moreover , we also find that SYD-1 binds to the cytoplasmic domains of both UNC-40 and SAX-3 . Together , these observations are consistent with a model whereby SYD-1 does not function in the UNC-40 individual pathway or the SAX-3 individual pathway , but rather functions in a common pathway with both UNC-40 and SAX-3 . However , we note that other interpretations for these results are possible and that future investigations will be required to further define the relationships between UNC-40 , SAX-3 and SYD-1 . Our results suggest that UNC-40 , SAX-3 and SYD-1 function interdependently , implying that SYD-1 can mediate combinatorial receptor signaling . We propose two models to explain how UNC-40 , SAX-3 and SYD-1 could mediate combinatorial signaling ( Figs S3 and S4 ) . In the heterodimer model , UNC-40 and SAX-3 would bind to each other and function as an UNC-6 independent guidance receptor . This heterodimer receptor would bind to SYD-1 and cause negative regulation of MIG-2 , which functions in the SAX-3 pathway . Alternatively , in the crosstalk model , UNC-40 and SAX-3 would not be physically associated with each other . In this crosstalk model , SYD-1 would bind to the cytoplasmic tail of the UNC-40 receptor and would negatively regulate MIG-2 , which functions in the SAX-3 pathway . We favor the heterodimer model because previous results have indicated that UNC-40 and SAX-3 can bind to each other [16] , our genetic analysis suggest that SYD-1 function requires both UNC-40 and SAX-3 , and our biochemical analysis show that SYD-1 can bind to both UNC-40 and SAX-3 . Future investigations may help differentiate between these two models , or potentially suggest other models . Our finding that a syd-1 null allele can enhance guidance defects in partially activated mig-2 mutants , but not in fully activated mig-2 mutants , implies that SYD-1 is involved in regulating the activation state of MIG-2 . However , SYD-1 lacks two critical residues that are required for GAP activity [34–35] . Thus far , attempts to detect SYD-1 GAP activity have been unsuccessful . Since we have found that the GAP domain of SYD-1 can bind to activated MIG-2 , we propose that SYD-1 functions as a scaffold that can promote negative regulation of MIG-2 activity . One possibility is that SYD-1 and MIG-2 may be part of a complex that also includes a GAP protein . Alternatively , it remains possible that SYD-1 could possess GAP activity on its own . We report that UNC-6 independent UNC-40 signaling is mediated by SYD-1 and that SYD-1 negatively regulates MIG-2 . Thus , UNC-6 independent UNC-40 signaling can promote axon guidance by negatively regulating MIG-2 . This idea is consistent with previous results showing that negative regulation of Rac promotes signaling downstream of Robo ( SAX-3 ) . In Drosophila , CrossGAP ( Vilse ) has been identified as a GAP that negatively regulates Rac activation to promote Robo signaling [36 , 37] . In mammalian neurons , srGAP1 has been identified as a GAP for Rac and Cdc42 that is required for Robo signaling [38–39] . Positive regulation of Rac has also been reported , whereby Son of Sevenless ( Sos ) , a Guanine Nucleotide Exchange Factor ( GEF ) , can positively regulate Rac to promote Robo signaling [40] . Thus , Robo signaling involves both positive and negative regulation of Rac . Consistent with this idea , guidance defects result from either Rac gain of function or Rac loss of function mutations [32 , 41] . Likewise , guidance defects are caused by loss of function or gain of function in CrossGAP , a GAP for Rac in the Drosophila [36–37] . Together , these observations indicate that axon guidance requires precise control of Rac activation state , which is achieved through both positive and negative regulation of Rac . SYD-1C resembles its mammalian homolog mSYD1A , in that both are predicted to contain an Intrinsically Disordered Domain ( ID domain ) at their N-terminus , rather than a PDZ domain [42] . ID domains can shift between disordered and ordered confirmations and can also engage in intramolecular interactions that affect the function of other domains [42 , 43] . Although SYD-1C and mSYD1A both contain predicted ID domains , the amino acid sequence of the ID domain is not conserved between SYD-1C and mSYD1A . Consistent with this different amino acid sequence , the function of SYD-1C and mSYD1A appears to be different , in that the former regulates axon guidance and the later regulates synaptic vesicle docking [42] . These different functions are likely to be mediated by different binding partners for the ID domains of SYD-1C and mSYD1A . Although our study has focused on UNC-6 independent UNC-40 function , we have also observed evidence for SLT-1 independent SAX-3 function . This idea is supported by the higher penetrance of HSN guidance defects in sax-3 null mutants relative to slt-1 null mutants . In fact , SLT-1 independent SAX-3 signaling has been observed in several different processes , including AVM and PVM axon guidance [12 , 23 , 25 , 44] . Much like in the HSN , the PVM neuron shows a higher penetrance of HSN guidance defects in sax-3 null mutants relative to slt-1 null mutants [12] . Genetic analysis has suggested that in the PVM neuron SAX-3 can inhibit UNC-40 signaling in the absence of SLT-1 [12] . This inhibitory interaction between SAX-3 and UNC-40 also occurs in the AVM neuron [12 , 44] . However , the AVM neuron shows a higher penetrance of guidance defects in slt-1 null mutants relative to sax-3 null mutants . The reason for the differences in slt-1 and sax-3 mutant defects in the AVM relative to the HSN and PVM are not known . However , these differences could be explained by a greater degree of redundant signaling in the HSN and PVM neurons relative to the AVM neuron . Since we have found that UNC-40 can function independently of UNC-6 , one might expect unc-40 null mutants to have a greater penetrance of guidance defects relative to unc-6 null mutants . However , we have found that unc-6 and unc-40 null mutants have a similar penetrance of HSN guidance defects . One possible explanation could be that UNC-6 can also function independently of UNC-40 in the HSN . Indeed , the existence of UNC-40 independent UNC-6 activity is apparent in AVM and PVM axon guidance , where unc-6 mutants have a higher penetrance of guidance defects relative to unc-40 mutants [12] . Moreover , ENU-3 is thought to form part of an UNC-40 independent UNC-6 signaling pathway in the AVM and PVM neurons [45] .
Genetic manipulations were carried out using standard procedures . The N2 Bristol genetic background was used in all experiments and all animals were maintained at 20°C on NGM plates seeded with OP50 bacteria . The following alleles were used and are considered to be null: unc-6 ( ev400 ) [46] , slt-1 ( eh15 ) [26] , sax-3 ( ky123 ) [47] , unc-40 ( e1430 ) [26] , syd-1 ( ju2 ) [17] , syd-1 ( tm6234 ) , syd-1 ( ju82 ) [17] , mig-2 ( ok2273 ) [48] . The following alleles are considered to be hypomorphic: sax-3 ( ky200 ) [47] , unc-40 ( ev546 ) [49] , unc-40 ( tm5504 ) [49] . The following alleles are considered to be gain of function: mig-2 ( gm38 ) [32] , mig-2 ( gm103 ) [32] . The unc-40 ( ev546 ) allele was obtained from Joseph Culotti . The syd-1 ( tm6234 ) allele was obtained from Shohei Mitani . All other alleles were obtained from the Caenorhabditis Genetics Center ( CGC ) . The pAGC5 plasmid includes Punc-86::syd1c::sl2::tagrfp and was created using Gibson Assembly and cloned into pCFJ910 , provided by Erik Jorgensen via Addgene ( Addgene Plasmid #44481 ) . The DNA sequence encoding TAGRFP was obtained from Julie Plastino [50] . pRSET-SYD-1C ( pCZGY658 ) , which encodes His6-tagged C-terminus of SYD-1C protein ( AA320-721 ) , was generated using the Gateway cloning system ( Invitrogen ) . Transgenes expressing SYD-1C in the HSN neuron were created by injecting the pAGC5 plasmid at 4 ng/μl with 50 ng/μl of odr-1::RFP and 50 ng/μl Bluescript into wildtype worms . Two independent transgenic lines were established and crossed into the syd-1 ( ju2 ) ; unc-6 ( ev400 ) genetic background . HSN axon guidance was analyzed as described below . Axon guidance was analyzed as described previously [51 , 52] . The zdIs13 transgene [53] , which expresses Ptph-1::gfp , was used to visualize the HSN axon in young adult animals . Except where noted , HSN guidance defects were scored as defective if the axon failed to reach the ventral nerve cord . For experiments involving the sax-3 ( ky200 ) hypomorphic allele , only weak defects were observed , where the axon trajectory is abnormal , but the axon eventually reaches the ventral nerve cord . Therefore , in experiments involving the sax-3 ( ky200 ) hypomorphic allele , HSN axons were scored as defective if their trajectory was abnormal , but they eventually reached the ventral nerve cord . Recombinant GST and GST-MIG-2 proteins were produced in Escherichia coli DH5α with plasmid pGEX6P and pGEX4T-MIG-2 ( obtained from Hiroshi Qadota [54] , respectively , by incubating for 5 h at 37°C in the presence of 0 . 3 mM IPTG . His6-tagged C-terminus of SYD-1C containing RhoGAP-like domain ( His-SYD-1 ) was produced in E . coli BL21 ( DE3 ) with pRSET-SYD-1C ( pCZGY658 ) by incubating for 16 h at 22°C in the presence of 0 . 3 mM IPTG . GST- and His6- tagged proteins were extracted by sonication in PBS pH 7 . 4 with 0 . 5% Triton X-100 and protease inhibitors ( Roche ) and centrifuging . GST-MIG-2 was incubated with 0 . 1 mM of GTPγS or 1 mM of GDP ( Sigma-Aldrich ) in PBS pH 7 . 4 with 0 . 5% Triton X-100 , 10 mM MgCl2 , 10 mM EDTA , 2% glycerol and protease inhibitors for 30 min at 30°C , and then supplied additional MgCl2 to 75 mM . For pull-down analysis , His-SYD-1 was immobilized to HisPur Cobalt Resin ( Thermo ) , incubated with GST or GST-MIG-2 proteins for 1 h at 4°C , washed with PBS pH 7 . 4 with 0 . 1% Triton X-100 and 20 mM imidazole for 4 times , eluted with 300 mM imidazole in PBS pH 7 . 4 , 0 . 1% Triton X-100 . The eluate and input were subjected to SDS-PAGE , and GST- and His- tagged proteins were detected by western blotting with mouse monoclonal Anti-GST antibody ( Upstate Biotechnology ) and Anti-His-tag mAb ( Medical and biological laboratories ) , respectively , using ECL Advance Western Blotting Detection Kit ( GE Healthcare ) . SYD-1C protein was produced with the TNT SP6 quick coupled in vitro transcription and translation system ( Promega ) and labeled with biotin . GST::UNC-40 was produced in bacteria and coupled to Glutathione-Sepharose . Binding assays were conducted for 16 hours at 4°C in PBS with 0 . 1% Triton X-100 , 1% BSA and protease inhibitors . After binding , samples were washed 3 times with wash buffer ( PBS and 0 . 1% Triton X-100 ) . Bound material was detected by SDS-PAGE electrophoresis and western blotting with Strepavidin-Alkaline Phosphatase . | The nervous system is comprised of a complex network of axonal connections . This network is formed during development , when axons navigate to their target regions . Axon navigation requires multiple signaling pathways to detect and respond to extracellular guidance cues . Many of the guidance cues , receptors and signaling pathways have been identified . However , little is known about how the information encoded by different guidance cues is combined to arrive at a directional response . A key part of how this occurs is likely to involve combinatorial receptor signaling , when different guidance receptors work in combination with each other . However , the mechanisms that underlie combinatorial receptor signaling remain mostly unknown . In C . elegans , ventral axon guidance is mediated by the UNC-40 ( DCC ) and SAX-3 ( Robo ) guidance receptors . Here , we use genetic and biochemical analysis to identify a molecular mechanism that can mediate combinatorial signaling involving UNC-40 , SAX-3 and the RhoGAP-like protein SYD-1 . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | SYD-1C, UNC-40 (DCC) and SAX-3 (Robo) Function Interdependently to Promote Axon Guidance by Regulating the MIG-2 GTPase |
Indirect reciprocity , in which individuals help others with a good reputation but not those with a bad reputation , is a mechanism for cooperation in social dilemma situations when individuals do not repeatedly interact with the same partners . In a relatively large society where indirect reciprocity is relevant , individuals may not know each other's reputation even indirectly . Previous studies investigated the situations where individuals playing the game have to determine the action possibly without knowing others' reputations . Nevertheless , the possibility that observers of the game , who generate the reputation of the interacting players , assign reputations without complete information about them has been neglected . Because an individual acts as an interacting player and as an observer on different occasions if indirect reciprocity is endogenously sustained in a society , the incompleteness of information may affect either role . We examine the game of indirect reciprocity when the reputations of players are not necessarily known to observers and to interacting players . We find that the trustful discriminator , which cooperates with good and unknown players and defects against bad players , realizes cooperative societies under seven social norms . Among the seven social norms , three of the four suspicious norms under which cooperation ( defection ) to unknown players leads to a good ( bad ) reputation enable cooperation down to a relatively small observation probability . In contrast , the three trustful norms under which both cooperation and defection to unknown players lead to a good reputation are relatively efficient .
We often help others even when the helping behavior is costly . The Prisoner's Dilemma game and its variants are used for examining cooperative behavior in such social dilemma situations . Several mechanisms can explain the emergence and maintenance of cooperation [1] , [2] . Direct reciprocity , one such mechanism , is relevant when the same pair of players repeatedly interact [3] , [4] . To avoid retaliation from a peer player , it is beneficial for both players to maintain cooperation . However , direct reciprocity cannot explain cooperation in relatively large societies , where players do not repeatedly meet each other . Indirect reciprocity is a main mechanism for cooperation in cases where players never interact with the same partners [5]–[8] . In indirect reciprocity , players are motivated to help others and receive help from different others . There are two types of indirect reciprocity mechanisms: upstream and downstream reciprocity [2] , [9] , [10] . The two types of indirect reciprocity differ in the direction of the chain of helping behavior . In upstream reciprocity , a player is motivated to help after the player has been helped by someone . In downstream reciprocity , a player will be helped after the player has helped someone . In downstream reciprocity , players possess unique reputation scores , determined by their past actions toward other players . Players help others with a good reputation but not those with a bad reputation . Nowak and Sigmund proposed a computational model in which players helping others are regarded to be good and those withdrawing help are regarded to be bad [6] , [7] . According to their model , helping others to maintain a good reputation is more beneficial than withdrawing help to gain momentary profits . Empirical studies also support downstream reciprocity [11]–[14] . In the present study , we focus on downstream reciprocity and simply refer to it as indirect reciprocity . The rule according to which players decide either to cooperate or to defect based on the reputations of the relevant players is called the action rule [15] , [16] . The discriminator that helps those with a good reputation and does not help those with a bad reputation is an example of the action rule . The rule for assigning a reputation to players based on their actions is called the social norm [15] , [16] . Nowak and Sigmund's norm is termed image scoring [6] , [7] . Theoretically , the discriminator is an unstable action rule under image scoring because the discriminator is invaded by the unconditional cooperator [17]–[19] . The discriminator is stable under some more complex social norms including standing [15] , [17] , [18] , [20]–[23] , judging [15] , [21]–[24] , and shunning [10] , [23] , [25] . These complex social norms require more information about other players than image scoring does , such as the co-player's reputation , in addition to the information on the player's action toward the co-player . Unless an authority maintains the reputation of all the players , as in the case of online marketplaces [26] , [27] and communities of medieval merchants [28] , the information about players , which is indispensable for indirect reciprocity , must spread from players to players via gossiping [10] , [14] , [15] , [29] . However , except in a sufficiently small population , the accuracy and span of gossip may be limited [16] , [30] , [31] . In such a case , the information about players is shared incompletely in the population , and individuals often need to make decisions when the information about the relevant players is unknown . Several studies have addressed the case in which players do not necessarily know the reputation of others [6] , [7] , [18] , [19] , [21] , [30] , [31] . However , these studies have two limitations . First , it is assumed in these studies that only players playing the game , not the third-party observers of the game , incompletely perceive the reputation of other players . The role of the third-party observer is to generate the reputation about players and disseminate it to other players in the population . The observer can propagate the reputation about players to others only when the observer knows the reputation about the players in question . Figure 1 illustrates the point . In a one-shot game , a player knows or does not know the co-player's reputation ( A ) . In addition , an observer , who watches the game but does not play the game , knows or does not know the co-player's reputation ( B ) . Previous studies considered incomplete observation of type A but not B . In reality , however , the interacting player and the observer are roles that the same individual may play on different occasions such that both roles may accompany incomplete access to information about others . Second , these studies examined the sustainability of a few exemplary combinations of the social norm and action rule ( e . g . , combinations of the image scoring norm and the discriminator action rule [6] ) . The choice of the pairs of social norm and action rule is subjective . On the other hand , exhaustive studies in which all the pairs of social norm and action rule within a certain class are examined are not concerned with the issue of incomplete information [15] . These studies considered erroneous behavior , such as wrong assignment of the reputation to other players . However , the error probability is eventually set to be infinitesimally small . We assume that the information about the reputation is available to interacting players and observers with an arbitrary probability between and . In the present study , we perform an exhaustive search to explore the possibility of indirect reciprocity under the social norms that permit observers to assign reputations to unknown players . The manner in which individuals may know the reputation about others generally depends on details of information spreading processes ( e . g . , gossiping on a social network ) . We do not consider explicit mechanisms of information spreading and model the incomplete observation by the probability with which a player and an observer know the reputation of the co-player in a one-shot game . In particular , we investigate two types of observation: concomitant and independent observation ( see Results ) . Our exhaustive analysis reveals that the trustful discriminator that helps players with a good or unknown reputation and does not help players with a bad reputation is the only self-supporting action rule under several social norms . Even if the fraction of players knowing others' reputation is relatively small , the population can be sufficiently cooperative .
We generalize the model of indirect reciprocity derived from the donation game with binary reputation values [7] , [15] , [16] , [18] , [19] , [21]–[23] , [25] , [30]–[32] with an additional assumption that players know the reputation of a fraction of other players . Consider an infinitely large population . From this population , we arbitrarily select two players , one as a donor and the other as a recipient . The donor either cooperates ( ) with or defects ( ) against the recipient . If the donor cooperates , the donor pays cost , and the recipient gains benefit . We assume such that the defection is rational for the donor in a one-shot game , whereas cooperation contributes to the welfare of the population . We repeat the same procedure until each player is paired with a sufficient number of others but never with the same opponent . In this way , we exclude direct reciprocity . Consequently , the participation of each player in the games as a donor and a recipient is equally probable . Each player possesses a binary reputation value , i . e . , good ( ) or bad ( ) . We assume that a third player serves as an observer of a one-shot game and assigns or to the donor . In a one-shot game , the donor and the observer know the reputation of the recipient with probability ( ) . The probability that the reputation of the recipient , which is actually or , is unknown ( ) to the donor and the observer is . The recognition of the reputation by the donor and that by the observer are assumed to occur concomitantly or independently ( see Results ) . In contrast to previous studies , observers as well as donors in our model are imperfect with regard to knowing the recipients' reputation . The donor's action ( or ) depends on the recipient's reputation ( , , or in the donor's eyes ) . For example , a player that cooperates with a recipient , represented as , and also cooperates with and recipients ( i . e . , and ) is referred to as the unconditional cooperator ( ) . A player obeying the action rule , , and is called the unconditional defector ( ) . A player obeying the action rule , , and is a discriminator that also cooperates with recipients whose reputation is unknown to the donor; this discriminator is denoted by . Because an action rule is specified by the allocations of or to each of , , and , there are action rules . The observer updates the reputation of the donor based on the donor's action ( or ) and the recipient's reputation ( , , or in the observer's eyes ) . We refer to the update rule as the social norm . The class of social norms that we are considering is called the second-order assessment [10]; the update rule depends on two kinds of information: the donor's action and the recipient's reputation . When ( therefore , no recipients ) , simple standing , stern judging , and shunning , which are schematically shown in Fig . 2 , belong to this class . To simplify notation , we henceforth refer to simple standing , stern judging , and shunning as standing , judging , and shunning , respectively . For example , in the case of standing , the observer assigns reputation when the donor cooperates with a good recipient ( ) or when the recipient is bad ( , ) and assigns when . Because a second-order social norm is specified by the allocations of or to each of , , , , , and when , there are social norms . We also assume that the donor receives a new reputation opposite to that intended by the observer with a small probability . With probability , the observer assigns a reputation to the donor according to the social norm . This error models the limited ability of the observer . Another reason for introducing the error is that and players must coexist in the population for distinguishing the efficiency of different social norms and action rules . We analyze the stability and cooperativeness of the homogeneous population of each of the action rules under each of the social norms by adopting the exhaustive search method introduced in Refs . [15] , [16] . Given a value of , we check whether each combination of the social norm and action rule ( there are combinations in total ) satisfies the following two properties . Stability: For a given social norm , an action rule is a strict Nash equilibrium ( NE ) , if the payoff of the action rule against itself is greater than the payoff of any other action rule against the focal action rule . Cooperativeness: For a given social norm , an action rule is cooperative , if players in the homogeneous population in which everyone adopts the focused action rule cooperate with a sufficiently large probability . The precise procedure is as follows .
In this section , we assume that the recipient's reputation in a single game is known or not known by the donor and the observer concomitantly . There are two possible situations in a single game ( see Fig . 3 ( A ) ) . With probability , both the donor and observer know the recipient's reputation . With probability , neither the donor nor the observer knows the recipient's reputation . and , used in Eqs . ( 1 ) and ( 2 ) , are given by ( 7 ) ( 8 ) where is the probability that the action of the donor obeying is regarded to be by the observer , is the recipient's reputation in the donor's eyes , and is the recipient's reputation in the observer's eyes . Note that and if the donor's action is regarded to be and except in the case of the assignment error , respectively . We found that , except for , there are 24 pairs of social norms and action rules in which the action rule is a strict NE . The number of pairs should actually be considered as 12 because the system is invariant if we flip and in all the entries of the social norm and the action rule [15] . For example , consider the following two pairs X and Y of social norm and action rule . X consists of the social norm under which donors always receive and the action rule , i . e . , , , and . Y consists of the social norm under which donors always receive and the action rule , , and . Because we obtain Y by flipping and in X , X and Y are essentially the same . Among the 12 strict NE pairs , three pairs are cooperative . The unique action rule that is cooperative under each of the three social norms is . The three social norms are schematically shown in Fig . 3 , where rows represent the donor's actions and columns represent the recipient's reputations . They are common in that the cooperation with a or recipient is regarded to be and the defection against a or recipient is regarded to be . Under these social norms , observers suspect that donors defecting against recipients are defectors . Therefore , we refer to these social norms as suspicious social norms , namely , suspicious standing , suspicious judging , and suspicious shunning ( Fig . 4 ( A ) ) . The suspicious social norms generalize standing , judging , and shunning , which are the unique stable and cooperative second-order social norms when everybody knows the reputation of each other ( i . e . , ; Fig . 2 ) [23] . Under all the three social norms , is stable in the shaded parameter region in Fig . 4 ( B ) , i . e . , ( 9 ) Equation ( 9 ) is also required for indirect reciprocity in a model with a different assumption for incomplete observation of reputations [6] , [7] , [10] . Generally speaking , the probability of knowing the reputation of others must be greater than the cost-to-benefit ratio for sustaining indirect reciprocity . When , is invaded by six action rules , i . e . , all the other action rules except . In this section , we assume that the recipient's reputation in a single game is known or not known by the donor and the observer independently . There are four possible situations in a single game ( see Fig . 3 ( B ) ) . First , both the donor and observer know the recipient's reputation , with probability . Second , only the donor knows the recipient's reputation , with probability . Third , only the observer knows the recipient's reputation , with probability . Finally , neither of them knows the recipient's reputation , with probability . and , used in Eqs . ( 1 ) and ( 2 ) , are given by ( 10 ) and ( 11 ) We found that , except for , there are essentially 27 pairs of social norms and action rules in which the action rule is a strict NE . Seven of these 27 pairs are cooperative . As in the case of the concomitant observation ( see subsection “Concomitant Observation” above ) , the unique action rule that is cooperative under each of the seven social norms is . Figure 5 ( A ) , 5 ( C ) , 5 ( E ) , and 5 ( G ) represents the seven social norms . The corresponding parameter regions in which is stable under these social norms are shown in Fig . 5 ( B ) , 5 ( D ) , 5 ( F ) , and 5 ( H ) . The three social norms shown in Fig . 5 ( A ) and 5 ( C ) are those found in the case of the concomitant observation ( Fig . 4 ( A ) ) , i . e . , suspicious standing , suspicious judging , and suspicious shunning . Under suspicious standing and suspicious shunning ( Fig . 5 ( A ) ) , is stable in the shaded parameter region in Fig . 5 ( B ) , i . e . , ( 12 ) Under suspicious judging ( Fig . 5 ( C ) ) , is stable in the shaded parameter region in Fig . 5 ( D ) , i . e . , ( 13 ) The condition is similar to that for the concomitant observation ( Eq . ( 9 ) ) . When , is invaded by six action rules , i . e . , all the other action rules except . In contrast to the case of the concomitant observation , there are upper bounds of for to be stable under the three suspicious social norms . When in Eq . ( 12 ) or in Eq . ( 13 ) is violated , is invaded by for the following intuitive reason . Because of the probability ( ) with which the assignment error occurs , the reputation of some players is . Let us suppose that a recipient's actual reputation is correctly known by the donor but not by the observer; the recipient's reputation in the observer's eyes is . This event can occur in the case of the independent , but not concomitant , observation . In this situation , a donor defects against the recipient and gains a reputation . Meanwhile , an donor cooperates and gains a reputation . Then , donors in later rounds help but not . Therefore , invades . The three social norms shown in Fig . 5 ( E ) constitute another set of generalizations of standing , judging , and shunning . They differ from the suspicious social norms ( Figs . 4 ( A ) , 5 ( A ) , and 5 ( C ) ) in that the defection against a recipient having reputation in the observer's eyes is regarded to be . Under these social norms , observers trust donors defecting against recipients by supposing that the donors are discriminators defecting against recipients and not that the donors are mere defectors . Therefore , we call them trustful social norms , i . e . , trustful standing , trustful judging , and trustful shunning . Under the three trustful social norms , is stable when ( 14 ) which is a stricter condition than . does not invade under these trustful social norms . Intuitively , this is because defection against a recipient in the eyes of the observer is regarded to be , which cancels the superiority of over that is present under the suspicious social norms . However , must be larger than that in the case of the suspicious social norms to prevent invasion by other action rules . This is because observers do not assign a reputation and cannot discriminate mere defectors from discriminators when the recipient's reputation is in the observer's eyes . When , is invaded by six action rules , i . e . , all the other action rules except . The social norm shown in Fig . 5 ( G ) is not a variant of standing , judging , or shunning . Because cooperation with recipients is only regarded to be when the recipient's reputation is known under this social norm , we name this social norm suspicious-Theognis after the ancient Greek poet Theognis of Megara , who said “He that doeth good to the baser sort suffereth two ills—deprivation of goods and no thanks” [33] . Suspicious-Theognis is the same as the suspicious judging ( Fig . 5 ( C ) ) except that under suspicious-Theognis , defection against recipients in the eyes of the observer is regarded to be . This assignment event can occur only when the recipient actually has a reputation . In this situation , the donor never defects; the donor defects only when the recipient actually has reputation . Consequently , the player's payoff is the same under suspicious judging and suspicious-Theognis , whereas the parameter region in which is stable against the other action rules differs for the two social norms . Under suspicious-Theognis , is stable in the shaded parameter region in Fig . 5 ( H ) , i . e . , ( 15 ) The condition is severer than , which corresponds to suspicious judging ( Eq . ( 13 ) ) . Regardless of the value of , is necessary for cooperation under suspicious-Theognis ( Eq . ( 15 ) ) ; however , as , only is needed under the other six social norms including suspicious judging . When , is invaded by six action rules , i . e . , all the other action rules except . If , invades for the same reason as that for the three suspicious social norms shown in Fig . 5 ( A ) and 5 ( C ) . Paradoxically , the condition under which is stable is severe when is large . When observers know recipients' reputations , they always assign to donors defecting against recipients . Therefore , when is large , is invaded by other defective action rules . In the limit , is unstable regardless of the value of . In contrast , the other six social norms shown in Fig . 5 ( A ) , 5 ( C ) , and 5 ( E ) converge to the conventional standing , judging , or shunning norms in the limit . Our results obtained in this and the previous sections are consistent with those in the previous literature obtained for [23] . In Model , we assumed that donors and observers know the reputation of recipients with the same probability . However , this probability may also be different for donors and observers because a player may have different interests or attention levels depending on whether the player faces a game as donor or observer . In the case of concomitant observation ( see Results ) , this distinction is irrelevant . Let and be the probabilities that the donor and the observer know the recipient's reputation in a single game , respectively . In the case of independent observation , the parameter regions in which is stable are shown in Table 1 . Table 1 indicates that all the four conditions contain the factor in their lower bounds of . This implies that if donors know recipients' reputation with a large probability , is relatively resistant to invasion by six action rules , i . e . , all the other action rules except . Three of the four conditions ( except for the trustful social norms shown in Fig . 5 ( E ) ) have upper bounds of that also contain the factor . Therefore , if donors know recipients' reputations sufficiently frequently , is invaded by . The reason for this is the same as that described in subsection “Independent Observation” above . donors defect against recipients if they know that the recipients' reputations are , whereas such defection is regarded to be if the observers do not know the recipients' reputations . In contrast , donors do not receive reputation via this route . However , because the three upper bounds of contain the factor or , a large value of prevents the invasion by . This is because , if the observers know the recipients' reputations sufficiently frequently , donors' defection against recipients is judged as . As explained in “Independent Observation” , the situation in which the donor does and the observer does not know the recipient's reputation crucially affects the upper bounds of the parameter region in which is stable . The lower bound of for suspicious-Theognis ( Fig . 5 ( G ) ) contains the factor . Under this social norm , the blindness of the observer enlarges the stability region of . This occurs intuitively because if observers know recipients' reputation with a large probability , defection tends to be regarded as . To identify the most efficient of the seven social norms , we compare them in terms of the payoff that the player obtains . In the homogeneous population , the payoff of is given by . Therefore , the question of highest efficiency is reduced to the comparison of derived from the different social norms . In Eq . ( 6 ) , is satisfied under all the seven social norms because we have imposed cooperativeness . Thus , we compare in Eq . ( 6 ) . Because the payoff of under the suspicious judging and suspicious-Theognis norms is exactly the same , we compare the payoffs of under the six social norms shown in Fig . 5 ( A ) , 5 ( C ) , and 5 ( E ) . Figure 6 shows the social norms that realize the largest payoff of for various values of and . Trustful standing is the most efficient when ( 16 ) holds ( blue region ) . Suspicious standing is the most efficient when ( 17 ) holds ( green region ) . These two social norms are variants of standing . under the suspicious judging and suspicious-Theognis has an equal and the highest payoff when ( 18 ) holds ( yellow region ) . This parameter region ( yellow ) is narrower than those in which the variants of standing are the most efficient ( blue and green ) . Nowhere in the parameter region are variants of shunning the most efficient . When , only the variants of standing are the most efficient . When , the variants of standing and judging are the most efficient for different ranges of and . Variants of standing are the most efficient in a broad parameter region; this is intuitively because observers under variants of standing assign to donors more often than observers under variants of judging and shunning and because the fraction of cooperation increases with the number of players . However , to prevent the invasion by defectors , observers should assign to inappropriate donors .
The present study is motivated by the premise that in a relatively large-scale society , players and observers may not know each other even indirectly . Under any viable social norm , the unique action rule stabilizes a cooperative society . cooperates with good and unknown recipients and defects against bad recipients . behaves trustfully toward ( i . e . , cooperates with ) unknown recipients , and such a trustful discriminator also supports cooperation in other models of indirect reciprocity [6] , [7] , [18] , [19] , [31] , [32] . We emphasize that we did not prefabricate but derived it through an exhaustive search . Previous studies only focused on social norms of discrete orders . Under first-order social norms ( for observers ) , observers have no information about the reputation of players . Under higher-order social norms ( for observers ) , observers have the complete information about the reputation of players . We set for observers as well as for donors . The social norms that we discovered can be classified into suspicious social norms in which observers discriminate between cooperative and defective donors interacting with unknown recipients and trustful social norms in which observers always assign a good reputation to donors interacting with unknown recipients . In the case of independent observation , there is a trade-off between trustful and suspicious social norms . Trustful social norms are more efficient in the sense that they yield the highest payoff of when they are stable ( blue region in Fig . 6 ) , while suspicious social norms enable indirect reciprocity down to a smaller value of . We have only considered the case in which all the players in a population obey a unique social norm . Note that a few recent studies investigated competition between players obeying different norms [34]–[36] . In contrast , such a trade-off does not exist for donors; trustful donors are always better than suspicious donors in our model and in the previous models [2] , [7] , [18] . The exhaustive search method was pioneered by Ohtsuki & Iwasa [15] . In Ref . [15] , the combinations of third-order social norm and action rule under complete observation are exhaustively searched . By definition , the third-order social norms and action rules depend not only on the donor's action and the recipient's reputation but also on the donor's reputation . Ohtsuki & Iwasa [15] found that the eight third-order social norms , called the leading eight , sustain indirect reciprocity . The discriminator or the so-called contrite TFT is stable and cooperative depending on the social norm included in the leading eight . The leading eight possesses properties similar to those of the stable and cooperative second-order social norms that we discovered . The leading eight includes essentially second-order simple-standing and stern-judging , whose extensions were identified as stable and cooperative social norms in the present study . In contrast , shunning , which we discovered in the extended form , is not included in the leading eight . This discrepancy is caused by the different assumptions regarding incomplete observation employed in these studies; Ohtsuki & Iwasa set , and we set . If , observers obeying shunning always assign to donors when recipients have reputation . Therefore , the reputation dynamics leads to a large fraction of players . If , observers may assign to donors when the observers do not know the recipients' reputations . In fact , the results for shunning are qualitatively different between the cases and . We did not explore third-order social norms ( i . e . , social norms using donors' reputations ) with incomplete observation ( ) because it would be difficult to comprehend plethora of results obtained from the exhaustive search of third-order social norms with . Instead , we found that the trustful and suspicious second-order social norms , which are distinguished for , sustain indirect reciprocity . In the donation game under a second-order social norm , we should distinguish between three types of the observation probability , as shown in Fig . 7 . is the probability that the donor knows the recipient's reputation . is the probability that the observer knows the recipient's reputation and uses it to assign a reputation to the donor . is the probability that the observer observes the donor's action and assigns a reputation to the donor . Observers are confined to a first-order social norm when and can use complete second-order social norms when . If , the discriminator is stable under three second-order social norms , i . e . , simple standing , stern judging , and shunning [23] . Nowak & Sigmund [6] , [7] studied the case under image scoring ( i . e . , ) . When , cooperation is difficult for a small value of and a necessary condition for indirect reciprocity is given by [7] . Although our model is different from theirs , our results are consistent with this necessary condition for their model . They also performed numerical simulations in which a player observes a game with probability ( ) and updates the image score of the donor [6] . refers to the image score of only when plays with as donor . Panchanathan & Boyd [18] considered two different action rules , discriminator and contrite TFT , under a third-order standing norm . They found that both strategies can be ESS for . Brandt & Sigmund [21] numerically analyzed the case and . They showed that for a small , cooperation is relatively easily accomplished under image scoring and third-order standing than third-order judging . Following Mohtashemi & Mui [30] , Brandt & Sigmund [31] investigated the image scoring ( i . e . , ) when and ( ) increases with time . They found that the trustful discriminator and the unconditional cooperator can stably coexist . Finally , Brandt & Sigmund [19] elaborated the case , under image scoring ( ) in various situations . Table 2 summarizes the previous models . In the present study , we conducted an exhaustive search of stable and cooperative pairs of social norms and action rules when , , and . In the context of incomplete observation , most previous models of indirect reciprocity assumed that the ability of observers is either null ( ) or complete ( ) ( see Table 2 ) , which is in contrast with the graduated ability of observation ( i . e . , ) assumed for donors . If a player acts as a donor and an observer in different situations , it seems likely to assume real-valued ( ) . For this case , we showed that indirect reciprocity is possible for various values of and . Under incomplete observation , a small fraction of players may observe a donor 's action , and these observers may inform others of 's reputation via gossip [10] , [15] . Suppose that a player observes a one-shot game and propagates 's reputation to the entire population with probability and that nobody observes the one-shot game with probability . In this case , when is selected as a recipient in a later one-shot game , the donor and the observer of this game may concomitantly know 's reputation . Alternatively , suppose that the initial observers always exist and propagate 's reputation to a fraction , , of players directly or indirectly . If is later selected as recipient and the observer is always selected from the neighborhood of the donor in the social network of gossiping , it is probable that the donor and observer concomitantly know the recipient's reputation . Independent observation does not require these assumptions and may be more natural than concomitant observation . We showed that in our model , even with independent observation , cooperation is achieved in a large parameter region , albeit smaller than that for the concomitant observation . Previous studies focused on the situation that donors , but not observers , have incomplete information about the society . Without an authority responsible for reputation assignment , we believe that donors and observers are temporary and not fixed roles for individuals such that observers as well as donors are exposed to incomplete information . The present results provide an important step toward understanding indirect reciprocity in self-sustaining societies . | Humans and other animals often help others even when the helping behavior is costly . Several mechanisms can explain the emergence and maintenance of cooperation . In one such mechanism called indirect reciprocity , individuals are rated according to their past behavior toward others . Individuals help others with a good reputation but not those with a bad reputation . Indirect reciprocity is relevant in relatively large societies where individuals do not meet each other repeatedly . Then , unless an authority maintains the reputation of individuals , individuals would not know information about some others even indirectly via gossip . We investigated a model in which both individuals playing the game ( acting players ) and observers of the game , who evaluate acting players and start gossiping , incompletely perceive others . In the unique viable and cooperative strategy , one cooperates with good and unknown peers and defects against bad peers . Populations of suspicious observers under which cooperation ( defection ) to unknown peers is regarded to be good ( bad ) enable cooperation in relatively wide parameter regions . In contrast , populations of trustful observers under which both cooperation and defection to unknown peers are regarded to be good are relatively efficient . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"theoretical",
"biology",
"evolutionary",
"modeling",
"biology",
"computational",
"biology"
] | 2011 | Indirect Reciprocity under Incomplete Observation |
Neuronal synchronization reflected by oscillatory brain activity has been strongly implicated in the mechanisms supporting selective gating . We here aimed at identifying the anatomical pathways in humans supporting the top-down control of neuronal synchronization . We first collected diffusion imaging data using magnetic resonance imaging to identify the medial branch of the superior longitudinal fasciculus ( SLF ) , a white-matter tract connecting frontal control areas to parietal regions . We then quantified the modulations in oscillatory activity using magnetoencephalography in the same subjects performing a spatial attention task . We found that subjects with a stronger SLF volume in the right compared to the left hemisphere ( or vice versa ) also were the subjects who had a better ability to modulate right compared to left hemisphere alpha and gamma band synchronization , with the latter also predicting biases in reaction time . Our findings implicate the medial branch of the SLF in mediating top-down control of neuronal synchronization in sensory regions that support selective attention .
In order to operate in complex environments , it is necessary to selectively attend to relevant information while inhibiting distraction . It has been proposed that changes in neuronal synchronization implement the mechanism required for selective gating [1 , 2] . The increase in synchronization supports a gain increase [3] as well as information transfers to downstream regions by means of communication through coherence [4] . For instance , neurons in the monkey visual cortex activated by a given object show increased gamma-band ( 50–90 Hz ) synchronization when attention is allocated to that object [1 , 5] . These results generalize to human electroencephalography ( EEG ) and magnetoencephalograhy ( MEG ) studies that have identified increased gamma band activity associated with selective attention [6–8] . Alpha oscillations on the other hand have been proposed to reflect active inhibition of distracting information . This is underscored by alpha oscillations ( 8–12 Hz ) being relatively strong in regions anticipating distracting input [9–11] . Modulations in both the alpha and gamma band are predictive of performance in visual attention tasks [6 , 12–14] . Given that these neuronal oscillations are modulated by selective attention , they are under top-down control . The aim of this study is to identify the anatomical pathways supporting the top-down control of the oscillatory activity in sensory regions . Cue-directed shifts of attention are believed to be subserved by the dorsal attentional network [15] consisting of the frontal eye field ( FEF ) and intraparietal sulcus ( IPS ) , in contrast to the ventral attentional network that governs stimulus-driven attentional shifts [15] . Recent studies using transcranial magnetic stimulation ( TMS ) have implicated the dorsal network in providing top-down control of alpha [16–18] and gamma [18] oscillations . Communication within the dorsal network must be subserved by structural connections , and there is evidence that the development of frontoparietal white matter tracts co-occurs with recruitment of superior frontal and parietal cortex during attention and working memory tasks [19 , 20] . The superior longitudinal fasciculus ( SLF ) , a network of white-matter fiber tracts consisting of medial , middle , and lateral branches [21] , has recently been proposed to connect prefrontal control areas to posterior regions . In particular , the medial SLF branch ( SLF1 ) projects to areas overlapping with the dorsal network—namely posterior superior frontal cortex in and near to the FEF and the IPS [21] . The lateral branch ( SLF3 ) projects to nodes in the ventral network ( inferior frontal gyrus and temporoparietal junction [21] ) , while the middle branch ( SLF2 ) supposedly provides connections between the two networks . Individual differences in SLF2 volume have been shown to predict behavioral attentional biases [21 , 22] . Further , the number of SLF1 connections predicts the disruptive effects of FEF perturbation with TMS on visual task performance [23] . Given that individual differences in the SLF are behaviorally relevant , we hypothesize that the variance in these tracts also explains individual abilities to modulate alpha and gamma oscillations in sensory regions . In the present study , we performed both MEG and high angular resolution diffusion imaging ( HARDI ) magnetic resonance ( MR ) measurements in the same subjects . Oscillatory brain activity was quantified from the MEG data while the subjects performed a cued spatial attention task requiring attention to the left or right visual hemifield . From the MR data , we used whole-brain spherical deconvolution tractography [24 , 25] to reconstruct the SLF branches . We hypothesized that the medial branch ( SLF1 ) —connecting superior frontal to parietal cortex [21]—served as the structural pathway for controlling oscillatory brain activity in visual brain regions . Therefore , individual differences in SLF1 properties should predict individual ability to modulate visual cortical oscillations and thereby performance on a spatial attention task .
We first confirmed previous results demonstrating that both anticipatory alpha oscillations ( defined as 8–12 Hz activity in a 1 s window prior to presentation of the target and distractor stimuli ) and stimulus-induced gamma activity ( defined as 50–90 Hz activity in a 400 ms window following target and distractor presentation ) in occipital brain regions are modulated by direction of attention . Attentional modulation index ( AMI ) was calculated for each sensor j according to the formula AMIj = 100% * ( PowerAttention left , j—PowerAttention right , j ) / ( PowerAttention left , j + PowerAttention right , j ) . The sensor-level analysis revealed a robust increase in gamma band activity in response to the target contralateral to the attended hemifield ( Fig 2A and 2B ) . This finding is consistent with gamma band synchronization reflecting visual processing that is modulated by selective attention . The alpha band activity was strongly modulated in the cue-target interval and showed a relative decrease contralateral to the attended hemifield . The strong modulation during this delay is consistent with the notion that alpha band activity reflects the anticipatory allocation of attentional resources . No strong attentional modulation was observed in the intermediate beta-band or in other frequency bands . To determine the underlying cortical sources of these modulations , we used a frequency domain spatial filtering technique ( a beamformer approach [26] ) . To statistically quantify these modulations we used cluster-based permutation statistics [27] , a method controlling for multiple comparison in space ( see Materials and Methods ) . When comparing power values from “attention left” and “attention right” trials , we found robust modulations in the occipital cortex . When subjects were cued to the left , right occipital alpha power was lower than when they were cued to the right . The reverse pattern was observed in the left hemisphere ( Fig 2C ) . These differences were greatest in the superior occipital cortex ( MNI coordinates: left , −26 −92 38 , right , 34 −82 44; associated clusters: left , p = 0 . 02; right , p = 0 . 0008 , see S1 Fig ) . Conversely , when subjects were cued to the left , right occipital gamma power was higher than when they were cued to the right , and the reverse pattern was observed in the left hemisphere ( Fig 2D ) . These differences were greatest in the middle occipital cortex ( MNI coordinates: left hemisphere , −26 −94 16; right hemisphere , 34 −82 16; associated clusters: left , p = 0 . 002; right , p = 0 . 004 , see S2 Fig ) . Consistent with the literature , both anticipatory alpha oscillations and stimulus-induced gamma band activity in occipital cortex are robustly modulated by spatial attention [6–12] . Next , we sought to relate individual differences in modulations of the gamma and alpha band activity to properties of the SLF . Spherical deconvolution tractography [25 , 28] was used to reconstruct the SLF branches from the diffusion data . Consistent with previous research [21 , 23] , a network of three branches in each hemisphere was reconstructed ( Fig 3A ) . For each of the three SLF branches , a hemispheric asymmetry index was computed ( 100% ( volume_left–volume_right ) / ( volume_left + volume_right ) ; see Materials and Methods ) , quantifying whether each subject had greater tract volume in the left or right hemisphere . Nonoverlapping regions were identified as regions of interest ( ROIs ) in prefrontal cortex and then used for seeding the fiber tracking . This ensured that the fiber bundles were well separated . The medial SLF1 branches were defined as fibers passing through superior frontal gyrus , SLF2 as passing through middle frontal gyrus , and SLF3 as passing through precentral gyrus ( see Materials and Methods ) . Replicating previous findings [21] , the SLF3 was right-lateralized at the group level , whereas SLF1 and SLF2 did not show evidence of lateralization at the group level ( see Fig 3B ) . Furthermore , a modulation asymmetry index was also calculated for each subject’s MEG data indicating whether—for both alpha and gamma oscillations—that subject displayed a stronger degree of power modulation with attention in the left or right hemisphere ( ΔAMI = ( - AMIleft , j ) —AMIright , j; see Materials and Methods ) . We derived the alpha and gamma modulation values ( ΔAMI ) from the anatomical regions demonstrating strongest attentional modulation for each band , namely the superior occipital cortex for the alpha band and the middle occipital cortex for the gamma band ( see Fig 2 ) . Alpha and gamma asymmetry were not correlated with each other ( r = -0 . 148 , p = 0 . 47 ) . We then correlated alpha and gamma asymmetry with the volumetric asymmetry of the three SLF branches . Our main finding ( Fig 4A , top panel ) shows that gamma modulation asymmetry was strongly positively correlated with SLF1 hemispheric asymmetry ( r = 0 . 596 , r2 = 0 . 36 , p = 0 . 0016 , Spearman , significant at the p < 0 . 005 level after Bonferroni correction for three comparisons ) . This demonstrates that subjects who displayed relatively greater gamma modulation in the left hemisphere than in the right hemisphere also had relatively greater tract volume in the left than in the right hemisphere ( and vice versa ) . No correlation was observed with SLF2 or SLF3 ( in all cases p > 0 . 05 without Bonferroni correction ) . Our second main finding ( Fig 4B , top panel ) shows that alpha modulation asymmetry was strongly negatively correlated with SLF1 hemispheric asymmetry ( r = -0 . 503 , r2 = 0 . 25 , p = 0 . 0096 , Spearman , significant at the p < 0 . 05 level after Bonferroni correction for three comparisons ) . This means that subjects who displayed relatively greater alpha modulation in the left hemisphere than in the right hemisphere also had relatively greater tract volume in the left than in the right hemisphere . The difference in the signs of the correlation is explained by alpha power decreasing and gamma power increasing contralateral to attention ( see Materials and Methods for detailed explanation ) . No correlation was observed with SLF2 or SLF3 ( in all cases , p > 0 . 1 without Bonferroni correction ) . This is evidence that individual differences in SLF1 hemispheric asymmetry predict individual differences in the top-down modulation of neuronal synchronization in both the alpha and gamma band . To determine whether target-driven reorienting produced an asymmetry in the gamma band , we computed a reorienting index ( RI ) analogous to the AMI , according to the formula RIj = 100% * ( PowerAttention both target left , j—PowerAttention both target right , j ) / ( PowerAttention both target left , j + PowerAttention both target right , j ) , for the gamma-band data in the post-stimulus window . This did not reveal a pattern of lateralized modulation , and no correlation was observed with any SLF branch ( p > 0 . 1 without Bonferroni correction in all cases ) . Having demonstrated a link between hemispheric asymmetry of SLF1 and both anticipatory alpha and stimulus-induced gamma band modulations in visual cortex , we further tested if these effects were predictive of subjects’ task performance . Accordingly , we quantified the degree to which subjects benefitted ( in terms of reaction time and accuracy ) from a left versus a right cue in comparison to the control condition with no spatial cue ( see Materials and Methods ) . This hemifield specific asymmetry of the cueing benefit correlated with the hemispheric asymmetry of occipital gamma power modulation ( ΔAMI; Fig 5A , r = -0 . 40 , p < 0 . 05 ) but did not correlate with alpha power modulation ( Fig 5B , r = 0 . 03 , p = 0 . 89 ) . The negative correlation value means that subjects with relatively stronger gamma modulation in the left occipital cortex than in the right occipital cortex benefitted more from a right cue than a left cue . This is fully commensurate with the notion that visual cortical gamma modulation in the hemisphere contralateral to target presentation boosts effective synaptic gain and thus enhances stimulus processing . No correlation was observed between accuracy benefit and hemispheric asymmetry of occipital gamma modulation ( Fig 5C , r = 0 . 16 , p = 0 . 45 ) or alpha modulation ( Fig 5D , r = 0 . 31 p = 0 . 12 ) . Participants also performed a behavioral “landmark” task outside the MEG , designed to test spatial perceptual and motor response biases in the absence of directed attention [29–31] . Performance on this task was found not to correlate with hemispheric asymmetry of any SLF branch ( see S1 Text and S3 Fig ) . Although evidence exists for behaviorally relevant modulation of alpha and gamma oscillations in the occipital cortex [13 , 14] , there is evidence that the top-down control signals that produce these modulations originate in the frontal cortex [16 , 18] . Given that gamma oscillations likely represent a general-purpose mechanism for effective communication [32] , we further investigated whether SLF1 asymmetry predicted hemispheric asymmetry of gamma oscillations in prefrontal regions . To do this , we predefined two frontal ROIs: first , the FEF as defined by a meta-analysis of saccade studies [33] and , second , an adjacent region in the superior frontal cortex that has been identified as part of a frontoparietal network underpinning spatial attention and working memory [19 , 20] . To our surprise , hemispheric gamma modulation asymmetry ( delta AMI ) was found to correlate strongly with SLF asymmetry in the latter ROI ( Fig 6A , r = -0 . 47 , p = 0 . 017 ) . Notably , the correlations in superior frontal cortex are negative , while they are positive in the occipital cortex . Fig 6B shows statistical maps of the correlation of SLF1 asymmetry with gamma asymmetry for every grid point . Grid points in the frontal cortex show negative correlations , and grid points in the occipital cortex show positive correlations . This means that those subjects with a greater left than right SLF1 volume actually displayed relatively greater gamma modulation in the right than left superior frontal cortex . For the FEF as defined from the saccade literature , no correlation was observed with respect to hemispheric gamma modulation asymmetry ( r = 0 . 35 , p = 0 . 08 ) . Neither ROI showed a correlation with hemispheric alpha modulation asymmetry ( r = -0 . 33 , p = 0 . 097 , and r = 0 . 02 , p = 0 . 92 , respectively ) . No correlations were observed between SLF2 or SLF3 asymmetry and hemispheric alpha or gamma modulation asymmetry in the above ROIs ( p > 0 . 15 in all cases ) . Finally , we computed functional connectivity values between the superior frontal and occipital ROIs within the left and right hemispheres for each subject using power envelope correlations [34] and correlated the hemispheric asymmetry in functional connectivity with asymmetry of the SLF branches . No correlation was observed for the alpha or the gamma band data ( all p > 0 . 1 ) .
As reported in numerous studies , we have shown that stimulus-induced gamma band activity increases with spatial attention . Further , alpha oscillations decrease in anticipation of an upcoming stimulus . Importantly , we have now demonstrated a relationship between hemispheric asymmetry of the medial branch of the SLF ( SLF1 ) and individual differences in the ability to exert top-down control over both anticipatory-alpha and stimulus-induced gamma oscillations . To our knowledge , this is the first evidence demonstrating that individual differences in frontoparietal white matter tracts predict the ability to modulate occipital cortical oscillations . This is strong evidence that the SLF1 is a structural pathway mediating top-down signals that control attentional modulations in visual cortex by modulating neuronal synchronization . There is evidence suggesting that attention-modulated neuronal synchronization in the gamma band increases effective synaptic gain , and this synaptic gain increase enhances the impact of a neuronal population on connected downstream regions [1 , 35] . Crucially , the ability to modulate gamma band activity in the present study was found to be predicted by the SLF1 . Top-down signals from frontal cortex may thus serve to enhance gamma band synchronization and thus effective communication between visual cortex and downstream brain regions [2] . Emphasizing the relevance of these connections , hemispheric gamma band asymmetry was itself found to predict reaction times on the behavioral cueing task . This implies a causal chain by which a structural feature—hemispheric SLF1 asymmetry—can impact behavioral outcomes via its effect on neuronal dynamics . In contrast , no relationship was found between alpha oscillations and accuracy , in contrast to previous reports [36 , 37] . Gamma power has previously been shown to lock to the phase of ongoing alpha oscillations [38] , suggesting an intimate relationship between bottom-up drive ( indexed by the former ) and pulsed inhibition ( indexed by the latter ) . The present findings suggest that attentional modulation of alpha and gamma oscillations may not be related in such a simple fashion . The direct relationship between alpha and gamma oscillations should be a topic for future studies . The relationship between hemispheric asymmetry of tract volumes and modulation of occipital cortical oscillations warrants further investigation . We propose that larger tract volume results in a higher fidelity of the top-down signal . A larger number of top-down connections from frontal control regions could result in a stronger propagation of the top-down signal by increased signal transmission . Tract volume is likely to depend on several factors including number of axons , proportion of myelinated axons , and axonal diameter [21] . Future work should therefore focus on identifying contributions of these factors to the effect on oscillatory modulation observed in the present study . A previous HARDI study from Thiebaut de Schotten and colleagues found that SLF2 asymmetry predicted attentional task performance , whereas in the present study we found a relationship with SLF1 . This is most likely explained by differences in the tasks . Although both studies used Posner paradigms [39] , Thiebaut de Schotten and colleagues used 50% cue validity ( Thiebaut de Schotten et al . [21] , Supplemental Materials , page 12 ) . Accordingly their subjects may have adopted a more stimulus-driven strategy engaging the ventral attentional network [15] , consistent with the notion that the SLF2 supports communication between the dorsal and ventral networks [21] . The present study uses 100% valid cueing allowing preallocation of attention and likely engaging the dorsal attentional network . The present findings complement and extend these previous findings , demonstrating that in the context of high cue validity the dorsal network ( and thus SLF1 ) is more strongly implicated . The present study demonstrated that frontal top-down signals propagated via SLF1 impact visual cortical oscillations . Data from nonhuman primates implicate beta-band ( 18–34 Hz ) oscillations in the FEF as controlling shifts of covert attention [40] , and entrainment of 30 Hz activity in FEF using TMS has been shown to enhance visual perceptual sensitivity on a visual detection task in humans [41] . However , and consistent with our main hypotheses , initial sensor-level analysis of the MEG data ( Fig 2 ) rather revealed robust attentional modulation during the cue-target interval in the alpha band and during the post-stimulus period in the gamma band , consistent with previous studies [1 , 6–11] . As well as the beta band , there is also some evidence that gamma-band phase interregional synchronization between frontal and posterior cortex is modulated by direction of attention [6] , making this another candidate mechanism for top-down control . Future studies should attempt to further elucidate the precise form these attentional top-down control signals take . The sources of the modulation of anticipatory alpha and stimulus-induced gamma oscillations were identified in the occipital cortex . The degree to which this attentional modulation was stronger in one hemisphere correlated strongly with hemispheric asymmetry of SLF1 volume . Crucially , however , a region in the superior frontal cortex also showed a similar effect in the gamma band , but with the opposite sign . This means that—whereas greater SLF1 volume in the left hemisphere ( versus right ) predicted stronger attentional gamma modulation in the left occipital cortex ( versus right ) —in the superior frontal cortex , greater SLF1 volume predicted weaker ipsilateral gamma modulation as compared to contralateral . Since modulation asymmetry is a measure of interhemispheric difference in modulation , this suggests a coupling of attentional gamma modulation between frontal cortex and contralateral visual cortex . Some evidence of such contralateral connections has been seen in previous TMS studies [18 , 42] . Furthermore , besides being the hypothesized frontal terminus of SLF1 [21] , this frontal region is also adjacent to the human frontal eye field [33] , a key node in the dorsal attentional network known to be involved in top-down allocation of attention [43–45] . Notably , one TMS study explicitly demonstrated a link between the disruptive effect of TMS to the right FEF on a visual perception task and properties of the SLF1 [23] , suggesting that this white-matter tract indeed serves as the structural basis for communicating signals from FEF to other nodes in the dorsal attentional network . Whilst we demonstrate a role for a cortico-cortical connection in top-down control of attentional oscillations , it is important to also consider cortico-subcortical connections . A recent nonhuman primate study demonstrated functional and structural connectivity between pulvinar and several visual areas , with the former serving to synchronize neocortical regions during a visuospatial attention task [46] . The cortico-cortical pathway we report on should be considered complementary to the subcortical pathway . The pulvinar may drive local synchrony between visual cortical regions preferentially during attention , whilst the frontal cortex provides top-down control signals that boost or attenuate the amplitude of attentionally relevant oscillations in response to task demands . Delineation of the respective contributions of both cortico-cortical and cortico-subcortical pathways should be the object of further study . In conclusion , our data demonstrate for the first time ( as far as we are aware ) evidence for a cortico-cortical pathway providing top-down control of attentional modulations of behaviorally relevant neuronal oscillations in occipital cortex . This provides experimental support for the notion that modulation of visual cortical oscillations—and thus of effective synaptic gain—is the mechanism by which the dorsal attentional network asserts goal-directed attention .
Twenty-eight right-handed subjects ( 15 males , 13 females , mean age of 24 y and 5 mo ) participated in the experiment . All subjects underwent standard screening procedures for MEG and MRI . All experiments were carried out in accordance with the Declaration of Helsinki and following ethical approval by the local ethics board ( CMO region Arnhem-Nijmegen , CMO2001/095 ) . One subject elected not to complete the diffusion scanning , meaning diffusion data were unavailable , and for one subject SLF branches could not be reconstructed . Therefore , the analyses were conducted on the remaining 26 datasets . | Directing attention to a part of visual space produces patterns of "brainwaves" or neuronal oscillations in the human visual cortex ( the part of the brain at the back that processes incoming information from the eyes ) ; oscillations at low frequencies are believed to help the brain block out irrelevant or distracting information , whereas high-frequency oscillations signal processing of relevant information . The instructions to increase or decrease these oscillations likely originate in the front part of the brain . In this study , we investigated the structural "highways"—bundles of white matter—that connect the front and back of the brain together . Not only did we show that these highways are asymmetric—i . e . , some participants have a larger fiber bundle in the left hemisphere of their brains , and some in the right—we also showed that these asymmetries predicted whether subjects were better able to control the neuronal oscillations in their left or right hemispheres . This , in turn , predicted whether the participants were faster in detecting targets in the right or left half of the screen . Thus , we showed that these structural highways are important in helping the brain pay attention to parts of visual space . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Frontoparietal Structural Connectivity Mediates the Top-Down Control of Neuronal Synchronization Associated with Selective Attention |
Viruses have evolved strategies to protect infected cells from apoptotic clearance . We present evidence that HIV-1 possesses a mechanism to protect infected macrophages from the apoptotic effects of the death ligand TRAIL ( tumor necrosis factor–related apoptosis-inducing ligand ) . In HIV-1–infected macrophages , the viral envelope protein induced macrophage colony-stimulating factor ( M-CSF ) . This pro-survival cytokine downregulated the TRAIL receptor TRAIL-R1/DR4 and upregulated the anti-apoptotic genes Bfl-1 and Mcl-1 . Inhibition of M-CSF activity or silencing of Bfl-1 and Mcl-1 rendered infected macrophages highly susceptible to TRAIL . The anti-cancer agent Imatinib inhibited M-CSF receptor activation and restored the apoptotic sensitivity of HIV-1–infected macrophages , suggesting a novel strategy to curtail viral persistence in the macrophage reservoir .
The envelope glycoproteins of primate lentiviruses harbor domains that mediate the interaction with receptor/co-receptor proteins on the surface of susceptible cells and promote fusion between viral and cellular membranes during virus entry [1] . However , some studies suggest that activities of human lentiviral envelope proteins extend beyond their role in viral entry . For example , the envelope protein signals through receptor and co-receptor molecules after envelope binding , an activity that may alter target cell function to increase the cell's permissivity to virus infection [2 , 3] . In addition , the HIV-1 envelope , like other retroviral envelope proteins [4] , contains domains that interact with the intracellular signal transduction machinery to promote changes in cell function [5 , 6] , such as the induction of pro-inflammatory cytokines [7] . The prevention of apoptosis of the infected cell is an important modification to host cell function , particularly during chronic viral infections [8] . As a consequence , viruses have evolved measures that either confer resistance to apoptotic cell death or kill the cells delivering the apoptotic signal . Adenoviruses encode RID proteins that promote the internalization of death receptors for Fas , tumor necrosis factor–related apoptosis-inducing ligand ( TRAIL ) , and tumor necrosis factor receptor 1 ( TNFR1 ) , thereby allowing infected cells to withstand these apoptotic stimuli [9–11] . As part of a counterattack strategy , human cytomegalovirus upregulates death-inducing ligands on the infected cell , triggering apoptosis of human cytomegalovirus–specific T cells [12] . Most of the data on HIV-1 immune evasion strategies have been derived from experiments in lymphocytes . For example , the HIV-1 Nef gene has been shown to block the function of ASK 1 ( apoptosis signal-regulating kinase 1 ) in infected lymphocytes to protect these cells from Fas and TNFR-mediated apoptosis [13] . At the same time , HIV-1 Nef induces Fas ligand expression on infected cells to kill cytotoxic T cells expressing Fas on the cell surface [14] . Despite the existence of viral mechanisms to protect lymphocytes from the host apoptotic response , productively infected lymphocytes are rapidly cleared by the cytopathic effects of virus infection [15] . By comparison , the turnover of infected macrophages is slow . Extrapolation of the decay characteristics of plasma virions during highly active antiretroviral therapy suggests that the half-life of the infected macrophage reservoir in the tissues is on the order of 2–4 wk [15] . However , a greater half-life is suggested by studies with highly pathogenic SHIV ( simian immunodeficiency virus [SIV]/HIV chimera ) variants , which demonstrated no decay in the macrophage reservoir over a 3–4-mo interval when viral spread was prevented by the antiretroviral PMPA [16] . Whether HIV-1 has a mechanism to sustain the persistence of infected macrophages is unknown .
HIV-1–infected macrophages release the pro-survival cytokine M-CSF ( macrophage colony-stimulating factor ) and the β-chemokines MIP-1α and MIP-1β [17 , 18] . To characterize the biological role of M-CSF in HIV-1 replication , we initially identified the viral gene product responsible for its induction . To restrict viral replication to a single cycle , macrophages were infected with vesicular stomatitis virus ( VSV ) –pseudotyped , X4-tropic viruses containing inactivating mutations in structural and accessory genes . Levels of virus production ( extracellular reverse transcriptase [RT] activity ) and M-CSF release were monitored at different intervals after infection . While the accessory proteins Vpu , Vif , Nef , and Vpr were relatively dispensable for the induction of M-CSF ( unpublished data ) , the release of M-CSF by macrophages infected with an HIV-1 envelope-minus variant ( HIV-1LAIΔenv ) was impaired relative to macrophages infected with a wild-type virus , despite indistinguishable levels of virus production in wild-type– and Δenv-infected cultures ( Figure 1A ) . There was a statistically significant ( p = 0 . 001 ) relationship between envelope and M-CSF when peak M-CSF levels in HIV-1WT , HIV-1Δ env , and mock-infected macrophages from nine independent experiments were compared ( Figure 1B ) . In addition , during infection of macrophages with wild-type R5-tropic HIV-1 , M-CSF production was significantly induced over mock-infected cultures ( p = 0 . 003 , n = 15; Figure 1C ) . The HIV-1 envelope protein induces signals through the chemokine receptors that influence the expression of cellular genes [19 , 20] . Macrophage infection by HIV-1 is mediated primarily by the co-receptor CCR5 [1] . Since M-CSF was induced by X4-tropic viruses that are unable to use CCR5 on macrophages ( Figure 1A and 1B ) , the induction appeared to be independent of co-receptor use . Macrophages infected with a pseudotyped HIV-1 variant that harbored a mutation in the CD4 binding epitope of envelope [21] also induced M-CSF production . However , M-CSF was not induced when macrophages were pulsed with gradient-purified R5-tropic HIV-1ADA virions or with HIV-1ADA Δenv virions ( Figure S1 ) . Collectively , these results indicate that the induction of M-CSF by the viral envelope glycoprotein was independent of receptor or co-receptor interactions . In addition , de novo synthesized envelope , but not virion-associated envelope , induced M-CSF production . Since cytokines can counteract apoptotic signals [22 , 23] , we analyzed whether the differential expression of M-CSF in HIV-1 wild-type– and Δenv-infected macrophages affected the expression of apoptotic mediators in these cells . When analyzed using pathway-specific gene arrays , macrophages infected with a Δenv virus but not with a wild-type virus expressed greatly elevated mRNA levels for the TRAIL Death Receptor TRAIL-R1/DR4 ( Figure S2 ) . This reflected an increase in surface TRAIL-R1 on HIV-1 Δenv-infected macrophages compared with wild-type–infected macrophages ( 23% TRAIL-R1 positive for Δenv versus 6% for wild-type–infected macrophages; Figure 2A ) . Furthermore , in a comparison of seven independent experiments , a statistically significant difference in TRAIL-R1 expression was observed following infection of macrophages with envelope minus virus relative to wild-type– and mock-infected macrophages ( p < 0 . 001; Figure 2B ) . However , expression of death receptors for Fas ligand and TWEAK ( tumor necrosis factor–like weak inducer of apoptosis ) were not affected by HIV-1 infection ( Figure 2A ) . We next examined whether the differential expression of TRAIL-R1 on HIV-1 wild-type– and Δenv-infected macrophages conferred differential sensitivity to TRAIL . Macrophages infected with a Δenv virus were highly susceptible to TRAIL killing , as evidenced by activation of caspase 3 , a hallmark of TRAIL-dependent apoptosis ( Figure 2C ) . In contrast , caspase 3 levels in wild-type–infected macrophages in both the presence and absence of TRAIL were similar to levels in uninfected macrophages ( Figure 2C ) . TRAIL also selectively curtailed the survival of macrophages infected with a Δenv virus . In the presence of TRAIL , the half-life of macrophages infected with a Δenv virus was ∼1 . 7 d , compared with ∼4 . 6 d for wild-type virus and ∼3 . 5 d for mock-infected cells ( Figure 2D ) . In the absence of TRAIL , Δenv-infected macrophages also possessed the shortest half-life , ∼38 d , while wild-type– or mock-infected macrophages exhibited a half-life of ∼70 d and ∼100 d , respectively ( Figure 2E ) . We also determined the relative concentrations of TRAIL required to impact viability and virus output of macrophage cultures infected with wild-type and Δenv viruses . In macrophages infected with a Δenv HIV-1 variant , a TRAIL concentration of 3 ng/ml−1 produced a 50% reduction in cell viability and virus output ( Figure 2F and 2G ) . In contrast , more than 30 ng/ml−1 and 50 ng/ml−1 of TRAIL was required for a 50% reduction in viability and virus output , respectively , from wild-type–infected macrophages ( Figure 2F and 2G ) . HIV-1 envelope also rendered infected macrophages resistant to TRAIL on the surface of activated T lymphocytes . When activated CD8+ T cells ( ∼16% TRAIL positive; unpublished data ) were co-cultured with infected macrophages , virus output from HIV-1 Δenv-infected macrophages was reduced by ∼90% , but virus output from wild-type–infected macrophages was not affected ( Figure 2H ) . Collectively , these results indicate that HIV-1–infected macrophages are resistant to TRAIL and that the viral envelope glycoprotein is required for this resistance . The viral envelope glycoprotein induced the production of M-CSF in infected macrophages ( Figure 1 ) . Since cytokines can promote cell survival by opposing apoptotic signals [22 , 23] , we examined whether M-CSF was required for the resistance of wild type–infected macrophages to TRAIL . When M-CSF was neutralized in macrophage cultures infected with wild-type HIV-1 , surface TRAIL-R1 expression increased ( Figure 3A ) . In parallel , neutralization of M-CSF in macrophage cultures infected with wild-type R5-tropic HIV-1 elevated TRAIL-R1 expression ( Figure S4A–S4C ) . Conversely , addition of M-CSF to the culture downregulated TRAIL-R1 expression on macrophages infected with an envelope defective virus ( Figure 3A ) . M-CSF neutralization restored TRAIL sensitivity to macrophages infected with wild-type virus; macrophage viability was reduced by 75% and virus output by 85% when M-CSF was neutralized in these cultures ( Figure 3B and 3C ) . M-CSF neutralization rendered wild-type virus–infected macrophages as sensitive to TRAIL as macrophages infected with Δenv virus . Together , these results demonstrate that M-CSF is required for the ability of the viral envelope glycoprotein to confer TRAIL resistance to infected macrophages . We next set out to identify host factors through which the viral envelope regulated the susceptibility of infected macrophages to TRAIL . We targeted the mRNA analysis to cellular genes involved in the TRAIL apoptosis pathway . We predicted that genes that mediate the protective effect of the viral envelope would also be upregulated by M-CSF . Two genes , Bfl-1 and Mcl-1 , which inhibit mitochondrial-dependent apoptosis [24 , 25] , were upregulated in macrophages infected with wild-type HIV-1 and in mock-infected macrophages that had been stimulated with M-CSF ( Figure 4A ) . However , Bfl-1 and Mcl-1 were not upregulated in macrophages infected with a Δenv virus ( Figure 4A ) . To determine whether these anti-apoptotic genes were necessary for HIV-1 to render macrophages resistant to TRAIL , the resistance of infected macrophages to TRAIL was examined after silencing these genes by RNA interference . In macrophages , RNA interference achieved a 70%–75% knockdown of Bfl-1 and Mcl-1 proteins and mRNA when assessed by Western blotting and quantitative RT-PCR , ( reverse transcription PCR ) , respectively ( Figure 4B and 4C ) . Silencing of either Bfl-1 or Mcl-1 restored the susceptibility of wild-type–infected macrophages to TRAIL ( Figure 4D ) . Levels of caspase activation in wild-type–infected macrophages in which Bfl-1 or Mcl-1 had been silenced approached levels observed in wild-type–infected macrophages in which M-CSF had been neutralized ( Figure 4D ) . In the absence of TRAIL , neither Bfl-1/Mcl-1 silencing nor M-CSF neutralization had a significant impact on the levels of macrophage apoptosis ( Figure 4D ) . Three other host genes , cIAP-1 , cIAP-2 , and XIAP , which antagonize caspase activation [24 , 26] , were upregulated by the viral envelope during the first days of macrophage infection , prior to the induction of M-CSF release . From similar RNA interference experiments , these IAP genes conferred moderate resistance of wild-type–infected macrophages to TRAIL-mediated apoptosis ( Figure S3 and unpublished data ) . Collectively , these results demonstrate that HIV-1 envelope glycoprotein in macrophages induces M-CSF , which in turn opposes the apoptotic effects of TRAIL by inducing the anti-apoptotic proteins Bfl-1 and Mcl-1 . The aforementioned results predict that agents that interfere with the function of M-CSF or of its receptor would antagonize HIV-1′s ability to protect infected macrophages from TRAIL-mediated killing . The anti-cancer drug Imatinib inhibits the tyrosine kinase activity of the bcr-abl oncogene and also shows considerable selectivity and efficacy toward the intrinsic tyrosine kinase activity of the M-CSF receptor [27] . Therefore , we examined Imatinib for its potential to kill HIV-1–infected macrophages . The M-CSF receptor undergoes autophosphorylation after ligand binding [27] . At concentrations obtainable in vivo , Imatinib inhibited receptor phosphorylation in primary human macrophages ( Figure 5A ) , indicating that Imatinib blocked signaling by the M-CSF receptor in human macrophages . Furthermore , while TRAIL-R1 expression in wild-type virus–infected macrophages was similar to uninfected macrophages , addition of Imatinib increased TRAIL-R1 expression to levels similar to that of Δenv-infected macrophages ( Figure 5B and 5C ) . The elevation of TRAIL-R1 expression on wild-type–infected macrophages by Imatinib was consistently observed on cells from three donors ( p < 0 . 001 ) , and Imatinib did not alter TRAIL-R1 levels on macrophages infected with an envelope-minus variant ( Figure 5C ) . Furthermore , Imatinib had no effect on TRAIL-R1 expression in uninfected macrophages ( Figure 5B and 5C ) . Imatinib restored the sensitivity of wild-type HIV-1–infected macrophages to TRAIL and augmented TRAIL susceptibility beyond that of Δenv-infected macrophages . Using Annexin V and propidium iodide to stain for apoptotic cells , the addition of Imatinib followed by TRAIL caused HIV-1–infected macrophages to undergo apoptosis ( Figure 5D ) . In the absence of stimuli , infected macrophages underwent a low level of apoptosis , which was not enhanced by TRAIL . Importantly , Imatinib , TRAIL , or their combination did not promote TRAIL-R1 expression or apoptosis in uninfected macrophages present in the cultures ( Figure 5D ) . Analysis of apoptosis in HIV-1–infected macrophages by DNA fragmentation further confirmed that Imatinib negated the resistance of wild-type–infected macrophages to TRAIL and facilitated extensive apoptosis , equivalent to macrophages infected with a Δenv virus and exposed to TRAIL ( Figure 5E ) . Macrophages infected with an R5-tropic wild-type HIV-1 were similarly resistant to TRAIL-mediated apoptosis . Imatinib likewise rendered these cells highly sensitive to apoptosis by TRAIL ( Figure S4D ) . Imatinib alone was sufficient to counteract the survival signals that keep HIV-1–infected macrophages alive , and long-term exposure to Imatinib resulted in apoptosis of wild-type–infected macrophages even in the absence of TRAIL ( Figure 6 ) . This finding was reflected by a decrease in Mcl-1 expression and a further increase in pro-apoptotic protein expression ( unpublished data ) . Overall , these data suggest that Imatinib may be an effective strategy to induce the TRAIL-dependent apoptotic death of HIV-1–infected macrophages in vivo . Our results also raise the intriguing possibility that Imatinib may be able to promote the death of infected macrophages without the need for TRAIL .
In this study , we present evidence for a novel envelope-dependent mechanism that allows HIV-1–infected macrophages to persist in the face of apoptotic clearance processes . Furthermore , we describe a chemotherapeutic agent , Imatinib , which has the potential to disable this viral defense . Figure 7 summarizes the mechanisms employed by the HIV-1 envelope glycoprotein to subvert the host apoptotic response to TRAIL in infected macrophages , as well as those processes countered by Imatinib to restore TRAIL sensitivity and induce apoptosis . Apoptosis by TRAIL can be propagated by signals via intrinsic or extrinsic pathways [24] . However , TRAIL-mediated apoptosis in macrophages was dependent upon the extrinsic or mitochondrial pathway of apoptosis . Despite the HIV-1 envelope-dependent upregulation of IAP family proteins ( which can block the extrinsic pathway ) early in infection , signals from the TRAIL receptor that promoted apoptosis through the mitochondrial pathway were dominant in HIV-infected macrophages . This finding agrees with Zheng et al . [28] , who report that human primary monocytes resist TRAIL-mediated apoptosis through Bcl-2–dependent ( mitochondrial ) mechanisms . The exact molecular basis for the regulation of pro-survival and anti-apoptotic factors by HIV-1 envelope is under investigation through mutagenesis studies and analysis of signal transduction pathways influenced by envelope expression . The dissemination of HIV-1 from infected macrophages to neighboring T lymphocytes requires intimate cell–cell contact , which can leave infected macrophages susceptible to apoptosis by death ligands on lymphocytes . TRAIL cytotoxicity is mediated effectively by CD4+ T lymphocytes and by CD8+ T cells and NK cells [29 , 30] . In vivo , macrophages can undergo apoptosis after activating CD4+ T cells . The induction of macrophage apoptosis by TRAIL on CD4+ T cells has been proposed to form part of macrophage homeostasis during antigen presentation [4] . The regulation of activated macrophages by TRAIL also serves to limit immune responses and to target macrophages infected with intracellular organisms for elimination [31] . For these reasons , the avoidance of apoptosis by TRAIL is likely an important process for the survival of HIV-1–infected macrophages in vivo . Macrophages infected by HIV-1 did not express receptors for other death ligands; neutralizing TRAIL—but not the Fas ligand or TWEAK—on T lymphocytes prevented the apoptosis of infected macrophages when the protective functions of the viral envelope were disabled . These findings suggest TRAIL is a major restriction for the persistence of HIV-1 in macrophages during interactions with immune cells . Mechanisms that restrict TRAIL-mediated apoptosis have been described for several unrelated viruses . Gamma-herpesviruses , including Kaposi sarcoma-associated human herpesvirus-8 , encode FLICE-inhibitory proteins ( FLIPs ) that interact with the adaptor protein FADD to inhibit the generation of active caspase 8 , which is necessary to trigger apoptosis by TRAIL [32] . Human adenovirus type-5 encodes three proteins ( RID ) that induce the internalization from the cell surface and lysosomal degradation of TRAIL receptors [10] . Human T cell leukemia virus type 1–infected T cell lines are also resistant to TRAIL-mediated apoptosis , presumably because of activation of TRAIL expression by the viral transactivator Tax [33] . Apoptosis of B cells by TRAIL is inhibited by the Epstein-Barr virus–encoded BHRF1 protein , which functions in the apoptotic pathway similarly to the host protein Bfl-1 that is induced by M-CSF in HIV-1–infected macrophages [34 , 35] . TRAIL-R1 is specifically downregulated in cells infected by human herpesvirus 7 and is associated with their resistance to TRAIL-mediated cytotoxicity [36] . Our demonstration that HIV-1 envelope glycoprotein also counteracts TRAIL-mediated apoptosis underscores the general importance of evading the immune pressure exerted by TRAIL in order for viruses to persist in the infected cell .
Neutralizing antibody to human M-CSF and isotype control antibody were used at 10 μg/ml−1 and obtained from R & D Systems . ELISA kits for human M-CSF , recombinant soluble TRAIL-R1 , and control receptor proteins were also supplied by R & D Systems . HSA ( murine CD24 ) , Fas and TWEAK-R antibodies for flow cytometry , and antibodies to CD3 and CD28 were purchased from BD Pharmingen . TRAIL-R1/DR4 antibody and soluble human recombinant TRAIL were obtained from Axxora LLC . Monoclonal and polyclonal antibodies to the M-CSF receptor and antibodies to phosphotyrosine and tubulin were purchased from Santa Cruz Biotech . Antibodies to Bfl-1 and Mcl-1 were supplied by Cell Signaling Technologies . Imatinib mesylate was obtained from Sequoia Research Products . The R5 tropic virus HIV-1ADA was prepared as detailed previously [18 , 37] . For the preparation of HIV-1 X4-tropic LAI and HSA viruses pseudotyped with the VSV envelope , 293 T cells were co-transfected with 25 μg of HIV-1 DNA and 25 μg of a VSV-G expression vector by Calcium Phosphate precipitation . Viruses containing supernatants were harvested 60 h post-transfection and standardized by RT assay as described previously [18 , 37] . HIV-1HSA [38] contains the mouse gene CD24 , a heat stable surface antigen , in place of Vpr . As a result , HIV-1HSA–infected cells can be identified by flow cytometry , and pseudotyping with VSV-G envelope effectively bypasses the requirement for Vpr in macrophage infection . Envelope-minus HIV-1LAI and HIV-1HSA variants were constructed by Nde I restriction site fill-in . HIV-1LAIΔCD4b contains a two amino acid deletion in a critical CD4-receptor binding domain of HIV-1 envelope that disrupts CD4 binding [21] . Lymphocytes and monocytes were obtained by leukapheresis from normal donors seronegative for HIV-1 and hepatitis B . Populations of CD8+ T lymphocytes were obtained by additional purification with antibody-coated magnetic beads according to manufacturer's instructions ( Dynal-Invitrogen ) . The lymphocytes were activated with antibodies to CD3 and CD28 ( 5 μg/ml−1 each ) for 48 h and washed in medium prior to co-culture experiments with macrophages . Monocytes were further separated by countercurrent centrifugal elutriation as detailed elsewhere [39] . Elutriated monocytes were differentiated from macrophages by culture for 4 d in medium containing M-CSF ( R & D Systems ) and for a further 3 d in medium without M-CSF . Macrophages were then used for virus infections within 1–5 d . ELISAs for determining apoptosis by the cleavage of caspase 3 ( to active form ) were obtained from Cell Signaling Technologies . Briefly , 1 . 4 × 106 macrophages infected with wild-type and envelope-minus viruses were lysed according to the supplier's protocol 1 h after the addition of soluble recombinant TRAIL ( 100 ng/ml−1 ) . Protein content was determined by Bradford Assay ( Bio-Rad ) , and 220 μg of protein in 200 μl from each culture was assayed in ELISA . Reagents were purchased from BioVision and used according to the supplier's protocol for the measurement of apoptosis by Annexin V and propidium iodide staining in flow cytometry . Apoptotic DNA fragmentation was measured by ELISA for histone-associated DNA fragments present in macrophage lysates prepared according to the manufacturer's instructions ( Roche ) 16 h after exposure to apoptotic stimuli . For the determination of macrophage longevity , cell death was quantitated by ELISA for the release of histone-associated fragmented DNA into macrophage culture supernatants ( Roche ) . Briefly , after the addition of soluble TRAIL to 0 . 7 × 106 cells , 10 μl of supernatant was harvested over time for ELISA . Supernatant from macrophages treated for 72 h with 100 μM Apoptosis Activator I ( EMD Biosciences ) , which resulted in comprehensive visual cell death , was used as a positive control and as a value for remaining viable cells calculated by subtraction . In other experiments , an MTT assay ( Sigma ) was used to measure cell viability . Infected macrophages were monitored by RT assay until viral replication approached peak . The medium was changed , and anti-CD3/CD28 stimulated autologous CD8+ T cells ( 2 . 5 × 106 ) after incubation with soluble TRAIL-R1 or control receptor for 1 h were co-cultured with infected macrophages for 4 h . The CD8 cells were then gently removed and virus production from macrophages determined by RT assay 24 h later . Macrophages , 1 . 4 × 106 , were infected with wild-type HIV-1 LAI and a Δenv variant and monitored by RT assay until viral replication approached peak levels . Macrophages were transfected with Lipofectamine 2000 containing 100 nM duplexed small interfering RNA ( siRNA ) to Bfl-1 , Mcl-1 , or scrambled siRNA ( Dharmacon ) for 4 h , then re-fed conditioned medium from uninfected macrophages . The transfection was repeated the next day . Cells were harvested for ELISA or RNA analysis 16 h after the second transfection . RNA interference ( RNAi ) –mediated mRNA decay was assessed on total RNA from 200 , 000 cells prepared by Trizol ( Invitrogen ) in SyBr Green real time RT-PCR ( Quantitect SyBr Green Kit; Qiagen ) using gene-specific primers from SuperArray . The levels of cellular mRNAs were similarly determined by SyBr Green real-time PCR or by ribonuclease protection assay ( BD Pharmingen ) , as described previously [37] . The statistical significance of data , where indicated , was determined by ANOVA or t-test using Prism 5 ( GraphPad Software ) . Mean values ± SEM are shown graphically; annotations indicate the confidence level ( p-value ) and number of replicates ( n ) . | Much of our understanding regarding mechanisms of HIV-1 persistence has been derived from studies with lymphocytes . However , mechanisms governing persistent infection of macrophages are less well understood . We investigated whether HIV-1 modulates macrophage function in a way that promotes their persist infection . We focused on a cytokine called macrophage colony-stimulating factor ( M-CSF ) , because this pro-survival factor is induced upon infection by HIV-1 . We found that the viral envelope gene was necessary for M-CSF induction of macrophages . M-CSF upregulated anti-apoptotic genes in macrophages and restricted the expression of the death receptor ( tumor necrosis factor–related apoptosis-inducing ligand [TRAIL]-R1 ) . As a consequence , HIV-1–infected macrophages were resistant to the apoptotic effects of TRAIL . If M-CSF was blocked by antibody or if the anti-apoptotic genes were silenced by RNA interference , the apoptotic sensitivity of HIV-1–infected macrophages was restored . Also , the anti-cancer drug Imatinib , which impairs activation of the M-CSF receptor , promoted the death of HIV-1–infected macrophages but not of uninfected macrophages . We believe that HIV-1 regulates M-CSF to extend macrophage survival and promote viral persistence in the host . Agents that interfere with M-CSF signaling , such as Imatinib , warrant further examination for activity against macrophage reservoirs in vivo . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases",
"cell",
"biology",
"macrophage",
"virology",
"in",
"vitro",
"hiv-1"
] | 2007 | Apoptotic Killing of HIV-1–Infected Macrophages Is Subverted by the Viral Envelope Glycoprotein |
The space-filling fractal network in the human lung creates a remarkable distribution system for gas exchange . Landmark studies have illuminated how the fractal network guarantees minimum energy dissipation , slows air down with minimum hardware , maximizes the gas- exchange surface area , and creates respiratory flexibility between rest and exercise . In this paper , we investigate how the fractal architecture affects oxygen transport and exchange under varying physiological conditions , with respect to performance metrics not previously studied . We present a renormalization treatment of the diffusion-reaction equation which describes how oxygen concentrations drop in the airways as oxygen crosses the alveolar membrane system . The treatment predicts oxygen currents across the lung at different levels of exercise which agree with measured values within a few percent . The results exhibit wide-ranging adaptation to changing process parameters , including maximum oxygen uptake rate at minimum alveolar membrane permeability , the ability to rapidly switch from a low oxygen uptake rate at rest to high rates at exercise , and the ability to maintain a constant oxygen uptake rate in the event of a change in permeability or surface area . We show that alternative , less than space-filling architectures perform sub-optimally and that optimal performance of the space-filling architecture results from a competition between underexploration and overexploration of the surface by oxygen molecules .
On its way from the trachea to blood in the lung , oxygen ( O2 ) travels through 14 generations of branching ducts forming the bronchial airways; 9 generations of ducts forming the acinar airways , which end in 300 million alveoli; and across the thin walls separating alveoli and blood capillaries ( Fig . 1; [1] , [2] ) . This architecture is a space-filling , fractal network at two levels ( Fig . 1; [1]–[4] ) , each creating a remarkable distribution system: The space-filling bronchial tree , in which transport is by convection , guarantees minimum dissipation ( pressure-driven flow [5]–[7] ) , including the 3/4 power law for metabolic rates [8] , and slows air down with a minimum number of ducts [4] . In the space-filling acinar tree , in which transport occurs primarily by diffusion , the network maximizes the gas-exchange surface area [2] , creates respiratory flexibility between rest and exercise [9] , and minimizes dissipation , too ( diffusion-driven flow [10] ) . Here we investigate performance metrics ( “engineering targets” ) of the acinar airways , the fundamental gas exchange system , that have not been studied previously . The metrics will be in terms of diffusive transport , local oxygen currents , and total oxygen currents . They will uncover an unexpected coexistence of new , seemingly mutually exclusive , optimum performance characteristics of the lung . A broad class of structure-function relations for diffusion of molecules to and across biological membranes , under various source and receptor geometries , including associated optimum architectures , has been reviewed in [4] . As air moves through successive branchings of the bronchial and acinar tree , its convective flow velocity decreases as the total cross-section area of the ducts increases . At branching generations , the flow velocity equals the diffusion velocity . After this point , although convective flow still exists , the primary transportation mechanism for oxygen becomes diffusion ( transition from convection to diffusion; Fig . 1; [9] ) . The value of depends on the flow velocity in the trachea , i . e . , on the breathing rate . The air at branching generation acts as a constant O2 source , from which O2 diffuses to alveolar membranes downstream , , and crosses the membrane into the blood ( perfusion ) . Most of the O2 exchange occurs downstream from , where O2 concentration gradients develop as progressively more O2 molecules cross the alveolar membrane . The acinus segment defines a gas exchanger . Each gas exchanger forms a space-filling network , and so does the system of all gas exchangers . The elementary building blocks of the network , small or large , are the alveoli . They manifestly span a 3-dimensional surface ( Fig . 1 , photographic inset ) . The drop in O2 concentration far from the source is referred to as screening [9] , and the oxygen current depends on the degree of screening present in different regions of the gas exchange system . There are two basic mechanisms how screening can vary . One is that the membrane permeability to oxygen , W ( length per time ) , may vary , such as under various disease conditions . If W is large ( small ) , a molecule crosses after few ( many ) collisions with the membrane , the O2 concentration gradient is large ( small ) , and regions away from the source are screened ( unscreened , [9] ) . The distance the molecule travels along the membrane ( acinar ducts ) before transfer occurs , is given by and called exploration length ( [4] , [11]; also see Methods section ) , with the O2 diffusion coefficient in air ( area per time ) . Thus a large permeability generates strong screening and a small exploration length , and vice versa . Under normal conditions , ( Table 1 ) . The second mechanism is that the transition from convection to diffusion , , varies with the breathing rate: As ventilation increases under the four levels of exercise considered in this paper—rest and moderate/heavy/maximum exercise ( see Methods section for the definition of the levels of exercise ) —the flow velocity in the trachea increases and pushes the O2 source , , deeper into the acinar tree . The two mechanisms reduce the question of performance of the lung as gas exchange system to the question , what are the relevant length scales downstream of at different levels of exercise ? Are they large or small compared to 33 cm ? Is the O2 current diffusion/access-limited ( small Λ ) or reaction/transfer-limited ( large Λ ) ?
The space-filling bronchial airways and space-filling acinar airways are different morphologically ( Fig . 1 ) . For the bronchial airways , the duct diameter after z branching generations , dz , is well-described by Murray's law , dz = 2−z/3d0 , for z = 0 , … , 14 [1] , and generates a tree whose canopy—the collection of all branch tips , has fractal dimension 3 [4] . For the acinar airways , z = 15 , … , 23 , the diameters remain nearly constant . The ducts and alveoli together span a network that is space-filling , with fractal dimension 3 , not just in terms of branch tips , but as a whole , including the ducts ( Fig . 11 . 6 in [1] ) . In this paper we investigate performance metrics of the acinar airways , through whose surface the gas exchange occurs . We model the O2 transfer from air to blood as a three-step process—diffusion through the acinar airways , diffusion across the membrane , and diffusion and binding to red blood cells [1] , [9]—and write the O2 current , I ( number of molecules transferred from air to blood per time ) , as ( 1a ) ( 1b ) Here , the membrane includes the alveolar tissue barrier and the plasma ( Fig . 12 . 7 in [1] ) . The driving force for the O2 transfer from air to erythrocyte is the partial pressure difference across the membrane and erythrocyte , where pa and pe is the partial pressure of O2 in air adjacent to the membrane and in the erythrocyte , respectively , both averaged over the whole lung ( Fig . 1 ) . In Equation 1a , we express the current in terms of ( total barrier: membrane and erythrocyte ) . Since the current across a series of resisters equals the current across any individual resister , we further express the current in terms of partial pressure difference across the membrane alone , . The inset of Fig . 1 shows the pressure profile across the membrane , with pressure pb on the erythrocyte side of the membrane . The respective proportionality factors define the lung and membrane diffusing capacity , Tlung and Tm ( number of molecules per time and pressure ) , used to determine the gas exchange capacity of the lung [1] , [12] . Equation 1b expresses the membrane diffusing capacity in terms of , the diffusion coefficient of O2 in the membrane ( area per time ) , the solubility of O2 in the membrane ( number of molecules per volume and pressure ) , the membrane thickness τ ( including the thickness of alveolar tissue and the plasma , Table 1 , [1] , [2] ) , and the surface area across which oxygen effectively diffuses ( Fick's law; effective surface area is defined in Equation 4 ) . To convert partial pressure difference to concentration difference , we used solubility β , which is the quantity of oxygen that becomes dissolved in a unit barrier volume if the partial pressure is raised by one unit , [1] . The quantity is the permeability of the membrane , defined as the number of O2 molecules crossing the membrane per unit surface area , unit time , and unit concentration difference between the two sides of the membrane , if both were air . Thus , the first part of Equation 1b describes the gas exchange as a three-phase system ( air , membrane , erythrocyte ) , and the second part reduces the description to two phases , involving the permeability , which refers only to the membrane and air , and the rescaled concentration difference , due to the fact that the medium on the far side of the membrane is not air , but erythrocyte . Under steady-state conditions , the oxygen concentration , at position x in the diffusion space , i . e . , acinar airways , obeys the stationary diffusion equation ( Laplace equation ) with appropriate boundary conditions: ( 2a ) ( 2b ) ( 2c ) The boundary condition 2b reflects that the gas exchanger , in which oxygen is transported by diffusion , is only a fraction of an acinus ( Fig . 1 ) , and that the air at the entrance has uniform oxygen concentration , . The boundary condition 2c states that the bulk diffusion flux in air in the direction normal to the surface , , is equal to the transmembrane flux ( current conservation ) . The condition accounts for screening , i . e . , that the oxygen concentration at the alveolar surface , far away from the entrance , may be much smaller than at the entrance , , and highlights the importance of the ratio , which has units of length . If this length is small , by virtue of the permeability being large , O2 is extracted quickly as it moves downstream from the entrance , and O2 visits only a small portion of the acinus before it crosses the membrane . Conversely , if this length is large , i . e . , if the permeability is small , O2 travels a long distance until it crosses the membrane . Thus is the length of a typical diffusion path of an O2 molecule before it crosses the membrane , measured along the alveolar surface . Whence the term ‘exploration length’ for Λ . We treat W , and accordingly Λ , as a variable , and investigate the performance of the gas exchanger by analyzing the effective surface area and oxygen current as a function of W and Λ beyond the physiological range . From the solution of the boundary-value problem , c ( x ) , we compute the current as ( 3 ) ( 4 ) where S is surface area; Ng is the number of gas exchangers in the lung; oxygen concentrations , solubilities , and partial pressures are as in Fig . 1; and Equation 4 expresses the integrated surface concentration , Equation 3 , in terms of the concentration at the source , , and the effective surface area of the gas exchanger , , which is the area that would carry the same current in the absence of screening [4] , [9] , [13] . Equation 4 reduces the question of the oxygen current across the lung to the knowledge of the solubility of oxygen in air , partial pressure drop of oxygen across the membrane , number of gas exchangers , and effective surface area of a single gas exchanger . A renormalization treatment decomposes the surface into regions completely accessible and completely inaccessible to O2 molecules and calculates as area of the accessible region [4] , [11] , [13] , [14] . For a membrane surface with fractal dimension and source small compared to the gas exchanger , this is illustrated in Fig . 2 and leads to four regimes , where are the total surface area of the exchanger , cross-section area of the source , and alveolar side length . This gives as a series of power laws of the exploration length , controlled by the fractal dimension , and identifies the relevant length scales downstream of . The four regimes and associated length scales follow from the four geometric situations depicted in Fig . 2 . In Equation 5a , the exploration length is less than the size of an alveolus , and only the region facing the source , with area Ss and depth less than an alveolus , contributes to the current . This is the case of complete screening . In Equations 5b and 5c , the exploration length is long enough that incoming molecules enter the hierarchy of small and large fjords of the fractal surface , but not long enough for the molecules to visit the entire surface . This is the partial screening regime . In Equation 5b the molecules visit a region one or several alveoli deep , but still shallow compared to the lateral size of the source , In Equation 5c , the depth of the visited region exceeds the lateral size of the source , and so does the lateral size of the visited region . In Equation 5d , the exploration length exceeds the perimeter of a planar cross section of the surface , and the molecules visit the entire surface before they cross the membrane . This is the case of no screening . Four structural parameters of the human lung are needed to calculate oxygen currents from Equations . 4 and 5: the alveolar side length , ; the cross-section area of the oxygen source , ; the surface area of the gas exchanger , ; and the number of gas exchangers , . The values of , and vary with zcd ( the values of zcd as a function of the breathing rate are obtained below ) . We consider the gas exchanger as a cubic array of cubes of side length [9] , [12] , in which each cube contributes four of its six faces as surface for gas exchange . So , where Vacinus = 0 . 187 cm3 and Sacinus = 54 cm2 are the acinus volume and acinus surface area , respectively [12] . This gives . For the cross-section area of the source , we take ( square cross section ) , evaluation of which with data from [12] gives the values listed in Table 2 . The cubic gas exchanger model is for convenience of calculation; a cylindrical/spherical model for the airways would give similar results . The surface area of the gas exchanger is obtained from , which reflects that the acinus comprises branching generations z = 15 , … , 23 [12] . The resulting values are shown in Table 2 . The number of gas exchangers is , where is the total surface area of the lung , taken as the arithmetic mean of the total alveolar surface area and total capillary surface area , following [2] . The value is obtained from data in [12] , and the resulting values for Ng are listed in Table 2 . The partial pressure differences across the membrane in Table 2 were obtained from , with experimental data discussed below . The membrane permeability in Table 1 is from Equation 1b and is evaluated using the listed values of diffusion coefficients , solubilities , and membrane thickness in Table 1 . For the sake of comparison , we also list the values of diffusion coefficients and solubilities of O2 , CO , and CO2 in water and air in Table 3 . The structural parameters , , and vary with the level of exercise because the branching generation , at which convection changes to diffusion , does so . Several authors [9] , [12] proposed a concept—“acinus Peclet number” Pa—to calculate . At branching generation z , the distance to travel to the end of the airway is of order , where λ is the mean length of an acinar duct and zmax is the total number of the branching generations ( λ≈1 mm [12]; zmax = 23 ) . So the mean diffusion velocity is , where is the O2 diffusion coefficient in air . The acinus Peclet number is then defined as the ratio of convection velocity to diffusion velocity , , where is the flow velocity at generation z ( volume of air per unit time , divided by the cross-section area of the bronchial or acinar duct ) , which depends on the breathing rate . The convection-diffusion transition occurs when Pa = 1 , and can be calculated ( rounded to the nearest integer ) from ( 6 ) Sapoval et al . [9] , [12] analyzed the cases of rest and maximum exercise and obtained zcd = 18 , 21 , respectively , for the two cases . To include intermediate breathing regimes , zcd = 19 , 20 , and revisit the cases zcd = 18 , 21 , we proceed as described in the next paragraph . This naturally leads to four distinct breathing regimes or levels of exercise , for which we assemble the experimental oxygen currents ( for comparison with computed currents , Equations 4 , 5 ) and membrane diffusing capacities ( for partial pressure differences , ) . The discrete breathing regimes result from our focus on discrete gas exchange units— zcd = 18 , 19 , 20 , 21 , leading to gas exchangers equal to 1/8 , 1/16 , 1/32 , 1/64 of an acinus ( dichotomous branching ) , The discrete screening regimes arise from geometric changes of the region incoming oxygen molecules visit before they cross the membrane . The oxygen current , expressed as volume of O2 at STP per unit time ( uptake rate ) , for a normal , healthy adult male at rest is about 270 ml/min [1] , [12] , [15]–[16] . The current at maximum exercise for well-trained athletes , which defines the maximum oxygen current a human lung can achieve , is about 5500 ml/min [17] . To obtain currents for the intermediate cases , zcd = 19 , 20 , we calculated U ( z = 19 , 20 ) from Eq . 6 , using 0 . 0363 cm ( z = 19 ) and 0 . 028 cm ( z = 20 ) ( data from [12] . We then used mass conservation ( equation of continuity at constant fluid density ) to estimate the volumetric flow rate in the trachea for z = 19 and 20 [4] . For zcd = 19 , 20 , the respective volumetric flow rates of air in the trachea are 30400 ml/min and 70760 ml/min , or less . At these two levels of volumetric flow rates , there are two studies reporting oxygen currents and diffusing capacities . For heavy exercise , Weibel ( Table 10 . 2 in [1] ) reported volumetric flow rates in the trachea , 68100 ml/min , which is very close to 70760 ml/min . So we associate heavy exercise with zcd = 20 . The corresponding oxygen current at this level is 2420 ml/min , and the diffusing capacity for oxygen of the whole lung is about 100 ml/min/Torr . For ( “moderate exercise” ) , we take the linear regression relation between oxygen current and volumetric flow rates in the trachea reported by Newstead [18] to calculate the oxygen current , and found it to be 1360 ml/min when the volumetric flow rate is 30400 ml/min . Borland et al . [19] measured the membrane diffusing capacity att an oxygen current of 1300 ml/min , which is very close to 1360 ml/min . We therefore associate Borland's oxygen current and membrane diffusing capacity with zcd = 19 . Conversion of currents into mol/s , conversion of lung diffusing capacities into membrane diffusing capacities , and averaging of multiple measurements , gives the values reported in Table 2 for zcd = 18 , 19 , 20 , 21 .
Figure 2 displays the screening regimes , length scales , and O2 current for a single gas exchanger at rest . It shows that only for is the surface unscreened and the effective surface area equal to the total surface area of the exchanger , ; for all larger permeabilities , we have . The most striking feature is the extended horizontal plateau . If W increases beyond the effective surface area drops , and the drop exactly cancels the increase in W , Equations 5c , leading to a constant current over more than three decades of W . The constant current provides protection against loss of permeability under environmentally adverse or disease conditions . The current remains stable even if the permeability drops from its normal value , , by a factor of seven . We refer to this robustness , , as maximum fault tolerance . It is similar to a “constant-current source” in an electric circuit , designed to deliver a constant current under variable load ( here , surface resistance ) . In the lung , it autonomously results from the hierarchical tree structure , without any feedback loop , and has not been observed before . Horizontal plateaus have been noted in other branched structures [20] . We computed O2 currents at four levels of exercise ( Table 2 ) . They are compared with experimental currents and other benchmarks in Figs . 3 and 4 . A rich spectrum of results and concepts emerge—number of gas exchangers , current across each exchanger , effective surface area in each exchanger , pressure difference —to monitor how the lung ramps up the current: Computed and experimental currents agree within a factor of 0 . 9–1 . 2 ( Fig . 3A ) . To the best of our knowledge , this is the first time currents are computed from first principles . The agreement is remarkable because the computed currents come from a model of gas exchange with minimum number of structural parameters and transport parameters . The model requires no data on branching ( number/width/length of daughter ducts ) , shape of alveoli , or membrane epithelia , and no computation of concentration maps , , in the airways . Increased ventilation under increasing exercise pushes the transition from convection to diffusion , , to more distant branchings [9] , which reduces the size of gas exchangers , , but increases their number , . The current across an individual gas exchanger remains approximately constant , , and so does the effective surface area , ( Equations . 4 , 5c , Table 1 and 2 ) . Thus , increased ventilation transforms 180 , 000 exchangers with 1 . 4 nmol/s per exchanger into 1 , 500 , 000 exchangers with 2 . 6 nmol/s per exchanger . This shifts the plateaus in Fig . 4 upward and increases the current , mostly by the increase in . It shows that the current increase occurs under strongly variable conditions in the bronchial airways ( convective air flow ) and capillaries ( blood flow ) , but only weakly varying conditions in individual gas exchangers , making the acinar airways a self-regulated system , largely decoupled from the dynamics of the input and output system . A figure of merit stripped of the dependence on is the pulmonary efficiency , defined as the ratio of effective to total surface area , Equation 7a , and measured as the ratio of experimental membrane diffusing capacity , , to morphometric diffusing capacity , [1] , [2] , [12] , Equation 4c: ( 7a , b , c ) where is the total surface area of the lung ( Table 2 ) . Our efficiencies from Eq . 7a are 13% , 26% , 49% , and 94% at rest and moderate/heavy/maximum exercise; the experimental values , from Equation 7c , are 11% , 26% , 40% , and 100% . The agreement is excellent ( Fig . 3B ) . The values also agree well with earlier calculations , 10–40% at rest , and 100% at maximum exercise [4] , [9] , [12] , [20]–[22] . The efficiency increases with increasing exercise because increased ventilation reduces screening—according to Eq . 7a by the decrease in ( 1/8 , … , 1/64 acinus ) ; according to Eq . 7b by the increase in —both at constant and constant lung volume . The tenfold increase in efficiency allows the lung to increase the O2 current by a factor of 20 with only a twofold increase in pressure difference across the membrane , Table 2 ) . So the area screened at low efficiency acts as “spare area” and is the principal source of the current increase . It transforms 180 , 000 screened gas exchangers with per exchanger into 1 , 500 , 000 unscreened exchangers with per exchanger . In Fig . 4 , which plots the oxygen current as a function of membrane permeability for all four levels of exercise , we now focus on currents at fixed permeability and variable exercise . The line intersects all four plateaus and runs almost through the “knee point” at maximum exercise . This optimizes switching from rest to exercise in extraordinary ways: If the lung operated in region C , it would have maximum fault tolerance over a maximum interval of permeabilities at maximum exercise , but waste more than 99% of its surface area due to massive screening ( Equations 5b , 5c , 7a ) . If the lung operated in region A , no waste would occur , but all fault tolerance would be lost and increased ventilation would not increase the current . Only in region B are four major engineering targets met—maximum current at minimum permeability ( knee point at maximum exercise ) ; maximum current increase from rest to exercise ( maximum response to increased ventilation ) ; no waste of resources ( surface area ) at maximum exercise; and maximum fault tolerance over a broad interval of lower than normal permeabilities at rest ( for ) . This should be contrasted with the expectation that a maximum current would require maximum permeability , all other parameters held constant; that a large current increase might require a large increase in partial pressure difference across the membrane or an increase in total surface area of the lung; that a maximum current might require overdesign of resources; or that a drop in permeability would inevitably lead to a drop in current . The location of near the knee point identifies a second notable property: far to the right of the knee , diffusing O2 molecules explore only a small part of the membrane surface ( “underexploration , ” short residence time of O2 in the gas exchanger ) ; far to the left , molecules explore the surface multiple times before crossing the membrane ( “overexploration , ” long residence time ) ; at the knee point , molecules explore the surface essentially once . Thus the knee , at maximum exercise , is the unique point at which O2 molecules visit the entire surface ( maximum exploration ) at minimum residence time . This optimizes transport at the level of microscopic dynamics . If , alternatively , the knee were further to the right ( as it would if ) , the lung could generate a current larger than 4 mmol/s if . The fact that nature has not selected this option suggests that it is more important to keep the permeability low , to maintain a strong barrier against intruders and fluid effusion from capillaries [18] , than to generate larger currents . In Fig . 5 , we ask what if the gas exchangers were less than space-filling , . At , as the fractal dimension drops from 3 . 0 to 2 . 5 to 2 . 0 ( flat surface ) , the current at maximum exercise would drop from 4 mmol/s to 0 . 6 mmol/s to 0 . 09 mmol/s; the current increase , from rest to exercise , would drop from 20- to 3- to 2-fold and all fault tolerance would be lost . Thus none of the achievements in region B of Fig . 4 could be realized with . E . g . , to achieve a current comparable to , but with a 2 . 5-dimensional surface , would require a membrane permeability of . Decreasing Df shifts the curves to the right and requires increasing permeabilities to achieve currents identical those at Df = 3 . The reason is that , at fixed side length of the gas exchanger , the surface area of the exchanger drops with decreasing Df so that a large W has to compensate for a small Sg ( Equations 5c , d ) . In Fig . 6 , we compare the renormalization method ( RM ) , Equation 5 , with earlier computations — ( A ) for three-dimensional models of a 1/8 acinus , in which the diffusion-reaction problem , Equation 3 , was approximated by random walks ( RW ) with absorbing boundary conditions [21] , [22]; and ( B ) for two-dimensional models of a 1/8 and 1/128 acinus , in which Equation 3 was solved by the finite-element method ( FEM; [4] , Text S1 ) . The RM , RW , and FEM currents agree within a factor of order one over 5 orders of magnitude of the permeability and nearly 3 orders of magnitude of the current . This is excellent agreement considering that the RW and FEM treatments trace out every duct detail ( Fig . 6 ) , while the RM evaluates Equation 3 with as sole structural input ( see also [13] , [23] ) . The sharp transitions in the RM currents , which result from the decomposition of the surface into regions completely accessible/inaccessible to O2 molecules ( Fig . 2 ) , are smooth in the RW and FEM treatment: e . g . , the horizontal plateau is transformed into a weakly W-dependent current . Thus the RM pinpoints changes in screening not easily detectable in purely numerical computations—similar to that , in adsorption of gases on solids , different models may or may not produce a knee and plateau in the adsorption isotherm [24] . In Fig . 6 , the knee marks the transition from overexploration to underexploration of the membrane surface and yields the cross-section area of the source; in adsorption the knee marks the transition from submonolayer to multilayer coverage and yields the surface area of the solid . The RM , RW , and FEM currents also agree , within a factor of order one , in terms of the knee points on the W axis . The RW currents for the two different realizations of a 1/8 acinus , but with identical source , merge at large W , in agreement with the RM prediction that screened currents depend only on the cross-section area of the source , Equation 5a–c . The RW currents at are 0 . 5–0 . 6 nmol/s , which is about half of the experimental value at rest , ( Table 2 ) , because the 's in [21]–[22] are smaller than in Table 2 , which is averaged over many 1/8 acini . This comparison of the RM , RW , and FEM currents provides important validation of the RM treatment and fractal model . Are there uncertainties of the model outside the diffusion-reaction framework , Equations 2–4 ? The results in Figs . 2–6 are based on a constant lung volume , . But breathing is dynamic and the volume of the lung changes periodically . If the fractional volume change is , then the fractional linear change , i . e . , the side length of alveoli and acini , will be ε/3 . At rest , the tidal volume for a normal adult male amounts to a volume change of [1] , whence . Evaluating Equation 4 , 5 , 7 with accordingly changed values of the structural parameters , we find that the fractional change in the oxygen current is 2 . 9% , and the pulmonary efficiency changes from 13 . 8% to 13 . 4% . We consider both changes insignificant . During maximum exercise , from functional residual capacity to peak inspiration , the lung volume changes as much as 50% [1] . In this case , the oxygen current changes by 17% , and the pulmonary efficiency increases from 94% to 100% . During the periodic expansion and contraction of the lung , the air flow and gas mixing in the acinar airways can be chaotic [25] , [26] . The chaotic patterns provide the time-resolved fluid-dynamic details of what happens at the transition between convection and diffusion . Upstream of the transition , velocity vectors point mostly downstream , along the duct axis; downstream of the transition , velocity vectors point mostly transverse to the duct axis ( diffusion to duct wall ) . In the transition region , some velocity vectors point along the duct axis , others transverse to the duct axis , giving rise to eddy-like instantaneous flow patterns . In the stationary diffusion-reaction framework , Equations 2 , 3 , these flow patterns are averaged out , and the average yields the boundary condition at the source , Equation 2b . One may view the flow patterns as fluctuations around the stationary average and examine their effect on the gas exchange . From the perspective of length scales , the effect is minimal: a typical eddy has a diameter of 0 . 005–0 . 01 cm [25] , [26] , which is small compared to the side length of the source ( for zcd = 21; for zcd = 18 ) , side length of the gas exchangers ( 0 . 36 cm for zcd = 21; 0 . 57 cm for zcd = 18 ) , and exploration length , . From the perspective of the current , the fluctuations have no effect: in the mean-field description provided by the diffusion equation , Equation 2 , they give the oxygen source , which is a dividing surface of zero thickness in Equation 2b , a nonzero thickness of the order of the eddy size , centered at the subacinus entrance . But the oxygen concentration is identical on both sides of the dividing surface , so the fluctuations are less than ca as often as they are larger , and the stationary oxygen current across the lung is the same as in the absence of fluctuations . In terms of dynamics , the eddies create a well-stirred chemical reactor , the prerequisite for a source with uniform time-averaged oxygen concentration [4] . The branching pattern of the fractal model in our study is symmetric , in line with that branching in the acinar airways is symmetric to a significant degree [12] . But the bronchial tree is asymmetric [27] , [28] , and the question arises how asymmetry would influence our calculations . Asymmetry is important in the bronchial tree ( convection , low z ) where the flow distribution depends sensitively on the aspect ratio of the daughter branches [6] , [29] . Asymmetry is not important in the acinar tree ( diffusion , high z ) because diffusion currents are driven by local concentration gradients , which depend predominantly on the distance to the nearest wall , and only little on geometric factors like width , length , and angle of daughter branches . Insensitivity of diffusion to asymmetry is supported by the RW results in [21] , [22] , which include various asymmetries , but depend mostly on the size of the subacinus ( Fig . 6A ) . However , little is known whether an asymmetric flow distribution in the bronchial tree can propagate all the way to the acinar tree and generate significantly different oxygen concentrations at the entrances of different gas exchangers . If so , then ca in Equations 1b , 2b , 4 should be an appropriately averaged entrance concentration . We have shown that oxygen exchange across the alveolar membrane can be successfully modeled as diffusion-reaction process bounded by a fractal , space-filling surface; that the fractal nature of the surface is key to the high performance of the gas exchanger; and that the operation of the system can be understood in terms of variable degrees of screening under different physiological conditions . The results in Fig . 3 ( validation of the model ) and Fig . 4 ( prediction of currents at arbitrary permeabilities ) were calculated by plugging seven numbers ( Rows 1&6 in Table 1; Rows 3–7 in Table 2 ) into Equations 4 , 5 , which requires no more than a pocket calculator . The success of the calculation in Fig . 3 demonstrates the power of the fractal model and associated physical insight—transformation of 180 , 000 screened gas exchangers into 1 , 500 , 000 unscreened gas exchangers , accompanied by an increase of the oxygen current by a factor of 20 , essentially at constant lung volume , surface area , partial pressure difference across the membrane , and membrane permeability . The success of the calculation in Fig . 4 provides a robust map of the vast territory of membrane permeabilities different from the normal physiological value . The success of the “pocket-calculator formula , ” Equation 4 , 5 , promises a robust map of respiratory performance of the lung in other species [1] , [2] , [8] . We demonstrated that the space-filling architecture provides optimum adaptation to changing demands—the ability to switch from a low oxygen current at rest to high currents at exercise ( vertical transition in Fig . 4 ) , self-regulated by diffusional screening , without external control circuits . Such adaptation is one hallmark of robustness in systems biology [30] . At the same time , the architecture provides optimum adaptation to changing resources—maintenance of a constant oxygen current in the event of a change in permeability , surface area , or other operational parameters ( homeostasis; horizontal transition in Fig . 4 ) , again without external controls . Such changes may occur in pulmonary edema , inhalation of aerosols , poor ventilation in asthma , pneumonia , emphysema , lung surgery , hyperbaric oxygen treatment . Insensitivity to specific operational parameters is the second hallmark of robustness in biological systems . While such insensitivity is of outstanding value for stable oxygen delivery under less than perfect conditions , it may make direct experimental tests or therapeutic applications , in which departures from normal oxygen delivery are observed , feasible only under severe departures from normality . Necessary for such tests and applications will be quantitative estimates of ΔW , ΔS , Δzcd , , and under various disease and treatment conditions . To the best of our knowledge , such estimates have yet to be developed . We reverse-engineered the lung's performance characteristics by monitoring how the oxygen current varies as we vary transport and structure parameters , here W and Df , over values far from those found in the lung , similarly to how reverse engineering of biomolecules requires experiments at temperatures far from ambient temperature [31] . The resulting understanding of how structure determines function , how a single three-dimensional surface can create a platform for coexistence of multiply optimized properties , gives new meaning to the statement that “Lebesgue-Osgood monsters are the very substance of our flesh” ( [3] , p . 149 , 159 ) . | The possibility of predicting oxygen currents in the human lung under varying conditions may give new understanding of the lung's operation , new therapeutic interventions , and new designs for non-biological transport systems . We introduce such a computation which requires only a pocket calculator and agrees with measured currents within a few percent . It treats the network of airways as a fractal surface and exhibits wide-ranging adaptation to changing process parameters , including tolerance to changes in membrane permeability , near-invariance of trans-membrane oxygen pressure at rest and exercise , and transformation of 180 , 000 gas exchangers into 1 , 500 , 000 exchangers from rest to exercise . We show that alternative architectures perform sub-optimally and that the observed performance results from a competition between underexploration and overexploration of the surface by oxygen molecules . | [
"Abstract",
"Introduction",
"Methods",
"Results/Discussion"
] | [
"physics",
"mathematics",
"physiology",
"computational",
"biology",
"biophysics",
"respiratory",
"medicine"
] | 2010 | Reverse Engineering of Oxygen Transport in the Lung: Adaptation to Changing Demands and Resources through Space-Filling Networks |
HIV-1-infected cells in peripheral blood can be grouped into different transcriptional subclasses . Quantifying the turnover of these cellular subclasses can provide important insights into the viral life cycle and the generation and maintenance of latently infected cells . We used previously published data from five patients chronically infected with HIV-1 that initiated combination antiretroviral therapy ( cART ) . Patient-matched PCR for unspliced and multiply spliced viral RNAs combined with limiting dilution analysis provided measurements of transcriptional profiles at the single cell level . Furthermore , measurement of intracellular transcripts and extracellular virion-enclosed HIV-1 RNA allowed us to distinguish productive from non-productive cells . We developed a mathematical model describing the dynamics of plasma virus and the transcriptional subclasses of HIV-1-infected cells . Fitting the model to the data allowed us to better understand the phenotype of different transcriptional subclasses and their contribution to the overall turnover of HIV-1 before and during cART . The average number of virus-producing cells in peripheral blood is small during chronic infection . We find that a substantial fraction of cells can become defectively infected . Assuming that the infection is homogenous throughout the body , we estimate an average in vivo viral burst size on the order of 104 virions per cell . Our study provides novel quantitative insights into the turnover and development of different subclasses of HIV-1-infected cells , and indicates that cells containing solely unspliced viral RNA are a good marker for viral latency . The model illustrates how the pool of latently infected cells becomes rapidly established during the first months of acute infection and continues to increase slowly during the first years of chronic infection . Having a detailed understanding of this process will be useful for the evaluation of viral eradication strategies that aim to deplete the latent reservoir of HIV-1 .
High levels of cell-associated HIV-1 RNA can be observed in peripheral blood of patients with undetectable plasma viremia during combination antiretroviral therapy ( cART ) [1]–[4] . The various HIV-1 RNA and DNA species that are present during the viral life cycle can serve as biomarkers for basal transcription in viral reservoirs with different properties [5] , [6] . Gaining a quantitative understanding of the development and turnover of HIV-1-infected subpopulations and viral latency is of particular interest in light of recent efforts in viral eradication strategies [7]–[10] . Highly sensitive assays for HIV-1 plasma RNA in patients on cART usually provide bulk measurements of viral activity and cannot distinguish between different infected subpopulations [11] . In contrast , the study by Fischer et al . [12] combined highly sensitive PCR assays for unspliced ( UsRNA ) and multiply spliced ( MsRNA-tatrev and MsRNA-nef ) HIV-1 RNA species with limiting dilution endpoint analysis of peripheral blood mononuclear cells ( PBMCs ) . In addition to intracellular RNA transcripts , extracellular virion-enclosed HIV-1 RNA that provides a marker for cells releasing virus particles was also measured . The study identified four distinct viral transcriptional classes: two overlapping cell classes of high viral transcriptional activity , representative of a virus producing phenotype; and two cell classes that express HIV-1 RNA at low and intermediate levels that match definitions of viral latency [12] , [13] . Analyzing the decay kinetics of plasma viral load in HIV-1-infected patients on cART using mathematical models has resulted in a detailed understanding of viral replication dynamics in vivo [14]–[16] . The plasma viral load typically exhibits three exponential phases during the first year after start of cART ( Figure 1 ) . Due to the rapid turnover of free virus in blood [17] , the viral decay phases are thought to reflect the contribution of different HIV-1-infected cell populations on viral production . The first phase with a half-life of 1 to 2 days is attributed to the loss of activated , virus-producing cells [18] , [19] . The second phase exhibits a half-life of 1 to 4 weeks and is considered to reflect the loss of so-called persistently infected cells with a lower state of activation [20] , [21] . The third phase decay has a long half-live of 39 weeks suggesting that latently infected cells are a primary candidate for this cellular compartment [22] , [23] , although slow release of virus from the follicular dendritic cell network is another possibility [24] . Although not shown in Figure 1 , in many patients , after the third phase a final low steady state level of plasma viremia is attained , that has been called a fourth phase [22] . This phase has also been attributed to release of virus from activated latently infected cells [22] . Other mathematical models have been developed that stratify the infected cells into additional subpopulations such as non-productively infected cells during the intracellular eclipse phase [25] and defectively infected cells [26] . Nevertheless , most studies to date are focused on the analysis of viral load and only indirectly allow inferring the kinetics of cellular subpopulations . Few studies have attempted to characterize the concentration of virus and several infected subpopulations based on data simultaneously [26] . Fitting mathematical models to multiple quantities of viral replication would result in refined parameter estimates for describing the generation and maintenance of latently infected cells . In this study , we developed a mathematical model that describes the dynamics of different transcriptionally active subclasses of HIV-1-infected cells and the viral load in peripheral blood . The model was fitted to previously published data from five chronically HIV-1-infected patients starting cART [12] . This allowed us to estimate critical parameters of the within-host dynamics of HIV-1 and the turnover of various subpopulations of infected cells . Finally , we simulated the development of the latently infected cell pool during acute infection , providing useful information for viral eradication strategies .
We first devised a detailed model of the within-host dynamics of HIV-1 that is based on the observations of different subclasses of HIV-1-infected cells in the study by Fischer et al . [12] . The five subclasses are HIV-1 DNA , low , medium and high HIV-1 RNA expressing and cells that have virion-enclosed HIV-1 RNA associated with them ( also see Methods ) . These subclasses show distinct decay dynamics in patients on cART ( Figure 2 ) . The slow decay of the subclass of PBMCs that contains proviral DNA ( DNA ) indicates that this cell population primarily contributes to the third phase decay and likely consists of defectively or latently infected cells to a large extent . The subclass of cells exhibiting UsRNA only ( Low ) decays slowly and most likely consists mainly of latently infected cells with low basal transcription of HIV-1 . The cells with medium transcriptional activity ( Mid ) appear to contribute to the second and the third phase viral decay , which is characteristic of persistently and latently infected cells . The early drop in PBMCs with a higher transcriptional activity ( High ) , which is more pronounced compared to cells with a low and medium transcriptional activity , that is followed by a slower loss of cells is reminiscent of activated , virus-producing and persistently infected cells . Finally , the PBMCs that have extracellular virion-enclosed HIV-1 RNA associated with them ( Extra ) show a very rapid loss before reaching the limit of detection . This is expected as they should represent the short-lived population of virus-producing cells [4] that contribute to the first phase of viral decay . The different subclasses of HIV-1-infected cells clearly overlap and are representative of heterogeneous cell populations . Furthermore , the life cycle of HIV-1 from infection of a cell to the release of virus particles can be divided into cell populations with different transcriptional activity [27] . We took both of these important characteristics into account in our model that consists of 12 subpopulations of cells that can be stratified according to their HIV-1 DNA and RNA content ( Figure 3 and Methods ) . In this model , we defined persistently infected cells ( and ) as long-lived cells that can produce viral particles . Latently infected cells ( and ) were assumed to transcribe HIV-1 RNA at low or intermediate levels [12] , [13] . Infected cells that are HIV-1 DNA positive , but HIV-1 RNA negative , were assumed to remain transcriptionally silent during the observation period and considered as defectively infected cells ( ) . Fitting the mathematical model to the data from five HIV-1-infected patients resulted in a good description of the viral and cellular decay kinetics during cART ( Figure 4 and Text S1 ) . The individual dynamics of each subpopulation of cells are shown separately ( Figure 5 ) . The model clearly describes more pronounced decay dynamics in infected cells with increasing transcriptional activity . Table 1 provides a summary of the geometric means as well as the ranges of the best fit parameter estimates that describe the virus dynamics in each of the five patients . We found that 1 . 1% ( 0 . 2%–7 . 0% ) of all CD4 T cells can be target cells for infection with HIV-1 . We also obtained estimates for the average lifespans of target cells ( 61 days , range: 11–528 days ) and latently infected cells ( 33 years , range: 168 days–505 years ) . While others have estimated the average half-life of latently infected cells to be 6 . 3 months [28] and 44 months [29] , our estimates are less precise due to the much shorter follow-up period after start of cART . However , the estimated activation rate of latently infected cells ( d , range: d ) that also influences the slope of the third phase decay in plasma HIV-1 RNA is consistent with previous findings [30] . The parameters , and denote the fractions of cells that end up in a particular subpopulation in a sequential process during the intracellular eclipse phase . From this , we can calculate the average proportion of newly infected cells that become a certain cell type ( Text S1 ) . In contrast to another study [26] , we find that only 63 . 4% ( 0 . 2%–7 . 0% ) of infected cells become activated , virus-producing cells ( ) . A substantial fraction of infected target cells results in defectively ( 14 . 0% ) and persistently infected cells ( 21 . 2% ) . The proportion of infected cells that become latently infected or die before ending up in one of the subpopulations is small ( 0 . 3% and 1 . 1% , respectively ) . Note that after activation , latently infected cells can then either become persistently infected or activated , virus-producing cells by moving through cell class . Transcriptional bursts that increase the level of viral RNA transcription occur on average every 12 . 7 days ( , range: 3 . 5–165 . 2 days ) and 9 . 7 days ( , range: 1 . 5–37 . 0 days ) in latently and persistently infected cells , assuming that bursts last for one day on average ( d ) . The total number of virus particles produced by a cell during its lifetime , the viral burst size , was estimated at 21′000 virions per cell ( range: 3′500–240′000 virions per cell ) . Note that we assumed that persistently infected cells in an elevated transcriptional state ( ) produce viral particles at the same rate as activated , virus-producing cells . However , the duration of virus release is shorter in persistently infected cells as they can revert to a lower transcriptional state ( ) . The majority of virus particles is produced by activated , virus-producing cells ( 68 . 3% , range: 5 . 6%–98 . 1% ) with the remaining proportion being produced by persistently infected cells . The high viral burst size suggests that the total number of virus-producing cells in peripheral blood must be small and we indeed found an average of only 25 . 7 cells ml ( range: 7 . 8–143 . 1 cells ml ) in the model during the chronic phase of infection . The parameters were estimated by fitting the virus dynamics model to data of patients chronically infected with HIV-1 . Although there are mathematical models that describe acute and chronic HIV infection together [31] , [32] , the virus dynamics during acute infection could differ significantly due to different parameter values and even model structures . Nevertheless , our model can still be used to simulate the virus dynamics during the acute phase and compare the results to experimental and clinical data . We used the average of the estimated parameters to simulate early infection with HIV-1 from a small viral inoculum in a hypothetical patient . We set copy per ml and assumed that the target cells are at steady-state ( ) . The rapid rise of plasma HIV-1 RNA during the first weeks of infection is followed by the chronic phase at which the virus concentration reaches its set-point level ( Figure 6 ) . The total pool of latently infected cells ( ) show somewhat different dynamics during acute HIV-1 infection . A very rapid expansion of latent cells during the viral growth phase is followed by a slower increase into the chronic phase of infection . From the time of peak viremia ( 22 days ) to the chronic phase ( 1000 days ) , the latently infected cell pool expands 14 . 3-fold from 9 . 8 to 140 . 4 cells per ml . The expansion of the total number of HIV-1 DNA positive cells from the acute ( 1813 cells per ml ) to the chronic phase ( 7608 cells per ml ) is smaller ( 4 . 2-fold ) . This is consistent with the 3 . 8-fold difference in the number of HIV-1 DNA copies that were measured in patients that initiated cART during the acute and chronic phase from another study [33 , and see Text S1] . The time after infection at which latently infected and HIV-1 DNA positive cells reach 50% of their chronic level is 441 and 451 days , respectively . Altogether , this illustrates the opportunity for eradication strategies during early cART interventions as the pool of HIV-1 infected cells seems to be substantially smaller during acute infection than during chronic infection .
We present the first mathematical model of virus dynamics that groups the different subpopulations of HIV-1-infected cells according to their transcriptional profile . The model assumes a heterogeneous population of latently and persistently infected cells having occasional transcriptional bursts to increase their level of RNA transcription which is consistent with experimental data from Fischer et al . [12] . Fitting this model to the unique data of virus transcription levels at the single cell level resulted in new estimates of the HIV-1 dynamics in vivo . We found that a large fraction of infected cells become either defectively or persistently infected cells . Furthermore , we found that the viral burst size can be high , between and viral particles per virus-producing cell . Lastly , we simulated the acute phase of HIV-1 infection in a typical patient . This illustrated that the latently infected cell pool becomes rapidly established during the first months of acute infection and shows a slow increase during the first years of chronic infection . Our study is unique in that we fit a mathematical model of HIV-1 within a host to data of the dynamics of different subclasses of infected cells . This is a substantial step beyond modeling studies that considered free virus in plasma , CD4 T cells and bulk measurements of viral activity only . The new quantitative insights into the replication dynamics of HIV-1 in vivo that this study provides will be useful for an improved understanding of HIV and the effects of novel treatment strategies . The measurements of HIV-1-infected cells and the virus concentration were performed in blood only . In our mathematical model , we thus assume homogeneous mixing of virus and cells throughout the body . It is important to note , however , that the characteristic decay profile in the study by Fischer et al . [12] could also be a result of differential trafficking of virus particles and HIV-1-infected subpopulations of cells between the blood and lymphoid tissue . It has also been suggested that the virion clearance rate from the blood corresponds to a virion efflux to other organs where the virus is ultimately cleared [34] . Furthermore , non-productively infected CD4 T cells could also indirectly die through ‘bystander’ effects [35] , [36] . Finally , the typical second-phase decay could also result from virus production in infected macrophages [20] or heterogeneity in activation rates of latently infected cells [30] , [37] . The concept of persistently infected cells has been previously used in mathematical models of HIV-1 dynamics to describe a population of long-lived cells that can contribute to the second-phase decay of virus during cART [20] , [26] . Since the cellular subclass with medium transcriptional activity ( Mid ) seems to be rather long-lived and strongly characterized by a decay dynamics that could contribute to the second-phase decay of virus , we assumed that the majority of persistently infected cells belong to this class . This is consistent with the notion that persistently infected cells could be in a lower state of activation [21] . The contribution of other subpopulations of cells to the subclass Mid is small as the average lifespan of those cells is longer ( ) or shorter ( ) than that of persistently infected cells . It remains to be determined whether persistently infected cells could indeed release viral particles as a result of an increase in their transcriptional level . However , the reversion of virus-producing cells into a lower state of activation has been proposed previously [30] . The data did not allow estimation of both the frequency and duration of transcriptional bursts that lead to the release of virions in persistently infected cells . We assumed that once persistently infected cells release viral particles , the probability to die through cell lysis is the same as the probability of reversion . For simplicity , we considered only one type of CD4 target cell whereas HIV-1 can infect activated but also resting CD4 T cells . Our estimate of the proportion of CD4 T cells that are target cells ( 1 . 1% ) is somewhat lower than the 6 . 5% of CD4 Ki-67 T cells in HIV-1-infected individuals that have been measured previously [38] . Also , the estimated average lifespan of target cells was longer than what others have estimated for activated cells [39] . The target cells in the model thus represent a particular subset of CD4 T cells that is smaller than the population of activated cells but has a longer average lifespan . The longer lifespan of target cells results from the assumption that the death rates of cells during the intracellular eclipse phase ( to ) and persistently infected cells ( ) remain the same after infection , i . e . , are the same as the death rate of uninfected target cells ( ) . While persistently infected cells are indeed defined as long-lived cells that can produce virus , some studies have suggested that infected cells in the eclipse phase could also be a target of cytotoxic T lymphocyte ( CTL ) killing and experience high death rates [25] , [40]–[42] . The early steps of proviral transcription also remain elusive . It has been suggested that the decay of non-integrated viral DNA in infected cells could render them CD4 target cells again [43]–[46] . The kinetics of HIV-1 DNA indeed show a small drop early after start of cART ( Figure 2 and ref . [47] ) . However , we have excluded this effect for simplicity . Ultimately , the mechanisms of viral latency in HIV-1 remain a matter of debate [48] . In our model , we assumed that after proviral insertion some cells fail to increase viral RNA transcription and become latently infected cells . Latency could also result from infection of resting CD4 T cells or de-activation of activated CD4 T cells . We have not included the latter two mechanisms in our model as the data would not allow us to distinguish between them . The complexity of the HIV-1 life cycle and its mathematical representation prevents the identification of a ‘true’ underlying model . We made several simplifying assumptions in our default model but we also studied a series of alternative models and found that some of those models also fit the data well ( Table S1 in Text S1 ) . Importantly , the estimates of critical parameters such as the viral burst size , the proportion of CD4 T cells that are target cells , and the fractions of cells that become defectively , latently or persistently infected in the alternative models that fit the data well were very similar to those estimated with the default model ( Table S2 in Text S1 ) . We were also able to reject some competing hypotheses about the life cycle of HIV-1 ( Table S1 in Text S1 ) . Removing the intracellular eclipse phase , that contains infected cells at different stages with increasing levels of viral transcription , impairs the model fit . Assuming that latently or persistently infected cells are homogeneous subpopulations results in a substantially worse fit to the data . The limited number of data points and patients prevented a more thorough analysis and resulted in substantial uncertainty in estimating the model parameters . The wide ranges of estimates in Table 1 illustrate that the reported parameter values need to be treated with caution . We also used the least-squares method to fit the model to the data and did not consider maximum likelihood approaches [49] , values below the limit of detection or nonlinear mixed-effect models [50] . It remains to be determined how well the parameter estimates that were obtained during the chronic phase of infection represent the situation of acute HIV-1 infection . It is re-assuring that the simulated virus dynamics of acute infection show a peak around three weeks after infection , which is in agreement with observations in patients [51] , [52] . Nevertheless , differences in immune activation during acute infection are likely to result in different proportions of cells becoming latent upon infection and different activation rates of latently infected cells . Hence , our results on the development of the latently infected cell pool during acute infection need to be interpreted with caution . We found the HIV-1 burst size in vivo to be large , corroborating previous estimates from Chen et al . [53] who found the average burst size in SIV-infected rhesus macaques to be between and . This is higher than other estimates that were in the range of virions per cell [54] , [55] and suggests that the number of virus-producing cells must be lower than previously anticipated . Measurements of extracellular virion-enclosed HIV-1 RNA ( ) in the study by Fischer et al . [12] suggest that the number of productively infected cells in peripheral blood is small which is also reflected in our model fits . In contrast to other studies that assumed the viral production rate in long-lived persistently infected cells to be lower than in activated , virus-producing cells [56] , we considered the viral production rates to be the same in both cell types . However , in our model persistently infected cells can have occasional transcriptional bursts from to , where they can release virus particles before reverting back to a lower transcriptional state or dying . Our simulations of the development of different pools of HIV-1-infected cells are in good agreement with observations in patients . We find that the total number of HIV-1 DNA positive cells rapidly build up during the acute stage of infection . A very similar expansion was found in a recent study that measured the total number of HIV proviruses in PBMCs during the first weeks of HIV infection [57] . Also , our predicted ratio of the number of HIV-1 DNA positive cells during acute and chronic infection is in the same range as previously reported [33] , [47] . The study by Murray et al . [47] further suggested that the level of HIV DNA continuously increases with duration of infection , reaching its 50% level at two years after infection . This contradicts earlier findings of stable levels of HIV-1 DNA positive PBMCs during the natural course of infection [58] . Our model predicts that the number of HIV-1 DNA positive PBMCs increases slowly during the first years of chronic infection and reaches its 50% level at 451 days after infection , corroborating the findings by Murray et al . [47] . An important question that remains is how many of HIV-1 DNA positive cells are latently or defectively infected . We found that the fraction of cells becoming defectively infected is surprisingly high . On the one hand , this could be a result of the assumption that HIV-1 DNA positive cells without viral RNA transcription remain silent . Some of these cells could actually be activated and start to produce UsRNA at low levels , i . e . , become cells of the latent class . Eriksson et al . [59] measured a 300-fold difference between the number of latently infected cells as measured with a viral outgrowth assay and the total number of HIV-1 DNA positive resting CD4 T cells . However , Ho et al . [60] showed a substantial fraction of noninduced proviruses in cells that have been stimulated in a viral outgrowth assay are replication-competent . They found that that the frequency of intact noninduced proviruses was at least 60-fold higher than the frequency of proviruses induced in a viral outgrowth assay . The median frequency of cells with intact non–induced proviruses per HIV-1 DNA positive resting CD4 T cells was estimated at 3 . 7% [60] . In our simulation , the fraction of latently infected cells ( ) in all HIV-1 DNA positive cells ( DNA ) is 1 . 8% ( 140 . 4/7608 ) during chronic infection . The striking correspondence of these numbers suggests that our mathematical model realistically describes the dynamics of the latent reservoir . Since the subpopulation of is much larger than , the majority of latently infected cells consist of PBMCs that contain solely HIV-1 UsRNA ( Low ) , indicating that this transcriptional subclass is a good marker for viral latency . This study provides an important step towards a more quantitative understanding of the dynamics of HIV-1 in vivo , in particular of the generation and maintenance of latently infected cells . A better understanding of the number of latently infected cells during acute infection is crucial for evaluating and predicting the outcome of early treatment and eradication strategies . Early cART treatment has been suggested to facilitate long-term control of HIV-1 [61] and studies have shown that it results in lower viral load levels during chronic infection [62] . Although the effects on viral load might only be transient [63] , early treatment can prevent the expansion of viral cellular reservoirs in peripheral blood [33] . More recent strategies aim towards depletion of this reservoir [9] , preferably during acute infection [64] . Predicting the chances of such eradication strategies critically depends on the ability to accurately quantify the pool of latently infected cells at various time points during HIV-1 infection . Our study supports the experimental finding that the latent reservoir becomes rapidly established during the first months of infection , and shows that the reservoir represents a significant proportion ( 1% ) of all HIV-1 DNA positive PBMCs during chronic infection . In addition , our mathematical model realistically describes the dynamics of different HIV-1-infected subpopulations of cells which will be useful for projecting the effects of eradication strategies .
We used previously published data from five chronically HIV-1-infected therapy naive patients that initiated cART using reverse transcriptase and protease inhibitors ( patient numbers: 103 , 104 , 110 , 111 , 112 ) [12] . Plasma HIV-1 RNA ( copies per ml ) and CD4 T cells ( per µl ) were measured at several time points during the first 48 weeks of cART . PBMCs were purified at weeks 0 , 2 , 4 , 8 , 12 , 24 and 48 after the start of cART as described in Fischer et al . [65] . Serial dilution of PBMCs and patient matched PCR quantification of HIV-1 RNA species and DNA was performed as described elsewhere in detail [12] , [13] , [66] , [67] . The freeze-thaw nuclease digestion method to differentiate between intracellular and virion encapsidated HIV-1 RNA has also been previously described in detail [4] , [33] . HIV-1 RNA or DNA positive cell fractions measured as cells per 10 PBMCs were converted to number of cells per ml of blood by multiplying with the number of PBMCs per ml . This ultimately lead to the stratification of cells to five ( partially overlapping ) subclasses [12]: For the subclass DNA , we make the assumption that there is only one proviral DNA copy per infected cell [68] . We devised a new virus dynamics model ( Figure 3 ) which is adapted from previously published models [19] , [25] , [26] , [30] . The various subpopulations of infected cells were stratified according to their HIV-1 DNA and RNA content . The model can be described by the following set of ordinary differential equations ( ODEs ) : ( 1 ) ( 2 ) ( 3 ) ( 4 ) ( 5 ) ( 6 ) ( 7 ) ( 8 ) ( 9 ) ( 10 ) ( 11 ) ( 12 ) ( 13 ) CD4 target cells , , are produced at rate and can become infected by virus particles , , at rate . denotes treatment efficacy , where before the start of antiretroviral therapy . Newly infected cells move through the intracellular eclipse phase , where denotes the stage of reverse transcription , the stage of proviral integration , and to subsequent stages with increasing transcriptional activity . After the intracellular eclipse phase , activated , virus-producing cells , , start to release free virus particles with a total viral burst size . Some of the cells during the intracellular eclipse phase can become defectively infected cells , , latently infected cells , , or persistently infected cells , . While we assume that defectively infected cells remain transcriptionally silent , both latently and persistently infected cells can exhibit transcriptional bursts that rise their transcriptional profile from Low to Mid and Mid to High , respectively . Latently infected cells in an elevated transcriptional state can become activated at rate or move back to the lower transcriptional state . Similarly , persistently infected cells that are highly transcriptionally active can release free virus particles at rate before they revert to a state of lower transcriptional activity or die . and describe cell death and viral clearance rates , respectively . Due to the complexity of the full model , we make a number of simplifying assumptions . First , we assumed several of the cell death rates to be the same: the death rates of virus-producing cells and the death rates of defectively and latently infected cells . The death rates of infected cells that are not virus-producing and do not solely belong to a resting phenotype , such as defectively and latently infected , were kept the same as the death rate of target cells ( ) . Second , the viral production rates in both virus-producing cells ( and ) are kept the same , i . e . , . Note , however , that persistently infected cells ( ) have a lower burst size than activated , virus-producing cells ( ) because they can revert to a non-productive state ( ) . The default model described above is compared to a number of alternative models with different assumptions of the viral life cycle ( Text S1 ) . The default model contains 22 parameters of which 10 are fixed to previously used values from the literature or based on assumptions ( Table 1 ) . The remaining 12 parameters were constrained based on literature values and consensus and we used the geometric mean of the restricted range as starting values when fitting the model to data . This proved to be a good strategy for estimating the model parameters . The set of ODEs were solved numerically in the R software environment for statistical computing [69] using the function ode from the package deSolve [70] . The 12 model variables were initiated with the target cells at their steady-state ( ) , copy per ml , and all other variables being zero . We assumed that the chronic state of infection is reached after 1000 days ( about three years ) , set [23] and further integrated the system during the time on cART ( 336 days ) . The concentration of free virus was measured directly but several of the infected cell populations contribute to the different subclasses of PBMCs ( Figure 3 ) : , , , and . We further assume that target cells , , correspond to a fraction , , of all CD4 T cells . All 12 parameters ( 11 model parameters and one scaling parameter ) were estimated by fitting the model to the data of each patient individually and minimizing the sum of squared residuals ( SSR ) between the prediction of the model and the data ( taking the natural logarithm ) . All data points were weighted equally . However , the higher number of data points for free virus compared to cellular subclasses ( e . g . , ) forced the model to fit the virus concentration better than the other variables . We used the minimization algorithm by Nelder & Mead [71] that is implemented in the function optim and the parallel package for parallel computation . The algorithm by Nelder & Mead is very robust in finding local optima . As a sensitivity analysis , we used different starting values for the parameters and the method SANN that is a variant of simulated annealing . Simulated annealing usually performs better in finding global optima but is relatively slow . In both cases , we found the best-fit parameter estimates to be the same or very similar to our default fitting strategy . Parameter estimates are presented as geometric means including the ranges over all five patients . Code files can be obtained freely upon request from the corresponding author . | Gaining a quantitative understanding of the development and turnover of different HIV-1-infected subpopulations of cells is crucial to improve the outcome of patients on combination antiretroviral therapy ( cART ) . The population of latently infected cells is of particular interest as they represent the major barrier to a cure of HIV-1 infection . We developed a mathematical model that describes the dynamics of different transcriptionally active subclasses of HIV-1-infected cells and the viral load in peripheral blood . The model was fitted to previously published data from five chronically HIV-1-infected patients starting cART . This allowed us to estimate critical parameters of the within-host dynamics of HIV-1 , such as the the number of virions produced by a single infected cell . The model further allowed investigation of HIV-1 dynamics during the acute phase . Computer simulations illustrate that latently infected cells become rapidly established during the first months of acute infection and continue to increase slowly during the first years of chronic infection . This illustrates the opportunity for strategies that aim to eradicate the virus during early cART as the pool of HIV-1 infected cells is substantially smaller during acute infection than during chronic infection . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"immunodeficiency",
"viruses",
"infectious",
"diseases",
"medicine",
"and",
"health",
"sciences",
"diagnostic",
"medicine",
"medical",
"microbiology",
"hiv",
"viral",
"pathogens",
"hiv",
"clinical",
"manifestations",
"microbial",
"pathogens",
"biology",
"and",
"life",
"... | 2014 | Quantifying the Turnover of Transcriptional Subclasses of HIV-1-Infected Cells |
We reported previously that a proportion of natural CD25+ cells isolated from the PBMC of HCV patients can further upregulate CD25 expression in response to HCV peptide stimulation in vitro , and proposed that virus-specific regulatory T cells ( Treg ) were primed and expanded during the disease . Here we describe epigenetic analysis of the FOXP3 locus in HCV-responsive natural CD25+ cells and show that these cells are not activated conventional T cells expressing FOXP3 , but hard-wired Treg with a stable FOXP3 phenotype and function . Of ∼46 , 000 genes analyzed in genome wide transcription profiling , about 1% were differentially expressed between HCV-responsive Treg , HCV-non-responsive natural CD25+ cells and conventional T cells . Expression profiles , including cell death , activation , proliferation and transcriptional regulation , suggest a survival advantage of HCV-responsive Treg over the other cell populations . Since no Treg-specific activation marker is known , we tested 97 NS3-derived peptides for their ability to elicit CD25 response ( assuming it is a surrogate marker ) , accompanied by high resolution HLA typing of the patients . Some reactive peptides overlapped with previously described effector T cell epitopes . Our data offers new insights into HCV immune evasion and tolerance , and highlights the non-self specific nature of Treg during infection .
Hepatitis C virus is a small positive sense single stranded RNA virus , which causes persistent infection that leads to cirrhosis , cancer and liver failure . In the acute phase of the infection , the host usually mounts strong CD4+ and CD8+ T cell responses , but this wanes in the next few months during the transition to persistence ( reviewed in reference [1] ) . Typically , in persistently-infected patients , the frequency of HCV-specific IFNγ-producing effector T cells is low ( usually <0 . 3% of PBMC by ELISPOT ) and that of IL2-producing cells is even lower [2] . T cells , particularly CD4+ T cells , proliferate poorly in response to HCV antigens [3] , although CD8+ T cells proliferate slightly better ( Li and Gowans , unpublished data ) . The reason behind the lack of adequate immunity to HCV in human is not well understood , although it is likely to be multi-factorial [1] , [4] . IL-10 producing type 1 regulatory T cells ( Tr1 ) may play a role in HCV persistence [5] , [6] , and more recently , several groups suggested that natural regulatory T cell ( Treg , a different type of suppressor cell to Tr1 ) may be also important [7] , [8] , [9] , [10] . The frequency of circulating CD4+CD25+ cells ( the cell population in which Treg are predominantly contained [11] ) in the blood of HCV carriers was higher than in healthy donors and individuals who had resolved the infection [7] . In addition , the percentage of CD4+CD25+ cells within the infected liver was much higher than in the peripheral blood [8] . ( A review of this topic was published recently [12] ) . One basic property of Treg is that , once activated via the T cell receptor ( TCR ) , they suppress a wide range of immune responses in vitro and in vivo in a contact-dependent manner [11] . Sugimoto et al . [13] initially showed that depletion of CD25+ cells enhanced the proliferation of the remaining PBMC , while Cabrera et al . [7] and several other groups [8]–[10] further showed that CD4+CD25+ T cells isolated from patients' PBMC could suppress the virus-specific CD8+ T-cell response , suggesting that this population contains HCV - specific Treg . The suppressor function of CD4+CD25+ T cells in response to polyclonal stimuli was further analysed recently in a longitudinal acute phase HCV cohort [10] , and it was found that Treg from patients who progressed to persistence were more suppressive than either those from patients who resolved the infection spontaneously or from uninfected healthy donors . In summary , these studies supported the concept that progression from acute to persistent infection is associated with functional changes in the Treg compartment . It is currently unknown , however , to what extend the total Treg pool in HCV-infected individuals is HCV-specific or how Treg react to viral infection as part of the adaptive immune response . Our group has previously reported [14] that a proportion of natural CD25+ cells isolated from the PBMC of HCV patients substantially upregulated CD25 expression in response to HCV peptide stimulation in vitro , and we proposed that virus-specific Treg were primed and expanded during the disease . Somewhat disturbingly , the frequency of the hypothetical HCV-specific Tregs far exceeded the well-documented low frequency of IFNγ producing anti-viral effector T cells in chronic infection [1] , prompting us to seek more insight to these cells in this study .
When the CFSE-CD25+/CD25− co-culture from patients was stimulated for 5 days with the HCV peptide pool ( pp ) , CD25 expression on the CFSE+ fraction was sustained or up-regulated compared to the non-antigen stimulated control ( Figure 1A ) . This observation is reproducible and statistically significant ( p<0 . 05 ) ( Figure 1B ) . When healthy donor cells were cultured under the same conditions , the CD25 expression profile in the HCVpp culture was similar to that of the non-antigen control ( Figure 1A , right panel and Figure 1B ) . In healthy donors , the baseline level of CD25 expression was sometimes higher ( Figure 1B ) compared to HCV patients , but there were no major differences between baseline and HCV pp stimulation . Consistent with the manufacturer's technical datasheet , freshly isolated cells expressed more homogenous and intermediate levels of CD25 ( Figure S1 ) . These data supported our previous observation with core and NS5 peptides [14] , that a proportion of natural CD25+ cells can sustain and/or up-regulate CD25 expression ( now termed CD25+/↑ cells ) in the presence of HCV peptides and this phenomenon is likely to be disease specific . The transcription factor FOXP3 plays a critical role in the development and function of natural Treg , but in humans this molecule is also transiently expressed by activated conventional T cells [15] , [16] . We have recently shown that epigenetic DNA modification of an evolutionarily conserved element within the FOXP3 locus , named Treg-specific demethylated region ( TSDR ) , correlates with a stable Treg phenotype [17] . In the current study , we applied this principle to determine whether the CD25+/↑ cells , which were previously shown to express FOXP3 [14] , are Treg or activated conventional T cells . HCVpp stimulated CFSE-CD25+/CD25− co-cultures were FACS sorted on day 5 into 3 fractions ( Figure 2A ) : CD25+/↑ cells ( P5 , >95% of which are CD4+ , Figure S2 ) , CD25low ( P6 ) and conventional T cells ( P7 ) . Analysis of DNA purified from the above sorted cells by bisulphate sequencing revealed ( Figure 2B , left ) a highly demethylated TSDR in the HCV-responsive fraction ( CD25+/↑ cells , P5 ) , which suggest that these cells are true Treg with stable FOXP3 expression and function . As expected , the TSDR in the conventional T cell fraction ( P7 ) remained highly methylated . The TSDR in the HCV-non-responsive fraction ( CD25low cells , P6 ) showed various degrees of demethylation , which reflects a mixed population of known or unknown cell types . Some P6 cells expressed FOXP3 ( Figure 2B , right ) , but the proportion varied greatly among patients ( from ∼5% to ∼40% , data not shown ) . To further understand the putative disease-associated CD25+/↑ Treg , genome-wide transcriptional profiles were generated on RNA isolated from the cells , cultured and sorted as described above ( Microarray datasets are deposited in Gene Expression Omnibus under series record GSE16576 , and can be reviewed via the following link: http://www . ncbi . nlm . nih . gov/geo/query/acc . cgi ? acc=GSE16576 ) . The Illumina platform was chosen because it requires only 100ng RNA , and given that cell numbers in P5 ( CD25+/↑ ) and P6 ( CD25low , or HCV-non-responsive natural CD25+ cells ) were limited , this allowed us to analyse each patient individually without pooling samples and thus permit rigorous statistical analysis . Of ∼46 , 000 genes ( or probe sets ) analysed , 307 genes were differentially expressed between P5 and P6 , followed by 272 genes differentially expressed between P5 and P7 and 155 genes between P6 and P7 ( Figure 3A ) . Some transcript changes were found in more than one comparison ( Figure 3B ) . This constitutes ∼1% of the entire known transcriptome , while the remaining ∼99% of genes were expressed at similar levels by all three T cell fractions . Table S1 provides the full list of genes that were differentially expressed in P5 compared to P6 or P7 ( Table S1-A ) , and in P6 compared to P7 ( Table S1-A ) . Figure 3C shows selected examples of these genes and demonstrates that the data are highly reproducible . The key Treg signature genes , such as FOXP3 , GITR , CD25 , IL7R and CTLA4 were differentially expressed as expected among the 3 fractions ( Table 1 and Figure 3C ) and provide confidence that the experimental system was able to generate quality data . A number of transcription factors ( Table 1 and Table S1 ) were among the differentially expressed genes . This is not particularly surprising because studies in mice suggested that transcription factors are among the genes regulating or regulated by Foxp3 [18] . Ingenuity Pathway Analysis ( Ingenuity Systems , www . ingenuity . com ) , a literature based online annotation tool , was used to identify the relationships and biological significance of the affected genes ( Figure S3 and S4 ) . This is the first study in which the putative HCV-specific Treg ( CD25+/↑ ) were analysed against the putative non-HCV-specific Treg ( P6 ) , as well as conventional T cells ( P7 ) . Most interestingly , a group of genes ( Table 1 , Figure 3C and Table S1 ) that were known to be implicated in T cell survival or proliferation ( within the top function , immune response , in Figure S3 ) were differentially expressed by P5 compared to P6 and/or P7 . This includes the up-regulation of BCL2 and BCL2L1 ( anti-apoptosis ) , TNFRSF1B and FLT3LG ( promote T cell proliferation and activation ) , IL7 ( T cell survival signal ) and IL32 ( a cytokine released following T cell activation , reviewed in reference [19] ) , and the down-regulation of the pro-apoptosis gene BMF . This pattern suggests that cells in P5 are likely to be more activated and perhaps have a survival advantage over cells in P7 and/or P6 . Figure S3 summarizes the major networks of interactions between these affected genes . It is known ( reviewed in [11] ) that Treg must be activated via their TCR to gain suppressor function , and we applied this principle to test the activation status of CD25+/↑ cells ( N = 3 ) . We used CD4+ conventional T cells as control because the CD25+ cells isolated from PBMC were almost exclusively CD4+ ( Figure S2 ) . The responder cells were a short term autologous CD8+ T cell line driven by HCVpp . The sorted cells ( see Figure 4A for a simple illustration and Figure 2A for technical details ) were added to responder cells at a ratio of 1∶2 and cultured for 7 days . CD25+/↑ cells strongly suppressed HCV-specific CD8+ T cell proliferation , as measured by Ki67 expression on the responder cells ( as the effector frequency is low in HCV patients we found that the Ki67 assay is more sensitive than 3HTdR incorporation in assays with low proliferating cell numbers ) . Cells from P6 suppressed to a lesser degree , reflecting that this was a mixed population of various cells of unknown nature , while conventional CD4+ cells had no suppressive activity ( Figure 4B ) . These results were confirmed in studies with cells from two additional patients ( data not shown ) . In addition to suppression , P5 also expressed a higher level of IL32 mRNA than P6 ( Table 2 , in 3 of 4 patients ) and P7 ( Table 2 , in 4 of 4 patients ) , analysed by qRT-PCR . The role of IL32 in HCV infection is unknown and requires future investigation . Taken together , P5 at the population level correlated with cytokine production and suppressor function , although at present we do not have a reporter molecule that could independently validate the TCR recognition of HCV antigens at the single cell level , a challenging area that is currently being investigated in our laboratories . A number of genes related to B cell phenotype and function , such as toll like receptors , CD19 , CD72 , CD86 , BLNK , etc . were up-regulated in P6 . Interestingly , the same category of genes was also up-regulated in healthy donor natural Treg compared to conventional T cells ( Barry , unpublished data ) . The implication of this is currently unclear . Genes related to CD8+ T effector cell functions ( such as CD8 , perforin and granzymes ) were upregulated in P7 ( Table 1 and Table S1 ) , consistent with the fact that this was the only fraction which contained CD8+ T cells , while the original CD25+ fraction ( now P5 and P6 ) contained mainly CD4+ cells ( Figure S2 ) . The HCV NS3 protein has been proposed as a suitable immunogen for vaccine development [20] . The NS3 peptide array ( provided by BEI resources , ATCC ) consists of 97 overlapping peptides that cover the length of this protein ( Table S2 lists the sequence of each peptide ) . We tested each of the peptides for their ability to induce CD25+/↑ cells following individual peptide stimulation ( N = 8 ) . Our working hypothesis is that such a phenomenon directly or indirectly reflects Treg recognition of HCV antigens . Comprehensive HLA typing of all common loci including class I ( HLA-A , B , C ) and class II ( HLADRB1 ) was performed for each patient by DNA-based sequencing methods ( Figure 5 and Table S3 ) . We found , as expected , that the HLA diversity amongst individuals was high , which may explain why the reactive peptides were not overtly consistent among patients . While the exact location varied among patients , for a given patient , only a few peptides could induce CD25 up-regulation ( Figure 5 ) , which is consistent with our earlier findings with the HCV core protein [14] . Some of the reactive peptides are located close to or overlapped with previously described T cell epitopes ( Table S4 ) . The implications of this need to be further investigated . The mechanisms of the positive responses are unknown but our data suggested that it could be related to the HCV-specific nature of Treg . To test this working hypothesis , we designed a HLA ( DRB1*1301 ) -peptide ( WKCLVRLKPTLHGPTPLL , the p92 ) tetramer , which is , to our knowledge , the only HCV HLA class II -peptide tetramer developed based on non-T-helper responses . Compared to a HLA mismatched control , more tetramer+ cells were detected in the patient with DRB1*1301 ( 7% in SA67 compared to 1 . 2% in PH 35 in Figure 6A ) , suggesting the staining signal is likely true . The control tetramers 0701-p92 ( mismatched HLA loaded with the same peptide ) and 1301-empty ( the correct HLA but loaded with no peptide ) showed minimal background staining , further suggesting that the staining is genuine . Importantly ( Figure 6B ) , a high proportion ( >60% ) of the tetramer+ Treg cells were CD25+ , while the vast majority ( >90% ) of tetramer+ T-helper cells were CD25− , supporting our hypothesis and also implying that the tetramer+ T-helper are likely not functional ( given that CD25 is an activation marker for conventional T cells ) .
Conventional protocols to culture human Treg usually involve long term expansion in the presence of high doses of rhIL2 . We have previously described a novel co-culture system [14] , which we believe to be more physiological . In this system , PBMC-derived CD25+ cells are labelled with CFSE , mixed with CD25− cells from the same donor and finally stimulated with HCV peptides . This approach , used throughout the current study , allowed us to identify a HCV-specific response within the natural CD25+ cell population by observing their response to HCV antigen with conventional T cells as an internal control . We found that the CD25+ population isolated from PBMC of HCV patients , despite a failure to proliferate ( which is consistent with the literature that Treg are hypo-proliferative in vitro ) , responded to HCV peptide stimulation by sustaining and/or up-regulating CD25 surface expression , a phenomenon that does not occur , or at least to a lesser degree , in healthy donors . It is not known if human Treg can down regulate CD25 expression in vitro in the absence of antigen , but we think this can not be excluded . In naïve inbred pathogen-free mice , CD25+ cells isolated from PBL are almost entirely Foxp3+ natural Treg , but in adult humans , the CD25 expression level is more heterogeneous , as this population is expected to contain activated effector T cells and other known or unknown cell types , particularly during infection . The transient expression of FOXP3 by activated human conventional T cells [15] , [16] further complicates the interpretation of human data . We found that natural Treg and Treg converted in vivo under tolerogenic conditions [21] exhibited a completely demethylated TSDR , whereas activated conventional T cells and TGF-β induced Treg contained almost 100% methylated CpG motifs . We therefore proposed the TSDR methylation status as a reliable criterion for the identification of natural and stable subsets of induced Tregs [17] . Using the same criteria , we confirm here that the CD25+/↑ cells in our culture are not activated conventional T cells or TGF-β converted unstable Treg , but are “hard-wired” stable Treg . Since the origin of human Treg is unclear [22] , [23] , CD25+/↑ cells could either belong to the natural Treg lineage , or be converted from peripheral HCV-specific conventional T cells during the infection , but if it is conversion , the conversion is thorough , as demonstrated by the epigenetic imprint . More Treg were found in HCV-infected liver than periphery blood [24] , where a surprisingly high proportion ( ∼80% ) of T cells expressed FOXP3 . In vivo expansion of HCV-specific Treg is possible , as Treg from a HCV-experienced chimpanzee had a lower TCR excision circle content compared to naïve animals [25] . The induction and expansion of HCV-specific Treg could have profound effects on the quantity and quality of the anti-viral effector T cell responses . We next generated gene expression profiles of CD25+/↑ cells ( P5 ) , using CD25low ( P6 ) and conventional T cells ( P7 ) as controls , to understand the molecular program that governs the role of these cells . In addition to typical Treg gene patterns , which are either consistent with our FACS data or with the literature , P5 also expressed genes patterns that are less known , such as the survival profile . In an independent study ( Barry , et al , unpublished data ) we generated transcriptional profiles for ex vivo isolated ( FACS sorted CD25high cells ) resting , as well as polyclonal stimulated Treg and conventional T cells from healthy donors . Comparing our current dataset to the healthy donor dataset provides hints as to transcriptional changes which could be unique in HCV patients and thus likely to be associated with HCV infection . BCL2 , BMF , IL7 , IL32 , CISH , CCL5 , CCR7 , IFNαR2 , IRF4 and IRF8 ( Table 1 and Table S1 ) are all among this “unique” list , and these genes are known to be critical in regulating cell survival or play important roles in immune responses against pathogens . Development of these data is necessary and is currently ongoing in our laboratories . It was recently reported that the gene profile of ex vivo isolated total Treg from HCV patients was very similar to that of healthy donors [26] , as only 5 genes were differentially expressed between the two and the change ranged from 0 . 4 to 2 . Interestingly , none of these 5 genes was identified in our experiments . We think that Treg and non-Treg compartments are both likely to be affected by the disease , a detail which would not be revealed by comparing total Treg of patients and healthy donors . The continued expression and/or up-regulation of CD25 on a proportion of Treg in response to HCV peptide stimulation in vitro is an event associated with HCV infection , because it does not occur , or is greatly reduced , in healthy donors . This could be a consequence of TCR engagement by the HCV antigen in the context of the peptide/HLA complex , a view supported by the suppression assay data , or alternatively , IL2 ( and/or other soluble factors ) produced by effector T cells within the co-culture may affect CD25 expression on Treg independently of antigen recognition . In the latter scenario , the apparent antigen specificity of Treg is likely to reflect the antigen specificity of the effector T cells . However , the effector frequency within PBMC was very low , as suggested by the literature ( reviewed in reference 2 ) . Supernatant IL10 and IFN-γ levels ( measured using Cytokine Bead Array , BD Biosciences ) also did not consistently correlate with culture conditions viz . the CFSE-CD25+/CD25− co-cultures and the CD25− PBMC cultures with or without antigen , from patients or from healthy donors ( data not shown ) , and IL2 was generally below the detection limit ( data not shown ) . This is consistent with our microarray data , as none of the key gene signatures for Th1/Th2 , Th3 and Th17 ( IL2 , IL4 , IL10 , IFNγ , IL12p70 , IL17 , TGFβ , IL6 , etc . ) were differentially expressed among the fractions upon HCV antigen stimulation . Thus it is unlikely that the common soluble factors produced by conventional T cells or other antigen non-specific cells in culture could determine the apparent Treg responsiveness . Ideally we should use a Treg-specific activation marker for epitope mapping , but since there is no such marker we used CD25 as a surrogate marker . In almost every patient , the most reactive NS3 peptide induced higher CD25 expression on Treg compared to anti-CD3 ( Figure 5B ) . Given that anti-CD3 induced more conventional T cells to express CD25 than any of the peptides ( Figure 5B and data not shown ) , these data support the concept that soluble factors alone do not completely correlate with the magnitude of the Treg response , as the level of IL2 in the anti-CD3 culture must be otherwise sufficient to achieve the highest CD25 expression . We attempted to match the reactive peptides against published data on effector T cell epitopes , but found this difficult , as studies using class II tetramers only focus on a few epitope/DR pairs , while in studies which did not use tetramers the HLA typing data were incomplete or missing . Further validation of the putative Treg epitopes and their HLA restriction are required , but nevertheless , our data show that the breadth of the reactivity is rather narrow , while the response itself is robust . Due to the lack of any Treg specific surface marker and a simple functional readout for these cells , it has not been possible to develop tetramers that are restricted to Treg . Using two class II HLA tetramers previously developed based on T-helper responses , Heeg et al [27] detected FOXP3+ cells during acute infection and reported that the frequency of tetramer+FOXP3+ cells was low and did not correlate with disease progress or outcome . It is unclear at present how this reflects a global picture of Treg/Teff balance , as it is not known to what extent the Treg repertoire overlaps with that of Teff , or if Treg and Teff clones of the same antigen specificity would expand/contact with the same kinetics . Unfortunately , our tetramer data is limited at present and could not answer these questions . Further studies are required , but since it is impossible to develop tetramer for every T cell epitope , we believe that it is important to develop a higher throughput or a more practical Treg antigen specificity readout so that a more global picture can be obtained . This study opens a window to explore the role of Treg and their target antigens in a chronic viral infection of humans . The target antigens recognised by the FOXP3+ lineage in humans are largely unknown and systems to guide the discovery of these antigens would benefit future studies in HCV vaccines and immunotherapy .
The study was approved by the Alfred Hospital Ethics Committee and the Victorian Department of Human Services Human Research Ethics Committee . Written informed consent was obtained from each subject . HCV-infected participants ( N = 31 ) were recruited from the Alfred Hospital , Melbourne and from an ongoing study of hepatitis C virus in the social networks of injecting drug users . All participants were HCV mono-infected , with either genotype 1a or genotype 3a viruses , and one participant resolved the infection spontaneously . A few patients were treated previously ( unsuccessfully ) with interferon/ribavirin and the remainder were untreated . Healthy donors were represented by local volunteers or blood donors from the Australian Red Cross Blood Transfusion Service , Melbourne Branch . The HCV peptide array , which contains 18-mer peptides overlapping by 11aa covering the entire HCV polyprotein , for genotype 1a and 3a were provided by BEI Resources , ATCC . A peptide pool ( pp ) working stock ( containing 100 µg/ml of each peptide ) was prepared in DMSO/RPMI . The final concentration of HCVpp within the culture was 0 . 2 µg/ml in initial experiments and 0 . 15 µg/ml for subsequent experiments , or as indicated . PBMC from patients or healthy donor controls were separated by Ficoll Paque centrifugation and CD25+ cells were isolated from PBMC using CD25 microbeads ( MiltenyiBiotec ) according to the manufacturer's instructions . The CD25+ cells , typically 1–2% of total PBMC , were labelled with CFSE ( Sigma-Aldrich ) and mixed back with unlabeled CD25-depleted PBMC at a ratio 1∶10 . The CFSE-CD25+/CD25− co-culture was stimulated with or without genotype matched HCVpp in RPMI-1640 , 2 mM L-glutamine , 100 IU/mL penicillin-streptomycin ( Invitrogen ) and 5% human AB serum ( MP Biosciences ) in 24-well tissue culture plates ( Interpath , Australia ) . Cells were harvested on day 5 for flow cytometry analysis or sorting . In some experiments , culture supernatants were collected prior to cell harvesting for cytokine analysis at later stage . In general , fluorescent dye-conjugated antibodies and isotype controls were purchased from BD Biosciences . PE-conjugated anti-human FOXP3 , isotype control and FOXP3 staining buffer set were purchased from eBiosciences . Intra-nuclear staining of FOXP3 , as well as Ki67 , was performed according to the manufacturer's instructions . Flow cytometry was performed using a FACScalibur flow cytometer ( BD Biosciences , ) and Cellquest software . For data analyses , an initial lymphocyte gate was set based on SSC/FSC and additional gates introduced as required . Results are presented as the percentage , or mean fluorescent intensity ( MFI ) of positively stained cells within the gated population . Sorting of HCV peptide-stimulated CFSE-CD25+/CD25− co-cultures from HCV patients was performed using a FACSaria located in a PC3 facility . The cultures were sorted on day 5 into 3 fractions as specified , based on their CFSE labelling and CD25 expression . The primary gate was set on lymphocytes based on SSC/FSC and an additional CD3 gate ( for methylation analysis and microarray ) or CD4 gate ( for in vitro suppression assay ) was introduced to the CFSE- population to refine the conventional T cell population . For this series of experiments , we used cells from male patients , as this overcomes the potential X-chromosomal inactivation of one FOXP3 allele , which usually affects the methylation analysis of Treg in females . Genomic DNA was isolated from sorted cells ( Figure 2A ) using NucleoSpinTissue XS kit ( Macherey & Nagel , Düren , Germany ) following the protocol for cultured cells . Bisulfite treatment of genomic DNA was performed as described previously [28] TSDR-primers ( 5′ to 3′ direction ) p-TGTTTGGGGGTAGAGGATTT and o-TATCACCCCACCTAAACCAA , amplifying Amp5 [17] were used for bisulphite-specific PCR and sequencing reactions . The primers “p” and “o” produce amplicons based on the +1 strand . PCR was performed in a final volume of 25 µl containing 1x PCR Buffer , 1U Taq DNA polymerase ( Qiagen ) , 200 µM dNTPs , 12 . 5pmol each of forward and reverse primers , and 7ng of bisulphite-treated genomic DNA at 95°C for 15 min and 40 cycles of 95°C for 1 min , 55°C for 45 sec and 72°C for 1 min with a final extension step of 10 min at 72°C . PCR products were purified using ExoSAP-IT ( USB Corp . ) and sequenced using the PCR primers and the ABI Big Dye Terminator v1 . 1-chemistry ( Applied Biosystems ) followed by capillary electrophoresis on an ABI 3100 genetic analyzer . AB1 files were interpreted using ESME . Total RNA from sorted cells ( P5 = CD25+CFSE+ , P6 = CD25−CFSE+ and P7 = CD3+CFSE− , as illustrated in Figure 2A ) was isolated using RNeasy Kit ( QIAGEN Australia ) according to the manufacturer's instructions . The RNA quality was ascertained by the Agilent Bioanalyser 2100 using the NanoChip protocol . The microarray experiments were performed , according to the technical manual from Illumina , by the Australia Genome Research Facility . In brief , 100 ng RNA was amplified using the Illumina Total Prep RNA amplification kit ( Ambion Cat . No . IL1791 ) to generate biotinylated cRNA . An aliquot ( 1 . 5 µg/30µl ) of the labeled cRNA for each sample , prepared in a probe cocktail that included GEX-HYB Hybridization Buffer , was hybridized to an Illumina Sentrix Human-6 Expression BeadChip-v2 . 0 at 58°C for 16 hours . After hybridization , the chips were washed , coupled with streptavadin-Cy3 and scanned in the Illumina BeadArray Reader . The scanner operating software , BeadStudio , converts the signal on the array into a TXT file for downstream analysis . Data analysis and visualization were performed using BeadStudio Gene Expression Module v3 . 3 software ( Illumina Inc . , San Diego , CA ) . With Illumina gene expression array , each probe is measured at least 30 times independently on random distributed beads . This large number of technical replicates allows robust estimation of the hybridization intensity and the measurement error for each probe . The signal for each probe or probe set ( gene ) was averaged and the background ( the average signal from the large number of randomly distributed negative control beads ) subtracted , and then normalized using quantile algorithms that account for variations between probes and between chips . A detection P value , calculated by comparing the distribution of the transcript signal to that of the negative control signal , was set at ≤0 . 001 to identify transcripts that were expressed ( with a confidence of ≥99 . 9% ) above background . Genes with detection P value≤0 . 001 in at least one of the three fractions were selected for further analysis . To detect changes in gene expression between samples , the differential P value ( Diff Pval ) was calculated using the Illumina custom error model , which allows 5% false discovery rate being automatically adjusted . The cut off for the Diff Pval was set at ≤0 . 05 ( a confidence of ≥95% that the given gene is expressed at different levels between the sample and control ) . We used the Ingenuity Pathway Analysis online software ( Ingenuity Systems , www . ingenuity . com ) to help further group the genes in term of networks and functions . RNA was isolated from sorted cells as above . Real time RT-PCR assay was performed using Mx3000P QPCR system ( Agilent Technologies ) . The gene expression assays for IL32 and house keep control GAPDH , as well as One-Step Master Mix Reagents , were purchased from Applied Biosystems ( Foster City , CA , USA ) . The cycle conditions are 30 min at 48°C for cDNA synthesis , 10 min at 95°C , followed by 50 cycles of 15 sec 95°C , 60°C 1 min . Data were analysed using MxPro software supplied by the manufacturer . The co-culture was sorted by FACSAria to CD25+CFSE+ ( hypothetical HCV-specific Treg ) , CD25−CFSE+ ( Treg of other specificity and other unavoidable contaminating cells ) and CD4+CFSE− ( conventional CD4+ ) in a PC3 facility . The target cells were represented by an autologous HCV-specific CD8+ T cell line , for which an equal number of CD8+ T cells and CD14+ monocytes were mixed and cultured in the presence of 0 . 15 µg/ml HCVpp for 5 days . The in vitro assay was set up in U-bottom 96-well plates in triplicate . Each well , in a final volume of 200 µl , contained 1×105 sorted cells , 2×105 target cells and 2×104 feeder ( autologous immature dendritic cells generated as described previously [29] ) and the antigens HCVpp ( 0 . 1 ug/ml final of each peptide ) . At the end of the culture period ( day 7 ) , cells were pooled from the triplicate wells , stained for Ki67 expression and analysed by flow cytometry , gating on CD8+ lymphocytes ( note that the sorted cells in this experiment were CD4+ ) . The CFSE-CD25+/CD25− co-cultures were set up essentially as described above , except in a 96 well format , containing 2×105cells in 200 ul medium . Each individual NS3 peptide ( Table S1 ) , genotype-matched , was added to each different well at 10 µg/ml final . Anti-CD3 ( clone 32-2A2 , Mabtech ) was used as a positive control at 0 . 1 µg/ml final . The cultures were harvested on day 5 and analysed for CD25 expression by flow cytometry . The criteria for reactive peptides were described previously [14] . The p92 , WKCLVRLKPTLHGPTPLL , is located towards the C terminal of NS3 of HCV genotype 3a ( Table S1 ) . PE conjugated HLA class II-peptide tetramer complexes ( DRB1*1301-p92 , DRB1*0701-p92 and DRB1*1301-empty ) were synthesized at the Benaroya Research Institute , USA . For staining , the CFSE-CD25+/CD25− co-culture was harvested at day 5 , washed and resuspended in fresh RPMI medium ( same as for culture but without HCV peptides ) at 1×105 cells in 50 ul per well . To each well 1 ul of a tetramer was added and the cells incubated for 3 h at 37°C , then 30 min at 4°C to stain surface molecules CD25 and CD4 . High-resolution HLA Class I and II typing was performed by direct DNA sequencing methods as previously described [30] . Ambiguities were resolved following sequencing with allele-specific subtyping primers . Sequence electropherograms were analysed using Assign™ ( Conexio Genomics ) . Allele assignment was based upon identity at exons 2 and 3 and consistently allocated for the most common expressed allele in the relevant population . | Hepatitis C virus persistently infects ∼3% of the world population , leading to life threatening liver diseases and liver failure . It is not well understood why the human immune system often fails to clear the virus , although it is likely multi-factorial . It is accepted that effector T cells are critical for clearing infections , but their function can be suppressed by the somewhat elusive regulatory T cells . Our hypothesis , supported by new data , is that a proportion of the regulatory T cells are specifically stimulated by the virus and that these cells are a stable cell population . We find evidence that these suppressive cells have a distinct set of genes activated and importantly might have a survival advantage over effector T cells , which helps to explain why natural regulatory T cells may influence the outcome of HCV infection . We propose that the new information provides a better explanation of chronic HCV infection and will let us focus on the key experiments to test the hypothesis and to design better treatments . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"immunology/cellular",
"microbiology",
"and",
"pathogenesis",
"immunology/immunomodulation",
"immunology",
"immunology/immunity",
"to",
"infections"
] | 2009 | Analysis of FOXP3+ Regulatory T Cells That Display Apparent Viral Antigen Specificity during Chronic Hepatitis C Virus Infection |
Leprosy remains a public health problem in Brazil . Although the overall number of new cases is declining , there are still areas with a high disease burden , such as Pará State in the north of the country . We aim to predict future trends in new case detection rate ( NCDR ) and explore the potential impact of contact tracing and chemoprophylaxis on NCDR in Pará State . We used SIMCOLEP , an existing individual-based model for the transmission and control of M . leprae , in a population structured by households . The model was quantified to simulate the population and observed NCDR of leprosy in Pará State for the period 1990 to 2014 . The baseline scenario was the current control program , consisting of multidrug therapy , passive case detection , and active case detection from 2003 onwards . Future projections of the NCDR were made until 2050 given the continuation of the current control program ( i . e . baseline ) . We further investigated the potential impact of two scenarios for future control of leprosy: 1 ) discontinuation of contact tracing; and 2 ) continuation of current control in combination with chemoprophylaxis . Both scenarios started in 2015 and were projected until 2050 . The modelled NCDR in Pará State after 2014 shows a continuous downward trend , reaching the official elimination target of 10 cases per 100 , 000 population by 2030 . The cessation of systematic contact tracing would not result in a higher NCDR in the long run . Systematic contact tracing in combination with chemoprophylaxis for contacts would reduce the NCDR by 40% and bring attainment of the elimination target two years forward to 2028 . The NCDR of leprosy continues to decrease in Pará State . Elimination of leprosy as a public health problem could possibly be achieved around 2030 , if the current control program is maintained . Providing chemoprophylaxis would decrease the NCDR further and would bring elimination forward by two years .
Leprosy , also known as Hansen Disease , is caused by Mycobacterium leprae . Every year more than 210 , 000 new cases are detected worldwide [1] . After India , Brazil has the second largest number of leprosy patients detected . In 2014 , there were 31 , 064 new cases reported , mostly from the Amazon region [1] . Brazil was unsuccessful in bringing the prevalence of leprosy below the official WHO’s ‘elimination as public health problem’ goal of less than 10 cases per 100 , 000 population by the year 2000 [2] . In 2010 the leprosy prevalence rate in Brazil was 15 . 6 per 100 , 000 and the new case detection rate ( NCDR ) 18 . 2 per 100 , 000 , although the NCDR had dropped by about 35% between 2001 and 2010 [3] . In 2011 , the Ministry of Health of Brazil ( MHB ) had set the year 2015 as the new target year to achieve elimination of leprosy [3] . This target is now part of the agenda of an ambitious governmental program for reducing poverty in Brazil , which is called “Brasil sem miséria” ( Brazil without misery ) [3] . The spatial distribution of leprosy in Brazil is known to be heterogeneous [4 , 5] . Therefore , the MHB selected 243 municipalities as priority for actions aimed at reducing the leprosy burden in order to achieve the elimination target by 2015 . Most of these municipalities are located in the Amazon region , and Pará State alone has 20% of all priority municipalities [3] . The NCDR of leprosy in Brazil between 1990 and 2010 shows a characteristic trend of an increasing rate up to 2003 followed by a gradual decrease afterwards . A similar trend can also be observed in Pará State , but with a NCDR that is more than two times higher ( 45 . 8 per 100 , 000 in 2014 ) [6] . Based on a simple extrapolation of the NCDR trend , policy makers expect elimination of leprosy to be achieved by 2015 in Brazil as a whole , but this will certainly not be the case for Pará State . Furthermore , simple extrapolation of trends fails to take into account dynamic processes that underlie the observed trend , such as the importance of contacts and in particular of households for the transmission of M . leprae [7] . To assess the future leprosy NCDR trend in Pará State in the Amazon Region of Brazil , we will apply the individual-based ( or microsimulation ) model SIMCOLEP [8] . It models the transmission and control of leprosy in a population structured by households and can be used to explore the potential impact of interventions targeted at households [8] . SIMCOLEP has been used to study the leprosy epidemic in northwest Bangladesh , India , Brazil and Indonesia [8 , 9] . It has also been applied to assess the various strategies for leprosy control such as contacts tracing , the effect of Bacillus Calmette-Guérin ( BCG ) vaccination in infants , and also possible future strategies such as chemoprophylaxis and early diagnosis of subclinical infection [10] . The impact of such targeted strategies is estimated in terms of leprosy incidence at the population level [8 , 10] . We therefore also address the question to which extent contact tracing and chemoprophylaxis for leprosy contacts contribute to the elimination of leprosy in Pará State . In summary , the aims of this study are: 1 ) to predict future trends in the NCDR; and 2 ) to explore the potential impact of contact tracing and chemoprophylaxis on the NCDR in Pará State .
The modelled household structures in the population were fitted to the observed distribution of household size in Brazil , obtained from IBGE [12 , 13] . In order to fit the household structure , parameters involving movements of individuals between households were calibrated ( see Table 1 ) . We only calibrated the following parameters: fraction of non-married males moving , fraction of individuals moving to child after becoming widowed , fraction of moving individuals that create their own household , and the weighting function to determine the household size to which individuals move . For all other parameters , we assumed that they would be similar to previous modelling work [8] . The simulated distribution of household size did not show any significant difference with the observed distribution ( p>0 . 05; χ2-test ) . Subsequently , we fitted the leprosy situation of the model to the observed NCDR in Pará State between 1990 and 2014 . NCDR data were obtained from the Sistema de Informações de Agravos de Notificação ( Sinan ) database [6 , 20] . Sinan is a national database for reporting communicable diseases in Brazil . We only calibrated detection delays and the contact rate in the general population ( cpop ) [9] . Detection delays were estimated following the NCDR trend in the data . As a starting point the most recent detection delay was set to 3 years [21] . Table 1 provides an overview of the estimated delays over time . The contact rate in the general population was calibrated such that the modelled NCDR would match the observed NCDR between 1990 and 2014 . The contact rate within households ( chh ) was set to the optimal value ( chh = 0 . 98 ) of previous work [8 , 9] . The best fit was determined by a log-likelihood function assuming a Poisson distribution . The model outcomes are based on an average of 100 simulation runs . A detailed description of the fitting procedure is found in Fischer et al . [8] . Simulations were run until the year 2050 to predict the future trends in leprosy . The baseline scenario represents the current leprosy situation and the existing control program as described above in Pará State . Additionally , we investigated the potentially impact of two scenarios of future control of leprosy: 1 ) discontinuation of contact tracing; 2 ) continuation of current control ( including contact tracing ) in combination with chemoprophylaxis . The first scenario was tested , because in practice coverage rates of contact tracing easily fluctuates [16] . To assess the importance of contact tracing alone for transmission , we tested to which extent discontinuation of contact tracing ( an extreme measure ) would potentially impact the NCDR trend in Pará State . In the second scenario , we assessed the impact of administering chemoprophylaxis to contacts given the continuation of contact tracing . Chemoprophylaxis ( i . e . single dose of Rifampicin ) was only given once after examination , assuming that it cures 50% of subclinical cases [22] . We further assumed that the probability for an individual to comply was 75% . Both scenarios started in 2015 and were projected until 2050 .
Fig 1 shows the results of the simulated leprosy NCDR against the observed NCDR in Pará State between 1990 and 2014 . The fitted values are found in Table 1 . The simulation provided a good fit to the observed data , and based on this fit it was possible to predict the future trend up to 2050 . Our prediction shows that the NCDR in Pará State continues to decline with an annual decrease of 8% to 2 . 17 per 100 , 000 cases ( 90% CI: 1 . 48–2 . 89 ) in 2050 . This corresponds with a drop in terms of annual new cases to around 400 new cases . The elimination target ( 10 cases per 100 , 000 population ) would be reached around 2030 . Discontinuation of contact tracing in Pará State would not increase the NCDR in the long run relative to the baseline scenario ( Fig 2 ) . In the first years a drop in the NCDR is observed , because more cases are missed as result of the discontinuation of contact tracing . These cases are expected to be detected at a later stage through passive detection . Afterwards , it slowly rises to the level of the baseline scenario . The relative impact of chemoprophylaxis compared to the baseline scenario shows a substantial impact on the NCDR of leprosy . In the long run the predicted NCDR is up to 40% lower than the baseline . The NCDR drops immediately below the level of the baseline scenario due to the cured subclinical cases . Over time the effect increases because it would prevent new infections . Fig 3A and 3B focus on the impact of chemoprophylaxis in comparison with the baseline . The NCDR of leprosy is predicted to decrease to around 1 . 13 per 100 , 000 ( 90% CI: 0 . 50–1 . 98 ) in 2050 , indicating an annual decrease of 9 . 5% ( Fig 3A ) . Moreover , the elimination target would be reached around 2028 , which is two years earlier than in the baseline scenario . Fig 3B shows the cumulative number of new cases detected . In total 40 new cases per 100 , 000 could be avoided over the whole period compared to the baseline . It also illustrates that chemoprophylaxis would prevent new infections on top of the cured subclinical infections over time .
The new case detection rate of leprosy in the whole country of Brazil as well as in Pará State has been showing a downward trend since 2005 . With the SIMCOLEP model we were able to reproduce the observed NCDRs between 1990 and 2014 . The modelled trend in Pará State after 2014 shows a continuous downward trend , reaching elimination ( less than 1 per 10 , 000 ) by 2030 . Systematic contact tracing alone would seem to have no additional impact on the NCDR in the long term , but contact tracing together with chemoprophylaxis for contacts would add a significant impact to the reduction of new cases in the long term . Our model predictions underscore the importance of the maintenance of early diagnosis ( i . e . short detection delay ) , treatment of newly found cases with MDT , and BCG vaccination coverage; these are our baseline conditions . We have shown that ceasing contact tracing will not change the NCDR trend much . Previous work found similar findings [10] . The reason for this is that patients are only found when they have symptoms , meaning that transmission might already have been taken place during the asymptomatic state . Hence , systematic contact tracing alone does not have a noticeable effect on the ongoing transmission of M . leprae and subsequent leprosy disease . However , contact tracing in itself will still benefit the individual with leprosy , because he or she will be detected and treated earlier . This would interrupt the disease process in most cases , preventing severe nerve damage ( grade-2 disability ) [23] . Systematic contact tracing , however , is effective when combined with an additional preventive intervention . Providing contacts with chemoprophylaxis showed to further reduce the number of new leprosy cases up to 40% in the long run . It also supports attaining the target of zero transmission in the population [24] . Chemoprophylaxis is yet to be introduced systematically in Brazil , but the results of our study indicate that this intervention will provide substantial benefit to the leprosy control program through reduced costs and disease burden . However , the feasibility to carry out this program on a large scale , such as in Pará State , remains a concern . The Ministry of Health is currently stimulating projects for evaluating contact chemoprophylaxis , but this is still on a small scale [2] . Our model predictions are based on the reported NCDRs in Brazil . The validity of our results therefore depends on the reliability of these reported data . In fact , the NCDR did not change much between 1990 and 2010; 2 . 0 per 10 , 000 in 1990 and 1 . 8 per 10 , 000 in 2010 . During this period however , there was considerable variability in the trend , with marked increases and decreases . In Pará State , the NCDR was twofold higher than in Brazil as a whole: 4 . 6 per 10 , 000 in 1990 and 5 . 0 per 10 , 000 in 2010 , and also showed marked variability [2 , 3] . The variability in the trend is explained largely by operational factors influencing the leprosy control program , and these have been taken into account in the model . The gradual implementation of SINAN can fully explain the sharp increase in detected cases from 1990 to 1997 , even though the underlying rate of all new cases ( detected and undetected ) may have remained the same . In the mid-1990s , another important intervention occurred in Brazil . Since 1994 the Family Health Program was introduced , with a focus on primary health care and health promotion , including active case finding as a main health surveillance methodology [25] . Leprosy , until then a disease diagnosed mainly by dermatologists , became an object for primary health care from 2000 onwards . This development is likely to be reflected in the second increase of the NCDR in the first decade of the current century , and particularly in Pará State . The sharp increase of the NCDR in the late 1990’s up to 2005 due to increased active case detection and registration indicates an enormous backlog of previously undetected cases that were now being found and treated . The reduction seen after 2014 would then be a return to a basic situation with equilibrium in terms of incidence and prevalence , but in a downward trend [25] . Our modelling results support the observed reduction in new case detection after 2014 . A remaining threat to the elimination of leprosy is missing cases . Previous work has shown that the prevalence of undiagnosed cases ( missing cases ) is substantial [26–28] . The data used in this study very likely underestimate the number of cases in Pará State . Because our predictions depend on available data , the problem of missing cases is also present in our predictions . It is likely that the actual number of new cases is higher than predicted . Also , the coverage of contact tracing , which was set to 59% based on a report of the Ministry of Health [16] , might be too optimistic . We therefore conducted a sensitivity analysis with a more realistic coverage of 40% to assess the impact of chemoprophylaxis on NCDR . Results showed that lowering the coverage would not affect our model predictions significantly ( S1 Fig ) . The results of our study not only depend on the validity of the data , but also on the availability of the data and the model´s underlying assumptions regarding leprosy . For example , data about the prevalence or incidence by household size and by contact were not available for Pará State . These data are needed to fit the contact rates within households directly . As a solution we used the optimal value from previous work [8] , assuming that the rate at which people have effective contact with other household members would not differ much between countries and regions . Also , our model assumes that susceptibility was fully randomly determined . Previous work has shown that this is a valid assumption , but could not rule out other mechanisms , such as genetic inheritance [8] . It has also shown that assuming random susceptibility will provide the most optimistic results . Our future projections did not account for any adverse events in the future , such as a famine , or any changes in leprosy control , which may alter the course of the epidemic . It is likely that in areas with declining NCDR , the incentive to find new cases may disappear [29] . Based on the observed data for Pará State , our modelling study confirms that the leprosy incidence appears to be decreasing in Brazil . Nevertheless , additional interventions should be implemented in view of the constant number of new cases detected in Pará state . Achievement of elimination could be brought forward with a number of years through systematic contact tracing combined with the application of chemoprophylaxis . | Leprosy remains a public health problem in Brazil . With over 30 , 000 new cases detected annually , it has the second-largest number of leprosy cases detected worldwide . Although the overall number of new cases is declining , there are still areas with a high disease burden , such as Pará State . In this study , we used the individual-based model SIMCOLEP to predict the future trend of the new case detection rate of leprosy ( NCDR ) in Pará State given the current control strategy . Additionally , we explored the potential impact of contact tracing and chemoprophylaxis on the NCDR . Our results show that the NCDR continues to decrease in Para State , reaching elimination around 2030 . Contact tracing alone would not further reduce the NCDR . However , when contact tracing is combined with chemoprophylaxis the NCDR would be reduced up to 40% and elimination could be brought forward by two years . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"mycobacterium",
"leprae",
"medicine",
"and",
"health",
"sciences",
"immunology",
"tropical",
"diseases",
"geographical",
"locations",
"india",
"preventive",
"medicine",
"bacterial",
"diseases",
"mathematics",
"extrapolation",
"neglected",
"tropical",
"diseases",
"vaccinati... | 2016 | Leprosy New Case Detection Trends and the Future Effect of Preventive Interventions in Pará State, Brazil: A Modelling Study |
With the emergence of leishmaniasis in new regions around the world , molecular epidemiological methods with adequate discriminatory power , reproducibility , high throughput and inter-laboratory comparability are needed for outbreak investigation of this complex parasitic disease . As multilocus sequence analysis ( MLSA ) has been projected as the future gold standard technique for Leishmania species characterization , we propose a MLSA panel of six housekeeping gene loci ( 6pgd , mpi , icd , hsp70 , mdhmt , mdhnc ) for investigating intraspecific genetic variation of L . ( Viannia ) braziliensis strains and compare the resulting genetic clusters with several epidemiological factors relevant to outbreak investigation . The recent outbreak of cutaneous leishmaniasis caused by L . ( V . ) braziliensis in the southern Brazilian state of Santa Catarina is used to demonstrate the applicability of this technique . Sequenced fragments from six genetic markers from 86 L . ( V . ) braziliensis strains from twelve Brazilian states , including 33 strains from Santa Catarina , were used to determine clonal complexes , genetic structure , and phylogenic networks . Associations between genetic clusters and networks with epidemiological characteristics of patients were investigated . MLSA revealed epidemiological patterns among L . ( V . ) braziliensis strains , even identifying strains from imported cases among the Santa Catarina strains that presented extensive homogeneity . Evidence presented here has demonstrated MLSA possesses adequate discriminatory power for outbreak investigation , as well as other potential uses in the molecular epidemiology of leishmaniasis .
Leishmaniasis , a vector-borne disease caused by protozoan parasites of genus Leishmania [1] , represents one of the highest disease burdens among the neglected tropical diseases in developing nations [2] . While not often fatal like the visceral form , the cutaneous form of the disease contributes substantially to leishmaniasis disease burden as it requires a lengthy and costly treatment regimen , results in apparent scarring , and can progress to a severely disfiguring mucosal form [1] . In recent years , leishmaniasis outbreaks have been described with increasing frequency [3]–[5] , including those in sub-tropical regions or regions not previously endemic across the global [6]–[8] . In Brazil , beginning in 2005 , an outbreak of human cutaneous leishmaniasis occurred in the southern Brazilian state of Santa Catarina , where the disease had not been observed previously as endemic . Overtime , cutaneous leishmaniasis has emerged in the region with evidence of a continued transmission cycle [9] . The species responsible for this outbreak has been incriminated as Leishmania ( Viannia ) braziliensis [9] , the most widely distributed Leishmania species in Brazil to date [10] , [11] . However , many questions still remain regarding the outbreak , such as: is one main strain or various strains responsible for the outbreak; is the emergence of L . ( V . ) braziliensis in the region a recent event; and how are Santa Catarina strains related to other strains in Brazil ? A wide range of molecular tools are available for the investigation of molecular epidemiology of leishmaniasis , but choosing which method and/or markers to use continues to be a challenge [12] . Particularly for New World species , open access databases based on gold-standard genetic markers have not been developed . Currently , outbreak investigation of leishmaniasis , mainly conducted for visceral leishmaniasis outbreaks caused by L . ( Leishmania ) donovani species complex [13] , [14] , commonly employs multilocus microsatellite typing ( MLMT ) . This technique has been proven to discriminate at the intra-species level [15] with high discriminatory power and is useful for determining outbreak strain origin when a database of MLMT strains is available for the Leishmania species of interest [13] . At the present moment , an open access MLMT database for L . ( V . ) braziliensis , has not been developed . The high discriminatory power of this technique has its drawbacks depending on the type of epidemiological question or analysis . In some cases , almost 20 “different” genotypes can be identified in one focus [13] , [16] , [17] . Dividing the isolates into many different genotypes reduces the statistical power of analyses involving epidemiological variables , such as clinical and demographic characteristics of the patient . Such reductions in statistical power greatly reduce the ability of researcher to conclude the relationship of factors like clinical form and disease virulence with a particular genotype . Thus , epidemiological tools with appropriate discriminatory power , increased reliability and inter-laboratory reproducibility and comparability urgently are required . With these characteristics in mind , the method of multilocus sequence analysis ( MLSA ) provides a promising alternative . Projected as the future gold standard species typing method [12] , MLSA involves sequencing a panel of house-keeping gene loci based on the panel of enzymes used in MLEE [18] . Several markers of these conserved regions have already been described , including ten markers for L . ( L . ) donovani [19] , [20] , and six markers for New World species [18] , [21] . However , for L . ( Viannia ) species , these studies have mainly focused on interspecies discrimination and phylogenetic/taxonomic analysis and have employed only up to four markers . Given the challenges described above , we propose a panel of six gene loci , including three new markers described here for the first time , as an epidemiological tool for investigation of L . ( V . ) braziliensis outbreaks . In the present study , the recent outbreak in Santa Catarina is used to demonstrate the applicability of this technique in outbreak settings . The overarching objective of this work will be to generate interest in the community of leishmaniasis investigators to create an international sequence database based on these gene markers , as well as other markers from the original MLEE panel , for a more comprehensive and unified investigation into the distribution and epidemiological characteristics of Leishmania species .
Ethical approval for the use of patient data and their respective sample was received from the UFSC Ethics Committee . CLIOC is a Depository Authority of the Ministry of the Environment [Fiel Depositária pelo Ministério do Meio Ambiente , MMA] ( D . O . U . 05 . 04 . 2005 ) . Following Resolution 21 ( August 31 , 2006 – CGEN/MMA ) , authorization was not required for usage of samples previously deposited in CLIOC since the samples were used for research purposes only and data were analyzed anonymously . Leishmania ( Viannia ) braziliensis strains from eleven Brazilian states ( n = 53 ) were obtained from the Leishmania Collection of the Oswaldo Cruz Institute ( Coleção de Leishmania do Instituto Oswaldo Cruz- CLIOC ) in Rio de Janeiro , Brazil , and strains from Santa Catarina ( n = 33 ) were obtained from the cryobank of the Laboratório de Protozoologia of the Universidade Federal de Santa Catarina ( UFSC ) , Florianópolis , Santa Catarina , Brazil . Patient data from Santa Catarina used in this study were investigated as part of routine reportable disease surveillance and collection procedures have been previously described in [9] . Santa Catarina isolates were deposited in CLIOC and subjected to MLEE characterization , according to routine procedures employed by CLIOC . Leishmania promastigotes were cultured at 25°C in Schneider's medium supplemented with 20% heat-inactivated fetal bovine serum . DNA extraction was conducted using the Wizard DNA purification Kit ( Promega , Madison , USA ) , according to manufacturer's instructions . Amplification was performed for a panel of six housekeeping gene loci listed in Table 1 . Primers and PCR conditions have been previously described for 6-phosphogluconate dehydrogenase ( 6pgd ) , manose-6-phosphate isomerase ( mpi ) , isocitrate dehydrogenase ( icd ) [18] and for the heat shock protein 70 ( hsp70 ) [22] , [23] . Primers for mitochondrial malate dehydrogenase ( mdhmt ) and nuclear malate dehydrogenase ( mdhnc ) are described here for the first time . Both follow the reaction condition: for 50 µl , 0 , 2 mM of each primer , 100 mM Tris–HCl , pH 8 . 8; 500 mM KCl , 1% Triton X-100; 15 mM MgCl2 , 0 . 25 mM deoxyribonucleotide triphosphate ( dNTPs ) , 0 . 025 U FideliTaq/GoTaq polymerase and 50 ng DNA . Amplification conditions were 94°C for 2 min , followed by 34 cycles at 94°C for 30 s , 52°C for 30 s and 72°C for 1 min , with a final extension at 72°C for 5 min . PCR products were purified and subsequently sequenced with the same primers used in the PCR . Consensus sequences were obtained and edited in the software package Phred/Phrap/Consed Version: 0 . 020425 . c ( University of Washington , Seattle , WA , USA ) and only those with Phred values above 20 were used as contigs . Analyzed sequence fragment lengths for each marker are provided in Table 1 . Contigs of all strains were mounted and aligned in MEGA4 ( Molecular Evolutionary Genetics Analysis version 4 ) [24] . Ambiguous sites were divided into two of the possible alleles for all markers using the PHASE algorithm in DnaSP5 [25] . Clonal complexes ( CC ) were defined through BURST analysis in the software eBURSTv3 [26] . The BURST algorithm identified groups of mutually exclusive genotypes associated with a MLSA population and the founding genotype sequence within each group . Then , the algorithm provided the predicted descent from the founding genotype for all other genotypes [26] , [27] . For this analysis , criterion for CC formation was fixed at the most stringent level with at least five identical alleles for the six loci defining a CC . Sequences which were not able to be grouped into a clonal complex remained in the analysis as unique sequences . Haploid sequences rebuilt from the PHASE algorithm in DNAsp containing homozygous and heterozygous alleles were imported into STRUCTURE 2 . 3 . 4 ( University of Chicago , Chicago , IL , USA ) to investigate the population structure of the 86 samples of L . ( V . ) braziliensis based on the six MLSA loci . Using a Bayesian statistical approach , STRUCTURE applies a model-based clustering method to infer population structure and assign individuals to clusters based on multilocus genotype data [28] . Genetically distinct clusters ( K ) are identified based on the frequency of alleles , attributing the fraction of each genotype for each sample . In STRUCTURE , runs were performed using a burn-in period of 200 , 000 iterations followed by 600 , 000 running iterations . Runs were repeated three times to obtain data suitable for estimating the value of ΔK ( defined as the rate of variation of the log likelihood of data between successive values of K ) , which provides the most likely K value for the data to be used in STRUCTURE HARVESTER [29] . STRUCTURE HARVESTER generates graphs for the change in the log of k and calculation of ΔK of STRUCTURE results , which were compared for choosing the K that best fit the data . Next , CLUMPP version 1 . 1 . 2 [30] was employed to align the multiple replicate analyzes of the same data set . Hierarchical analysis of two to seven K clusters was performed to define the assignment of borderline strains . Based on clusters found in STRUCTURE , we used Microsatellite Analyser ( MSA ) [31] to estimate FST values and Genetic Data Analysis ( GDA ) version 1 . 1 [32] to calculate expected heterozygosity ( He ) , observed heterozygosity ( Ho ) , and inbreeding coefficient ( FIS ) . Recombination analysis was performed in Recombination Detection Program ( RDP ) [33] . To view genetic relationships ( phylogenetic network ) among strains and differentiation provided by the six markers , the median-joining network was mounted in the program SplitsTree 4 . 0 [34] . The median-joining network was constructed using concatenated character nucleotide sequences with ambiguous sites for all loci and strains . Nodes of the network , representing individual or groups of strains , were labeled by size , color and/or year/location to reflect epidemiological variables associated with the patient from whom the strain was isolated . Associations between genetic and epidemiological variables were analyzed in Stata SE 13 ( StataCorp LP , College Station , TX , USA ) . Chi-squared test , or Fisher's exact test when appropriate , was used to assess the relationships between categorical variables . Maps were created in ArcGIS 10 ( ESRI , Redlands , CA , USA ) .
BURST analysis identified three clonal complexes ( CC ) among the 86 strains of L . ( V . ) braziliensis , with over half ( 54 . 7% , 47/86 ) of the strains not belonging to any of the three CCs and remaining separate as unique sequence types ( Supporting Information S1 ) . A total of 76 distinct sequence types were observed among strains . The analysis was heavily weighted by the homogeneity and large number of strains from Santa Catarina included in the analysis , with the large majority ( 84 . 8% , 28/33 ) of Santa Catarina strains being grouped into one nearly exclusive clonal complex ( CC1 ) . Five out of six strains from Santa Catarina that did not group with CC1 were registered as imported cases in the epidemiological investigation . No association was found between CC and clinical form ( p = 0 . 660 ) . Figure 1 shows the geographical distribution of the CCs by state , revealing proportionally higher genetic variation in states from the Amazon biome ( 94 . 1% ( n = 16/17 ) unique sequence types ) ( Supporting Information S1 ) . Through calculation of ΔK in the STRUCTURE analysis , the L . ( V . ) braziliensis strains included in the present study from 12 Brazilian states were found to best fit into three clusters ( POP ) ( Supporting Information S1 ) . Overall , 41 . 9% ( 36/86 ) of strains belonged to POP1 , 40 . 7% ( 35/86 ) to POP2 , and 16 . 3% ( 14/86 ) to POP3 . As in the BURST analysis , the large majority ( 87 . 9% , 29/33 ) of Santa Catarina strains formed their own cluster ( POP2 ) , which also included four strains from Pernambuco , one from Mato Grosso and one from Bahia ( Figure 2 ) . The four Santa Catarina strains that did not cluster with POP2 were registered as imported cases in the epidemiological analysis . These four strains were the same strains from imported cases that did not cluster in the BURST analysis . Complete strain information can be found in Supporting Information S2 . As shown in Figure 3 , POP1 demonstrated the most extensive geographical distribution , including strains from all states analyzed in this study . A distinction can be made between the genetic variation and genetic structure of coastal states , which contain Atlantic forest , and northern states , which are located in the Amazon basin . States of the Amazon region were predominately comprised of POP1 strains , while strains of POP2 and POP3 were mainly found in coastal states . A significant association between the genetic cluster designated by STRUCTURE and leishmaniasis clinical form of the patient from which the strain was isolated was observed ( p = 0 . 030 ) ( Table 2 ) . Most strains from cases presenting the mucocutaneous clinical form ( 4/7 ) belonged to POP3 , including one case from Rio de Janeiro State , one from Pernambuco and two from Bahia . Based on the scale for the interpretation of FST suggested by Wright ( 1978 ) , the estimates showed significant genetic differentiation among the STRUCTURE clusters ( Table 3 ) . POP1 and POP3 showed moderate genetic differentiation ( FST = 0 . 1087 ) , while POP2 showed great genetic differentiation with POP1 and POP3 ( FST = 0 . 1540 and 0 . 2028 , respectively ) . POP1 had the highest average number of alleles per locus ( 23 . 3 ) , while both clusters POP2 and POP3 were similar in mean number of alleles , being approximately five alleles per locus . Positive values of FIS were found for all clusters . FIS values for POP1 and POP3 were particularly high ( Table 4 ) . All loci were polymorphic for POP1 and POP2 and five ( 83 . 3% ) of the six loci were polymorphic for POP3 . The marker 6pgd was not polymorphic for POP3 . In general , the new markers hsp70 , mdhnc and mdhmt showed the highest number of alleles of 35 , 40 and 44 , respectively , in comparison to 15–30 alleles for the other three markers . Results of the BURST and STRUCTURE analysis were found to be significantly associated ( p<0 . 001 ) ( Table 5 ) . The majority ( 37/48 ) of unique sequences in the BURST analysis were forced into their own population ( POP1 ) in the STRUCTURE analysis , representing mainly strains from the Amazon regions . Recombination events were detected by seven algorithms in RDP software ( p<0 . 05 ) . However , neither the beginning nor ending breakpoints could be identified , which may have resulted in recombinant misidentification . Nonetheless , one sample from Santa Catarina ( 185 ) and one sample from Bahia ( IOC/L 2871 ) were indicated as potentially parental or recombinant . Thirty-one samples from Santa Catarina had sequences with partial evidence of the same recombination event . The median-joining network was created from concatenated sequences of the six gene loci for the 86 strains of L . ( V . ) braziliensis from Brazil . Majority of Santa Catarina strains presented as an evident cluster . Other strains close to the Santa Catarina cluster were from Pernambuco ( n = 2 ) , Rio de Janeiro ( n = 1 ) , and Pará ( n = 1 ) . When the nodes of Santa Catarina strains were highlighted by case origin , all cases not clustered with the principal cluster were imported cases , with the exception of strain 605 ( Figure 4 ) . This 605 strain also was grouped within the main Santa Catarina CC and POP in both the BURST and STRUCTURE analyses . When the strains from cases of mucocutaneous and disseminated clinical form were highlighted , those from Bahia were clustered , while mucocutaneous cases from other Brazilian states appeared closer to the main cluster of Santa Catarina ( Figure 5 ) . When the median-joining network was reduced to only strains from Santa Catarina , the resulting network presented three principal branches . Marked by year and city of leishmaniasis case diagnosis , a main cluster can be observed in the center of the network , representing the epicenter of the outbreak which occurred in 2006 in the municipality of Blumenau ( Figure 6 ) . From this main epicenter , autochthonous cases branched separately , appearing to evolve over time and space to the neighboring municipality of the capital municipality of Florianópolis . The map in Figure 6 shows this main cluster of related Santa Catarina strains was distributed over a distance of 140 km in four years from Blumenau to Florianópolis .
MLSA revealed epidemiological patterns among L . ( V . ) braziliensis strains from twelve Brazilian states , even within the state of Santa Catarina where the strains presented extensive homogeneity . The addition of three markers , hsp70 , mdhnc and mdhmt to the previously described panel of markers increased the discriminatory power of the technique , permitting the identification of three genetic clusters within L . ( V . ) braziliensis strains . All three analyses ( BURST , STRUCTURE and median-joining network ) provided a complementary and integral part in the interpretation of the MLSA results . When used in tandem with MLMT , these two methods could provide a more robust approach to the molecular epidemiology of leishmaniasis and increased validity of the population structure model . A prospective study design that seeks to include a representative sample of the patient population and active collection of their Leishmania strains is needed to validate this method as a molecular epidemiology tool . However , the present study has provided sufficient evidence of the effectiveness of this method for pursuing further validation of MLSA for leishmaniasis outbreak investigation . | Molecular epidemiology of infectious diseases , which uses pathogen genetics to determine risk factors in the human population , is commonly employed to assist in outbreak investigation . While definitive genetic markers and techniques have been developed for several other bacterial , viral , and parasitic pathogens , the scientific community has yet to agree on an international standard for inter- and intra-species differentiation of Leishmania , the parasite that causes the disease leishmaniasis . As leishmaniasis represents one of the highest disease burdens among the neglected tropical diseases , development of molecular techniques , which allow for inter-laboratory comparability through international sequence databases , is imperative for moving forward with disease control . Based on the current standard technique employed for bacteria , the authors propose a panel of six genetic markers for multilocus sequence analysis ( MLSA ) for intraspecific differentiation of Leishmania braziliensis , the most widely distributed of the Leishmania species in Brazil . Using strains from a recent outbreak in the sub-tropical non-endemic southern Brazil in comparison with strains from eleven other Brazilian states , the authors provide a practical example of how this technique can be applied in a real world outbreak situation . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"molecular",
"epidemiology",
"infectious",
"disease",
"epidemiology",
"epidemiology",
"genetics",
"molecular",
"genetics",
"population",
"genetics",
"biology",
"microbiology",
"parasitology"
] | 2014 | Multilocus Sequence Analysis for Leishmania braziliensis Outbreak Investigation |
The idea that most morphological adaptations can be attributed to changes in the cis-regulation of gene expression levels has been gaining increasing acceptance , despite the fact that only a handful of such cases have so far been demonstrated . Moreover , because each of these cases involves only one gene , we lack any understanding of how natural selection may act on cis-regulation across entire pathways or networks . Here we apply a genome-wide test for selection on cis-regulation to two subspecies of the mouse Mus musculus . We find evidence for lineage-specific selection at over 100 genes involved in diverse processes such as growth , locomotion , and memory . These gene sets implicate candidate genes that are supported by both quantitative trait loci and a validated causality-testing framework , and they predict a number of phenotypic differences , which we confirm in all four cases tested . Our results suggest that gene expression adaptation is widespread and that these adaptations can be highly polygenic , involving cis-regulatory changes at numerous functionally related genes . These coordinated adaptations may contribute to divergence in a wide range of morphological , physiological , and behavioral phenotypes .
To what extent the evolution of gene expression cis-regulation drives the evolutionary innovations of life is an important unresolved question . While some contend that changes in cis-regulation are responsible for the majority of morphological adaptations [1] , others point out that only a few such cases have been demonstrated [2] , [3] ( we distinguish here between cis-regulatory changes that have been shown to affect phenotypes , of which there are a moderate number [4] , [5] , and those that have further been shown to be adaptive , of which there have been far fewer [2] , [3]; adaptive changes are those that are subject to positive selection as a result of increasing fitness ) . This long-standing paucity of examples of adaptive cis-regulatory divergence was due in large part to the fact that historically it has not been possible to formally demonstrate the presence of cis-regulatory adaptation from genome-wide data [3] . Sequence-based approaches have often been used to scan the genome for accelerated divergence in non-coding regions [6]–[9] , but what fraction of these represent positive selection on cis-regulation remains unknown; other possible explanations include changes in local mutation rate or biased gene conversion rate [10] , or selection on non-coding RNAs , recombination control elements , DNA replication origins , or any other non-coding feature of genomes ( e . g . [6] ) . Moreover , even when accelerated evolution does reflect cis-regulatory adaptation , the target genes often cannot be identified , since transcriptional enhancers can act on distant genes in many species . Alternatively , many studies have attempted to detect genes under positive selection from genome-wide gene expression data , but have been unable to demonstrate the presence of positive selection due to the lack of a null model of neutrality [3] , [11] . For example , the finding that gene expression divergence among three populations of Fundulus fish species correlates better with the species' environment than with their phylogeny [12] is consistent either with widespread adaptation to the environment , or with a neutral mutation affecting many gene expression levels being shared between two populations by chance; these cannot be distinguished without a null model of neutral change . Similarly , studies that rank genes by their ratio of gene expression divergence between species to diversity within species [13]–[14] can identify promising candidates for follow-up studies , but cannot distinguish between neutral and adaptive evolution without knowing how the expression of a “neutral gene” would evolve [3] . Several studies have succeeded in developing accurate neutral models of gene expression change by quantifying expression divergence when selection is artificially weakened in the lab [15]–[17] . In these studies positive selection on a gene's expression would be indicated by a greater divergence between species than expected from the neutral model; less divergence than expected would reflect negative selection . Although these studies have had the potential to discover positive selection , they have only uncovered negative selection—i . e . all genes have shown less divergence between species than expected under neutrality . However since these studies can only measure “average” selection pressures ( much like the dN/dS metric for coding regions ) , genes even with fairly frequent episodes of positive selection on expression would go undetected if they are most often subject to negative selection [3] . Therefore the lack of any positive selection on gene expression identified in these studies is not evidence against the existence of such positive selection . This landscape has changed with the recent publication of two studies of selection on genome-scale gene expression data in Saccharomyces yeast [3] , . In one of these [18] , we used the directionality of gene expression quantitative trait loci ( eQTL; reviewed in [20] ) to demonstrate that at least 242 gene expression levels ( and likely many more ) have been subject to lineage-specific selection ( i . e . different selective regimes between two lineages ) since the divergence of two strains of S . cerevisiae , and then employed population-genetic analyses to show that most of these represent positive selection , as opposed to relaxed negative selection . Although this work expanded the number of known cases of gene expression adaptation ( across all species ) by over 10-fold , it revealed little insight into the higher-level traits being selected . In another important recent study , Bullard et al . [19] examined the allele-specific expression ( ASE ) levels of gene sets ( e . g . pathways , co-expressed gene clusters , etc . ) in a hybrid between S . cerevisiae and another yeast , S . bayanus . ASE implies the presence of a cis-acting polymorphism affecting expression , and consistent directionality of ASE within a gene set implies lineage-specific selection ( see below for further explanation ) . This method has great promise for identifying the biological processes affected by gene expression adaptation , though it remains unknown if the gene sets implicated in this work have been subject to positive ( as opposed to relaxed negative ) selection [19] . Interestingly , parallel analysis of the genomic sequences of these same gene sets revealed no cases of either promoters or protein-coding regions under positive selection [19] . Here we apply a gene set-based test of selection on gene expression to M . musculus . Although mouse is a heavily studied model organism , both in the lab and in the wild , no cases of gene expression adaptation have been demonstrated in this species ( one example , the Agouti gene , has been found in Peromyscus deer mice [21] ) . Our results show that both “traditional” eQTL mapping in an F2 population as well ASE analysis in an F1 hybrid can be used to detect lineage-specific selection on gene sets , and that data from additional strains can be used to polarize the changes and infer the probable action of positive selection . Moreover , we expand the known extent of gene expression adaptation in M . musculus from zero genes to over 100 , and find that a great deal of such adaptation may occur in parallel on many genes of small effect , in contrast to all previously known cases of gene expression adaptation [1] , [2] aside from our work in yeast [18] . Finally , our results suggest that gene expression adaptations can affect behavioral and physiological phenotypes , in addition to their more well-established role in morphological evolution [1] .
The test of lineage-specific selection we use is based upon an idea first formalized by Orr [22] in an elegant test of selection on quantitative traits: under neutrality , QTLs for any given trait are expected to be unbiased with respect to their directionality . In other words , given two parents ( A and B ) of a genetic cross , A alleles at any QTL would be expected to be equally likely as B alleles to increase the trait value . If a significant bias is seen—e . g . , among 20 QTLs for a trait , the A allele increases the trait value at all of them—neutrality may be rejected in favor of lineage-specific selection ( in the absence of ascertainment bias [see Text S1] ) . At present , no gene expression levels have been mapped to a sufficient number of eQTLs to reject neutrality for any single gene . However , if the expression levels of an entire group of genes is treated as a single trait , and each eQTL used in the test is independent ( i . e . caused by a distinct polymorphism ) , then lineage-specific selection can be detected as a bias in the directionality of eQTLs for the gene set being tested [3] , [19] ( This approach will have the greatest power for gene sets containing genes that predominantly have the same direction of effect on a trait under selection; for gene sets with a significant fraction of genes that act in opposition , selection in one direction could result in upregulation of some , and downregulation of others . ) . The independence of eQTLs for different genes is critical for this test , since a single eQTL that affected many genes could lead to a strong bias in the directionality of effect even in the absence of lineage-specific selection ( Figure 1 , strain A versus B ) . To ensure that each eQTL is independent , we considered only local eQTLs—that is , eQTLs located at genetic markers that are close in the genome to the gene whose expression they control . These local eQTLs have been shown to be primarily cis-acting [23] ( so we refer to these as cis-eQTL for brevity ) , though we note that our test of selection is equally valid for local trans-acting eQTLs . Since a single cis-eQTL could conceivably control multiple nearby genes , and thus violate the requirement for independence , we also discard genes that are located close to others in the same gene set ( see Methods ) . At any eQTL , either the allele from parent A up-regulates expression ( and thus parent B's down-regulates ) , or the allele from parent A down-regulates expression ( and thus parent B's up-regulates ) . In our test we include an equal number of each type ( arbitrarily termed “+” and “–” ) , so that any gene set that is not under lineage-specific selection should have close to the same number of genes in each eQTL direction ( Figure 1 , strain A versus C ) . This null expectation requires no assumptions about gene sets or eQTLs or the complex biological networks involved , but follows simply from the fact that we constrain the total number of + and – eQTLs to be equal ( relaxing this constraint to allow different numbers of + and – eQTLs is straightforward , and requires only adjusting the null expectation; e . g . if we adjust our cutoffs so that 60% of all eQTLs are + , then any random or non-lineage-specific-selected gene set is expected to have ∼60% + eQTLs ) . A hypergeometric p-value , testing whether the observed data deviate from this expectation by having an excess of either + or – eQTLs ( Figure 1 , strain A versus D ) , constitutes the test . Although this method will have greater power for gene sets with many cis-eQTLs , any variation in the total number of cis-eQTLs per gene set ( whether due to real biological differences , or experimental design , e . g . gene sets not well-represented on the expression array ) will not lead to false-positive results , since these will affect + and – eQTLs equally . Further , the test is sensitive to both positive selection and relaxed negative selection acting on a gene set , as long as that selection is present in only one of the two lineages; thus it is a test of lineage-specific selection , although positive selection can be inferred with additional data ( see below ) . In this sense , it is similar to the McDonald-Kreitman test [24] , which also cannot distinguish between positive and relaxed negative selection [25] . However unlike the McDonald-Kreitman test , as well as nearly all other previous tests of selection ( on both gene expression levels and DNA/protein sequences ) , this is not dependent on any assumptions about either demographic histories or a subset of neutral sites ( see Text S1 ) . We applied our test of selection to eQTL data from a cross between two diverged inbred mouse strains , C57BL/6J ( B6 ) and CAST/EiJ ( CAST ) . B6 , like most commonly used lab strains , is a mosaic of several lineages [26] , but primarily Mus musculus domesticus . CAST represents Mus musculus castaneus , a subspecies thought to have diverged from the primary B6 progenitor strains ∼500 , 000 years ago [27] . This divergence , as well as recent selection during inbreeding in the lab , provides ample opportunity for adaptive changes to have accumulated in each lineage . To map cis-eQTLs , we produced 442 F2 animals , either with B6 as the F0 paternal strain ( referred to here as CxB F2 animals ) or maternal strain ( referred to as BxC F2 animals ) . Each mouse was genotyped at 1 , 438 informative genetic markers , and genome-wide gene expression was measured in adult brain , liver , and skeletal muscle ( see Methods ) . Cis-eQTLs were found by linear regression of gene expression levels on genotypes separately in each of four cohorts: CxB females , CxB males , BxC females , and BxC males . A total of 5 , 000 cis-eQTLs in each cohort—the strongest 2 , 500 in each direction ( corresponding to a false discovery rate [FDR] <10% in each cohort ) —were retained for analysis . Using the same number of + and – eQTLs allows us to apply our simple yet robust null expectation of neutrality to any gene set: regardless of what complex biological networks and population histories underlie the eQTLs , any gene set not subject to lineage-specific selection ( including random gene sets ) will show an approximately equal number of + and – eQTLs , following the binomial distribution . Throughout this work we report gene sets significant at either a high-confidence ( <2% FDR ) or medium-confidence ( <15% FDR ) cutoff , with FDRs estimated by testing randomly generated gene sets matched in size to the real ones ( see Methods ) . We began by testing gene sets from the Gene Ontology ( GO ) Consortium [28] , since these have been shown to be useful in a wide range of applications ( while any particular gene's GO classification and expression data may be imperfect , the sheer number of genes and expression measurements being used make this a potentially powerful test; any inaccuracies in the gene set assignments may lead to false negatives , but are unlikely to result in false positives ) . Applying the hypergeometric test to 531 GO gene sets ( each with at least 50 members ) separately in each tissue , we found one high-confidence set ( FDR = 1 . 5% , meaning that there is approximately a 1 . 5% probability that this enrichment is due to chance , given the number of gene sets tested , and the overlap in content between gene sets ) : genes in the “mitochondria” set were biased towards increased expression from B6 cis-eQTL alleles ( “B6-upregulation” ) in liver ( Table 1; see Table S1 for gene lists ) . These results were consistent across all four cohorts ( Figure 2a ) , not only at the gene-set level , but also for specific genes within those sets ( see Text S1 ) , underscoring their robustness . SNPs that could disrupt microarray probe hybridization are unlikely to explain the results , since these did not show any enrichment in the B6-upregulated mitochondria-related genes ( see Text S1 ) . The number of genes affected by selection can be estimated as the difference between the numbers of cis-eQTLs in each direction ( see Text S1 ) ; in mitochondria , this is estimated separately in each cohort as 32-35 genes in females and 47-48 genes in males ( Figure 2a , green numbers ) . We note this will be conservative if any of the CAST-upregulated cis-eQTLs were fixed by positive selection as well . No additional gene sets were observed with medium confidence . To increase our statistical power , we combined results across tissues , since many cis-eQTLs in our data were not tissue-specific . Seven additional gene sets were found: one at high-confidence and six at medium-confidence ( Table 1; see Table S2 for results from all 531 gene sets ) . Two of the seven sets were related to mitochondria at different levels of the GO hierarchy ( “mitochondrial inner membrane” and “intracellular organelle” ) , while the other five represented a diverse collection of functions . As an example , locomotory genes—which are biased towards CAST-upregulation in all three tissues—are shown in Figure 2b . Similar to the mitochondria gene set , the specific genes implicated in each cohort overlapped extensively ( see Text S1 ) . In sum , these results suggest that lineage-specific selection involving these subspecies can be inferred for several functional categories . We also applied our method to other types of gene sets . Testing 41 modules of genes co-expressed in each F2 population ( see Methods ) , we did not find any significant enrichments for biased directionality of cis-eQTLs . However testing 75 pathways from the KEGG database [29] , we found one at medium confidence ( FDR = 4 . 5% ) : the JAK/STAT pathway was biased towards cis-upregulation in CAST brain ( Table 1 ) . To complement the microarray-based approach described above , we turned to sequencing RNA isolated from F1 mice to directly identify allele-specific expression ( ASE ) . While this approach does not offer the richness in terms of understanding genetically regulated networks and their interactions that can be achieved in a large F2 cross , it does address two drawbacks of the microarray approach described above: 1 ) our microarrays cannot provide direct evidence of cis-regulation ( since local eQTLs can occasionally be trans-acting [23] ) , so we cannot be confident that our results truly reflect selection solely on cis-acting elements; and 2 ) there is considerable time and expense associated with rearing , genotyping , and expression profiling of hundreds of F2 mice . We and others have shown that high-throughput mRNA sequencing ( RNA-seq ) in F1 hybrid mice is an effective approach to studying ASE [30]–[32] . mRNA levels can be accurately estimated by simply counting the density of reads from each transcript . Since heterozygous SNPs are present at a 1∶1 ratio in the genome , any significant deviation from this ratio in the number of sequence reads that can be mapped to each individual allele ( as a result of containing a heterozygous SNP ) indicates ASE . When the allele-specificity associates in reciprocal crosses with SNP genotype—as opposed to parent-of-origin , as seen for imprinted loci [30]–[31]—this implies the presence of a cis-acting eQTL . These cis-eQTL target genes can then be used as input for our selection test , in exactly the same fashion as those found using microarrays in an F2 population . We searched for ASE in a set of ∼78 million sequence reads from F1 hybrid BxC and CxB embryos we generated previously [30] . Because this is not only a different technology , but also a different developmental stage ( embryonic day 9 . 5 ) and tissue ( whole embryos ) , we were encouraged to see several of our strongest hits replicate . For example , mitochondrial genes show a bias towards higher expression of B6 alleles , whereas locomotory-related genes show the opposite ( Figure 3a ) . Gene sets that were biased in adults but not in F1 embryos might be tissue and/or stage-specific , or may be missing due to lower power of our RNA-seq data for weakly expressed genes ( this is not an inherent limitation of RNA-seq , since power is limited only by the number of reads ) . In addition , genes lacking any B6/CAST sequence polymorphisms are not assayable by allele-specific RNA-seq . In addition to replicating some hits from adult mice , the F1 embryo data revealed new significant gene sets as well . Two gene sets reached high-confidence: “calmodulin binding” and “memory” ( Table 1 and Figure 3b ) , both showing a bias towards B6-upregulation . Although unannotated SNPs overlapping RNA-seq reads can cause a marginal alignment bias resulting in an apparent up-regulation of the B6 reference genome alleles , our analysis indicates this is unlikely to underlie the significance of these gene sets ( see Text S1 ) . Consistent with previous work in yeast [19] , we conclude that RNA-seq is a cost-effective alternative for measuring selection on cis-regulation , particularly between lineages with a high density of exonic sequence differences . An important question is whether the lineage-specific selection we detected has had any detectable impact on organismal phenotypes . Examination of the gene sets in Table 1 reveals that specific predictions can be made for the gene sets belonging to the GO “biological process” and “cellular component” ontologies ( Table 2 ) . For example , cis-eQTLs leading to higher expression of growth or locomotory genes may be ( at least naively ) expected to increase growth or locomotion , since these gene annotations were typically identified by observing a reduction of growth or locomotion in a gene knockout/knockdown model; genes leading to increased growth or locomotion when inactivated are far less common ( for example , among genes annotated as growth regulators [28] , 40 have a mutant phenotype of decreased body size , whereas only two are associated with increased body size [33] ) . These effects could either be strong , like all previous examples of adaptive cis-regulatory adaptation in metazoans [1] , [2]; or subtle , considering that many loci are being selected in parallel and thus may only exert major phenotypic effects in aggregate . We were unable to make any phenotypic predictions for the GO molecular function terms ( “calmodulin binding” , “G-protein coupled receptor activity” , “receptor activity” , or “enzyme inhibitor activity” ) , or the JAK-STAT pathway . If the loci we identified have major phenotypic effects , they should be detectable by QTL mapping in our F2 mice . One phenotype we predicted to be affected was measured for every F2 individual in our cross: naso-anal length , which approximately reflects the sum of growth over the lifetime of the mice . In females , we found two significant QTLs for length , on chromosomes 2 and 15 , while in males the strongest QTL was on chromosome 5 ( Figure 4 , red lines ) . In all three cases , the B6 alleles were associated with greater length , as expected since B6 alleles tend to increase expression of growth-related genes ( whose knockout/knockdown phenotype is typically a reduction in growth ) . Strikingly , the strongest QTL from each gender overlapped almost perfectly with two of the strongest ( genotype versus expression level r2>0 . 5 ) cis-eQTLs in the growth-related gene set ( Figure 4 , blue lines ) , and the weaker female length QTL coincided with a weaker ( r2>0 . 2 ) but still highly significant growth-related gene cis-eQTL ( Figure 4a , green line ) . This overlap is unlikely to occur by chance , considering that only ∼0 . 5% of cis-eQTLs are as close to the length QTLs as each of these are ( probability of overlap by chance , p<0 . 001; see Methods ) . The three genes are Dcaf13 ( also known as WDSOF1 ) , an rRNA processing factor; Ept1 , a CDP-alcohol phosphatidyltransferase ( orthologous to human SELI ) ; and Sp3 , a transcription factor . All three are well-conserved , and have been implicated in positive regulation of growth either by mouse knockout [34] , or RNAi experiments involving their orthologs in Caenorhabditis elegans [35] . This highly significant overlap suggests that these genes may be responsible for the length QTLs . To further test the hypothesis that the cis-eQTLs for these three genes affect mouse length , we applied a statistical approach for inferring causality of eQTLs for other traits [36]–[37] that has been extensively tested and validated using transgenic mice [38] . For all three genes , causality for length was strongly ( p<0 . 001 ) supported in at least one tissue . This provides further support for a role of these eQTLs in the length phenotype . An alternative method to assess the phenotypic importance of these gene sets is to compare the predictions to phenotypes of B6 and CAST mice . Although QTL mapping cannot be performed with only two strains ( typical mapping populations consist of hundreds of F2 individuals or recombinant inbred lines ) —and thus causal loci cannot be implicated—concordance of predictions with observed phenotypes can at least serve as evidence that the selection on cis-regulation of these gene sets is phenotypically relevant . To this end , we searched the literature for studies where phenotypes we predicted to be affected by selection ( Table 2 ) were measured in B6 and CAST . For three of our four predictions , we found multiple studies testing the relevant phenotypes . From the growth regulator gene set , we predicted larger size of B6 mice ( measured by length , as above , or by total body mass ) , and indeed they are known to have nearly twice the mass of CAST mice , from an early age through adulthood [39] , [40] . From the adult locomotory-related gene set showing CAST-upregulation ( found in both the microarray and RNA-seq data , Figure 1 strain B , and Figure 2a ) , we predicted higher locomotor activity in CAST , which has indeed been observed [40] , [41] . In fact , one study [41] found that daytime activity of CAST was over six times higher than that of B6 . The B6-upregulation of the memory-related gene set ( Figure 2b ) predicted increased memory in B6 ( since knockout of most memory-related genes results in reduced , not increased , memory ) . In two studies employing the Morris Water Maze ( MWM ) to measure learning and memory , B6 significantly outperformed CAST [40] , [42] . In fact , CAST showed no capacity at all for memory in this context ( see Text S1 ) . In sum , all three of our predictions that have been addressed in previous publications were confirmed by multiple independent studies . We did not find any studies contradicting these predictions . Our fourth prediction—that mitochondria would be more abundant in B6 , as a result of the B6-upregulation of many mitochondrial genes ( most notably genes related to the inner membrane , but also mitochondrial small ribosomal subunits [combined-tissue p = 4 . 5×10−8] , among others ) observed in both the microarray and RNA-seq data—has not , to our knowledge , been tested by previous studies . Therefore we isolated nuclear and mitochondrial genomic DNA from livers ( the tissue with the strongest B6-upregulation of mitochondrial genes ) of B6 and CAST adult mice , and measured the ratio of their mitochondrial to nuclear genome copy number by qPCR ( see Methods ) . Consistent with our prediction , we found a small but highly significant ( p<0 . 001 ) difference between B6 and CAST , with B6 showing a 12 . 9% increase in abundance . Therefore , all four of our predictions have been confirmed—three retrospectively and one prospectively—underscoring the ability of our selection test to predict phenotypic differences , and suggesting that these differences may have been shaped by lineage-specific selection on cis-regulation ( though we note that other traits could also have been affected by , or been the primary targets of , the lineage-specific selection in these gene sets ) . To better understand the selection that has acted on these phenotypes , we sought to determine on which lineage the majority of changes in each trait had occurred . This can be achieved by including an outgroup species in the analysis: for example , if a trait value in B6 is much further from the outgroup than is the CAST trait value , then the most parsimonious explanation is that the majority of divergence occurred on the B6 lineage . As with all parsimony-based methods , this indicates the most likely evolutionary scenario ( i . e . that requiring the fewest changes ) , but cannot formally rule out any less parsimonious explanation . To perform this analysis we searched for measurements of the four traits in Table 2 from additional mouse strains . Mus spretus ( SPRET ) is an ideal outgroup , being the species most closely related to Mus musculus . We found published measurements from SPRET for two of the traits , growth and memory . For growth , the adult mass of SPRET was found to be statistically indistinguishable from CAST [39]—and about half of that of B6—indicating that the change in growth likely took place along the B6 lineage . Similarly for memory , SPRET showed no evidence of recall in the MWM [42] , similar to CAST but in stark contrast to B6—again implicating the B6 lineage as the probable source of divergence . In fact , B6 showed significantly greater recall than all of the 12 other strains tested [42] . Although locomotory behavior has not been measured in an outgroup ( to our knowledge ) , it was measured in nine strains in addition to B6 and CAST [41] , including seven wild-derived strains that are more closely related to CAST than is B6 or other lab strains [43] . Since CAST had over twice the daytime locomotory activity of any other strain tested [41]—including the closely related wild strains—the majority of divergence can be inferred to have likely taken place on the CAST lineage , after its divergence from the other wild strains ( in this case , B6 is the outgroup ) . The much lower daytime activity level of B6 was similar to most of the wild strains , as well as another lab strain [43] . In sum , the phenotypic changes can be polarized for three of the traits . These results rest on the logic of parsimony: that a phenotypic change in one lineage is more likely than independent changes in the same trait—of the same direction and magnitude—in two lineages . Under the assumption that the phenotypic divergence was driven by ( and thus occurred on the same branch as ) the expression divergence , all three cases can be inferred to have likely been caused by cis-upregulation of the relevant gene sets . As mentioned above , our test of lineage-specific selection cannot by itself distinguish between positive selection and relaxed negative selection ( analogous to the McDonald-Kreitman test [24] , [25] ) . However recent evidence from saturation mutagenesis studies showing that the vast majority of random cis-regulatory mutations cause downregulation ( see Text S1 ) suggests that relaxed negative selection would likewise be biased towards downregulation . If this is indeed the case for the gene sets we have implicated , then relaxed negative selection is unlikely to explain the upregulation of these three traits/gene sets , leading to the conclusion that their divergence was most likely due to the action of positive selection for upregulation . However given the qualitative nature of this argument , we cannot yet quantify the precise probability that positive selection has been acting upon the cis-regulation of these gene sets .
Using a systematic genome-scale approach to inferring lineage-specific selection acting on cis-regulation , we found that over 100 genes belonging to several gene sets have undergone lineage-specific selection in mouse , which may have impacted diverse morphological and behavioral phenotypes . This work reports the first cases of adaptive cis-regulatory evolution in M . musculus , and expands the classes of traits ( in any species ) known to be affected by gene expression adaptation , which previously did not include any behavioral phenotypes . Methodologically , we augment previous work [19] by showing that adding information from an outgroup can suggest the likely action of positive selection ( as opposed to relaxed negative selection ) when that selection was for cis-acting upregulation . Two interesting questions for future work are how much of this selection occurred since the introduction of these strains to the lab , and for selection that occurred on the wild B6 ancestors , how much occurred in Mus musculus domesticus ( the primary ancestor of B6 [26] ) as opposed to Mus musculus musculus . Interestingly , wild M . m . domesticus tend to be larger than wild M . m . castaneus when reared in a common laboratory environment ( C . Pfeifle , personal communication ) , suggesting that this adaptation was likely to have occurred in the wild . Another question raised by these findings is what are the relevant “units of selection” [44] for these polygenic adaptations; though regardless of the answer , our conclusions regarding the extent of selection on cis-regulation will not be affected . Because the RNA-seq version of this approach can be applied rapidly and inexpensively to hybrids between any two diverged lineages ( including outbred lineages ) , we expect it will find use in a wide range of taxa . In fact , it can be applied to any ASE data from a hybrid between diverged lineages . Published ASE data sets from a variety of species ( e . g . [45] , [46] ) can now be similarly re-analyzed for cis-regulatory selection . This approach can also be applied to any of the numerous published eQTL data sets involving crosses between diverged parental lines . Our approach is quite different from all previous studies of metazoan cis-regulatory adaptation [1]–[4] , which have identified single genes with extremely strong effects on phenotypes such as pigmentation ( e . g . [21] , [47] , [48] ) or skeletal structure ( e . g . [49] ) . Our results reveal several important insights that could not have been found at this single-gene level . For example , the only previously known case of pathway-level gene expression adaptation was from our work on the ergosterol biosynthesis pathway in S . cerevisiae , where six genes clustered in the pathway have undergone selection for down-regulation [18] . Our present results extend this considerably , demonstrating that polygenic cis-regulatory adaptation can operate in parallel on dozens of genes within a single functional group or pathway , and that this has occurred in multiple gene sets during recent mouse evolution . Although each gene under such coordinate selection may be expected to have a less extreme phenotypic effect than those previously reported [1] , [2] , [21] , [47]–[49] , the sum of their effects could be quite strong . One important question that can now start to be addressed is how often cis-regulatory adaptation proceeds via dramatic changes in single genes , as opposed to more subtle changes distributed across an entire gene set [3] . Much of the answer may ultimately depend on factors such as the strength/duration of selection ( with intense/short-term selection pressure likely favoring extreme single-locus changes ) and the genetic architecture of the trait in question . A second open question is how often cis-regulatory adaptation occurs by upregulation versus downregulation of genes; our results suggest that the majority of the adaptation we discovered was due to upregulation , in contrast to most previous ( single-locus ) studies , which have predominantly identified cases of trait loss via downregulation [2] . Interestingly , we previously observed a preponderance of upregulation in a genome-wide study of gene expression adaptation in S . cerevisiae [18] , suggesting that this pattern may be widespread . Again , which of these is more common in a particular species may depend on the nature of the selective pressure and the underlying genetic architecture . Third , it has been proposed that gene expression adaptation may be responsible for most morphological adaptations in part because it offers a solution to the issue of pleiotropy . For a gene expressed in many tissues or stages of development , an amino acid change ( in a constitutive exon ) will affect the protein produced in all of these different contexts . Even if this change is adaptive in one or two of them , it has been argued that it would be highly unlikely to be advantageous in all of them [1] . In contrast , the modular nature of cis-regulation allows for a change in expression in just one tissue or stage , without affecting any other; thus pleiotropic constraints should not be as severe , and adaptation should be able to proceed [1] . Predictions from this are that genes expressed more broadly will be more likely to adapt via cis-regulation , and that these adaptations will only affect a small part of the genes' expression patterns . Two recent studies attempted to test this idea . In one of these [50] , genes near noncoding elements with accelerated evolution in the human lineage were proposed to have undergone human-specific selection on cis-regulation ( though the authors acknowledged that such acceleration need not indicate positive selection ) ; however no enrichment was found for these genes to be expressed in more tissues than average . In the other [51] , genes were classified as either “morphogenes” or “physiogenes” based on their mouse knockout phenotypes; morphogenes ( which tend to be expressed in fewer tissues ) had higher dN/dS ( an indicator of selection on protein-coding regions ) , while physiogenes had a higher magnitude of expression change between human and mouse , consistent with the prediction of greater adaptive expression change in broadly expressed genes . However this study did not distinguish between adaptive versus non-adaptive change , or cis versus trans regulation , or tissue-specific versus non-specific expression changes , so the relevance to theories of tissue-specific adaptive cis-regulatory evolution is not clear . Our results suggest that although most of the genes in our most significant gene sets are broadly expressed ( not shown ) , their expression in all three tissues was affected by the recent selection on cis-regulation we detected ( Table S2; all gene sets from Table 1 were significant in all three tissues , except for the JAK/STAT pathway ) ; thus these adaptations were not tissue-specific , so do not support pleiotropy-based arguments for the expected prevalence of tissue-specific gene expression adaptation ( we note that while the adaptations did not result in tissue-specific expression changes , the selection may have acted to change expression in just one tissue , with the rest changing as a side-effect ) . Of course , since we have only examined three tissues in two mouse strains , much more work is required to determine how general this conclusion is . Finally , because of its genome-scale perspective , our approach may eventually help to address many other fundamental questions that cannot be addressed by single-locus studies [3] , such as what fraction of gene expression divergence is adaptive , and what fraction of evolutionary adaptation occurs at the level of cis-regulation .
Ethics statement: All mouse work was conducted according to Institutional Animal Care and Use Committee regulations . C57BL/6J ( B6 ) mice were intercrossed with M . m . castaneus ( CAST/EiJ ) mice to generate 442 F2 progeny ( 276 females , 166 males ) . All mice were maintained on a 12 h light–12 h dark cycle and fed ad libitum . Mice were fed Purina Chow until 10 wk of age , and then fed western diet ( Teklad 88137 , Harlan Teklad ) for the subsequent 8 wk . Mice were fasted overnight before they were killed . Their tissues were collected , flash frozen in liquid nitrogen , and stored in −80°C prior to RNA isolation . RNA preparation and array hybridizations were performed at Rosetta Inpharmatics . The custom ink-jet microarrays used were manufactured by Agilent Technologies . The array used consisted of 2 , 186 control probes and 23 , 574 non-control oligonucleotides extracted from mouse Unigene clusters and combined with RefSeq sequences and RIKEN full-length cDNA clones . Mouse tissues were homogenized , and total RNA extracted using Trizol reagent ( Invitrogen ) according to manufacturer's protocol . Three micrograms of total RNA was reverse transcribed and labeled with either Cy3 or Cy5 fluorochrome . Labeled complementary RNA ( cRNA ) from each F2 animal was hybridized against a cross-specific pool of labeled cRNAs constructed from equal aliquots of RNA from 150 F2 animals and parental mouse strains for each of the three tissues . The hybridizations were performed to single arrays ( individuals F2 samples labeled with Cy5 and reference pools labeled with Cy3 fluorochromes ) for 24 h in a hybridization chamber , washed , and scanned using a confocal laser scanner . Arrays were quantified on the basis of spot intensity relative to background , adjusted for experimental variation between arrays using average intensity over multiple channels , and fitted to a previously described error model to determine significance ( type I error ) [52] . All microarray data are available at NCBI GEO ( GSE16227 ) . Genomic DNA was isolated from tail sections using standard methods and genotyping was performed by Affymetrix ( Santa Clara , CA ) using the Affymetrix GeneChip Mouse Mapping 5K Panel . The RNA-seq data were described previously [30] . All data are available at the NCBI SRA ( accession SRA008621 . 10 ) . eQTL scans were performed by linear regression of expression log ratios against genotypes ( coded as 0 , 1 , and 2 ) , separately in each tissue for each of the four cohorts ( CxB females , CxB males , BxC females , and BxC males ) . eQTL were designated as “local” ( and likely cis-acting ) if the regression between the expression level of a gene and a genetic marker within 1 megabase of the transcription start site was significant ( where significance was defined as the cutoff resulting in 2 , 500 eQTLs in each direction; see below ) . Testing for dominance ( comparing the average heterozygote value to the average of the two average homozygote values ) revealed evidence for non-additivity at only a small fraction of local eQTLs ( as expected for cis-eQTLs , which typically act additively ) , so dominance effects were not included in our eQTL mapping . We implemented the following strategy to isolate local eQTL effects in the presence of unlinked marker correlations . First the strongest local eQTL was identified , and expression of the target gene was then corrected for its effects by taking the residuals of expression when regressed against the eQTL genotype . The corrected expression level was then subjected to a whole-genome eQTL scan to identify the strongest trans-eQTL . Once this trans-eQTL was identified , its effects were regressed out of the original expression levels for the gene . These trans-corrected expression levels were then regressed against all local genetic markers once again , to identify the strength and direction of effect for the cis-eQTL . This process allows us to achieve a more accurate estimate of local eQTL effect sizes , even in the presence of unlinked trans-eQTLs or correlations between unlinked genetic markers ( we note that removing trans effects is not necessary for our test , though we have found it to improve our ability to estimate cis effects ) . More generally , our focus on local eQTLs allows us to isolate the effect of the local polymorphism ( s ) on gene expression , regardless of other effects ( e . g . environmental effects , trans-eQTL not captured in our regression approach , epistatic interactions , feedback , etc . ) ; of course such effects are widespread , but they will only weaken the correlation between a genetic marker's genotype and a nearby gene expression level , potentially causing us to miss some local eQTLs , but not resulting in false-positive results . A total of 5 , 000 genes with the strongest cis-eQTLs ( 2 , 500 in each direction ) in each tissue/cohort combination were analyzed . The decision to use an equal number of eQTLs in each direction does not reflect any biological aspects or assumptions , but instead is merely an arbitrary choice . Whether the total “true” numbers of cis-eQTLs in each direction are actually equal is not addressed here ( nor is it directly relevant for interpreting our test's results ) . Altering the proportion of eQTLs in each direction by up to 10% ( a 60/40 ratio ) in either direction did not have any impact on our results ( i . e . the gene sets in Table 1 were not affected , although FDRs were changed slightly ) . FDRs for each tissue/cohort combination were estimated by randomization . We first shuffled genotype labels so that one individual's entire set of genotypes was paired with another individual's expression levels . Then the entire eQTL detection procedure was carried out , and the number of cis-eQTLs above the cutoffs associated with the top 5 , 000 eQTLs in the real data were counted . Randomizations were repeated at least 1 , 000 times . The estimated FDR equals the average number of significant eQTLs in the randomized data divided by 5 , 000 ( the number in the real data ) . This procedure yielded a maximum FDR of 9 . 7% in the smaller cohorts ( BxC ) , and an FDR of <2% in the larger ( CxB ) ones . An equal number of eQTLs were used in each cohort so that results between cohorts would be directly comparable . We note that 5 , 000 eQTLs represents an average of ∼3 . 5 eQTLs per genetic marker , which is not surprising given that linkage disequilibrium extends for many megabases in a mouse F2 cross , so a single marker captures many polymorphisms . Gene ontology ( GO ) and Kyoto Encyclopedia of Genes and Genomes ( KEGG ) classifications were tabulated for each gene on the microarray . Only the 531 GO gene sets ( from all levels of the GO hierarchy and all three GO branches: Biological Process , Molecular Function , and Cellular Component ) and 75 KEGG gene sets containing at least 50 genes on our microarray were tested , since small gene sets have little statistical power in our test . If multiple genes from a gene set had cis-eQTLs and were located within 2 mb of each other in the genome , all but one in the cluster were discarded from the analysis , to ensure that the eQTLs being tested are all independent ( the 2 mb cutoff was chosen since the most distant known cis-regulation is an enhancer ∼1 mb from its target gene; so allowing 1 mb from a cis-regulatory mutation in each direction yields 2 mb ) . All of the cases of clustered eQTLs within a gene set showed the same direction of effect ( either up-regulated by the B6 allele , or by the CAST allele , but not a mixture of both ) , so the choice of which gene ( s ) to exclude had no effect on the test's results . Relaxing our distance cutoff results in a small increase in the sample size and gene set enrichment significance . Likewise , increasing the distance cutoff excludes a small fraction of genes , marginally decreasing the enrichment significance . The effect directions for the cis-eQTLs of a gene set were then tested for departure from the expected 1∶1 ratio of +/– alleles by comparing to the hypergeometric expectation . The results are similar to testing using the binomial expectation , but the hypergeometric takes into account the fact that if many + alleles have already been observed in a gene set , further genes in that set are actually slightly less likely to have + alleles by chance ( since the total number of + and – alleles included in our list is equal ) . Coexpression modules were constructed for each tissue as previously described [53] . A total of 41 modules containing at least 50 genes were tested ( 10 in brain , 14 in liver , and 17 in muscle ) . Hypergeometric p-values for each gene set in each tissue/cohort were then combined across cohorts by Fisher's method , to yield the single-tissue p-values for each gene set . The FDR was estimated in two ways . In the first approach , genotype labels were permuted as described above , and the entire eQTL detection and directionality test procedure was carried out . This yielded zero false positives even over many thousands of randomizations . However this randomization strategy does not account for the fact that a gene with a B6-upregulating cis-eQTL in one cohort is likely to have B6-upregulating alleles in other cohorts as well . In order to capture this effect in our permutations , we carried out a second randomization procedure . We used the cis-eQTL results from the real data , but randomly shuffled the gene set assignments for each gene . In this test , the consistency of eQTL directions across tissues and cohorts is perfectly preserved , and only the effect of the gene set assignments is randomized . With this procedure , false positives were found at all cutoffs tested; FDRs were estimated at several cutoffs , and are shown in Table 1 . We note that although the data from different tissues are not entirely independent , since they come from the same mice , this does not present a problem for estimating FDRs because we combined the p-values in the same way for both real and permuted data . In addition , the non-independence of gene sets is not a problem , since this overlap is perfectly captured by our randomization procedure . For the multi-tissue analysis , the three single-tissue p-values for each gene set were combined by Fisher's method , both for real and randomized data . This was expected to increase power because it decreases false-positive eQTLs , though it is also possible that the non-tissue-specific eQTLs this procedure enriches are more likely to be the result of recent selection . FDRs were estimated as described above for the single-tissue analysis . We also tested combining only results from mice of each gender , but did not find any sex-specific gene set enrichments . The RNA-seq data were analyzed as follows . Sequence reads overlapping heterozygous SNPs were assigned to alleles as described [30] . All reads from each allele of each RefSeq gene were then summed to generate the total number of reads from each allele . Distinct transcripts from the same gene cannot be discerned with this approach ( as with the vast majority of microarrays ) , so each gene was treated as if it produced a single transcript ( we note that since GO annotations are typically for genes , and not individual transcripts , having transcript-specific data would not substantially affect our results ) . SNPs with no reads from one allele were discarded , since these are likely to reflect SNP annotation errors . Binomial p-values were calculated for each gene , using the expected 1∶1 ratio of reads from each allele . The most extreme 25% of genes with allele-specific information ( 2 , 037 genes ) in each direction were retained for GO analysis . The GO analysis was carried out with the hypergeometric test as described above , except that no p-values were combined because only a single tissue/cohort was used . Randomizations were performed by replacing the cis-eQTL target genes with randomly chosen genes , and repeating the hypergeometric test . The probability of QTLs for naso-anal length overlapping with eQTLs for the growth regulator gene set was calculated as follows . The peaks for all three eQTLs shown in Figure 4 were within the 0 . 5 LOD support interval of the top three length QTLs ( one in males , two in females ) . Across all 3 , 834 eQTLs at this strength ( r2>0 . 2 ) , only 0 . 6% were within this interval of the male length QTL and 0 . 3% for each female length QTL . Since these are independent , and 27 eQTLs from the growth gene set reached this cutoff , the chance of all three overlaps is 27×0 . 006×26×0 . 003×25×0 . 003 = 0 . 00095 . Interestingly , the eQTL overlapping the strongest length QTL in each gender were both in the top 12 strongest growth eQTL ( r 2>0 . 5 ) , so even just the overlap of those two is significant at p = 0 . 002 . Testing the overlap with the three length QTLs in random groups of 27 eQTLs supported these calculations . In males there is one length QTL where the CAST allele is associated with greater length , but this was not included in our overlap analysis because we only posit that the alleles increasing B6 growth have been under positive selection and are present in the list of growth genes with B6-upregulating cis-eQTL . eQTL scans shown in Figure 4 were performed using CxB brain; brain was chosen because it is the tissue with the strongest growth gene eQTL direction bias , and CxB was chosen because it is the larger of the two cohorts . Expression levels were from CxB female brains in Figure 4a , and CxB male brains in Figure 4b , to match genders with the length QTL shown . We performed quantitative PCR with SYBR green , amplifying both nuclear and mitochondrial DNA from B6 and CAST liver tissue . The ratio of mitochondrial/nuclear DNA gives an estimate of the mitochondrial abundance in each strain , and the ratio of these ratios indicates their relative levels . The following primer sequences were used: nuclear , CCTTGGACATTAGCACATGG and AACTGTCTCCCCTGACCAAC; mitochondrial , ACAATGTTAGGGCCTTTTCG and GTTCCCAGAGGTTCAAATCC . No off-target effects were observed for either primer pair . Each reaction was repeated 48 times to ensure consistency . The 99% confidence interval for the B6:CAST ratio of mitochondrial/genomic DNA ( a ratio of ratios ) was 1 . 06 – 1 . 20 , and the 99 . 9% confidence interval was 1 . 04 – 1 . 23 . | Evolution can involve changes that are advantageous—known as adaptations—as well as changes that are neutral or slightly deleterious , which are established through a process of random drift . Discerning what specific differences between any two lineages are adaptive is a major goal of evolutionary biology . For gene expression differences , this has traditionally proven to be a challenging question , and previous studies of gene expression adaptation in metazoans have been restricted to the single-gene level . Here we present a genome-wide analysis of gene expression evolution in two subspecies of the mouse Mus musculus . We find several groups of genes that have likely been subject to selection for up-regulation in a specific lineage . These groups include genes related to mitochondria , growth , locomotion , and memory . Analysis of the phenotypes of these mice indicates that these adaptations may have had a significant impact on a wide range of phenotypes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genomics",
"genetics",
"biology",
"computational",
"biology",
"evolutionary",
"biology",
"genetics",
"and",
"genomics"
] | 2011 | Systematic Detection of Polygenic cis-Regulatory
Evolution |
Symptoms and signs of leptospirosis are non-specific . Several diagnostic tests for leptospirosis are available and in some instances are being used prior to treatment of leptospirosis-suspected patients . There is therefore a need to evaluate the cost-effectiveness of the different treatment strategies in order to avoid misuse of scarce resources and ensure best possible health outcomes for patients . The study population was adult patients , presented with uncomplicated acute febrile illness , without an obvious focus of infection or malaria or typical dengue infection . We compared the cost and effectiveness of 5 management strategies: 1 ) no patients tested or given antibiotic treatment; 2 ) all patients given empirical doxycycline treatment; patients given doxycycline when a patient is tested positive for leptospirosis using: 3 ) lateral flow; 4 ) MCAT; 5 ) latex test . The framework used is a cost-benefit analysis , accounting for all direct medical costs in diagnosing and treating patients suspected of leptospirosis . Outcomes are measured in length of fever after treatment which is then converted to productivity losses to capture the full economic costs . Empirical doxycycline treatment was the most efficient strategy , being both the least costly alternative and the one that resulted in the shortest duration of fever . The limited sensitivity of all three diagnostic tests implied that their use to guide treatment was not cost-effective . The most influential parameter driving these results was the cost of treating patients with complications for patients who did not receive adequate treatment as a result of incorrect diagnosis or a strategy of no-antibiotic-treatment . Clinicians should continue treating suspected cases of leptospirosis on an empirical basis . This conclusion holds true as long as policy makers are not prioritizing the reduction of use of antibiotics , in which case the use of the latex test would be the most efficient strategy .
Leptospirosis is a zoonosis of worldwide distribution , caused by infection with pathogenic spirochetes of the genus Leptospira . Human leptospirosis is an important health problem in Asia [1]–[3] and Latin America [4] , [5] . The source of infection in humans is either direct or indirect contact with the urine of an infected animal , whether livestock , domestic pets , rodents or wild animals . Veterinarians , abattoir workers and other occupations which require contact with animals are at risk of infection . Indirect contact can also cause infection in occupations such as rice field workers , sewer workers , and soldiers . Peak incidence occurs during the rainy season in tropical regions . Clinical manifestations of leptospirosis are non-specific , varying from subclinical infection , through self-limited anicteric febrile illness with or without meningitis , to severe and potentially lethal multisystem illness with jaundice and renal failure [6] , [7] . Recent findings show that leptospirosis is a common cause of undifferentiated febrile illness in developing countries [3] , [8] , [9] . Early diagnosis of leptospirosis is essential since antibiotic therapy provides greatest benefit when initiated early in the course of illness [6] , [7] . Diagnosis at an early phase , however , is hampered by the non-specific presentation of leptospirosis . A number of diagnostic tests for leptospirosis are available , all of these test are aimed to detect specific antibody against pathogenic Leptospira . These tests have so far shown low levels of accuracy , questioning their usefulness in the selection of appropriate antimicrobial treatment [10] . Nonetheless , these tests are often used in routine practice in many clinical settings . The objective of this study is therefore to determine whether it is cost-effective to perform these screening tests for leptospirosis in patients with acute febrile illness suspected of mild leptospirosis , and if so , which diagnostic assay is most cost-effective , in the outpatient setting .
We developed a decision tree with Markov nodes [12] to compare the costs and outcomes of a hypothetical cohort of 10 , 000 patients with suspected leptospirosis under each of the strategies ( Figure 1 ) . All patients began the simulation with acute fever and symptoms suggestive of leptospirosis . The model simulated the natural history of suspected leptospirosis progression over a 7-day period , during which the patients make various possible transitions; if they are sick , they may remain sick , become well , become sick with an antibiotic side effect , or develop a serious disease complication . A 7-day limit was applied because most patients became afebrile within this period , as shown in our concurrent clinical study [11]; the remaining proportion of febrile patients at the end of the 7 day period was also calculated . The model estimated the number of fever days associated with each strategy by summing the daily proportions of patients without fever . Given the short time horizon of 7 days , costs and outcomes were not discounted . Limited information was available for many of the parameters in the analysis . Where possible , we used the results of our concurrent clinical study of adult patients with acute undifferentiated febrile illness suspected of leptospiorsis [11 . This study was conducted between July 2003 and January 2005 at 5 hospitals in Thailand . The results of this study showed that leptospirosis and rickettsial infection ( mainly scrub typhus and murine typhus ) had similar clinical manifestations , and accounted for approximately 50% of the cause of acute undifferentiated fever . This was supplemented where necessary with further data from the literature and expert opinion . Tables 1 through 4 list the values used for the variables in the model , the range of values tested in the sensitivity analyses , and the data sources . Lateral flow ( Lepto Tek , BioMerieux , The Netherlands ) is a one step colloidal gold immunoassay . It is based on the binding of specific IgM antibodies to the broadly reactive heat- extracted antigen prepared from the non-pathogenic Patoc 1 strain [13] . This test was first tested in south Andaman , India in 2003 [13] . Microcapsule agglutination test ( MCAT , Japan Lyophilization Lab . Tokyo , Japan ) is a passive agglutination assay , using microcapsule particles of a stable synthetic polymer to which surface cell components of mixture of 6 sonicated Leptospira spp . ( serogroup Australis , Autumnalis , Hebdomadis , Canicola , Icterohaemorrhagia , and Pyrogenes ) are adsorbed [14] . This test was developed and first tested in Japan and had been used in Thailand since 1997 [15] . Latex agglutination test ( National Institute of Health , Ministry of Public Health of Thailand ) is a latex agglutination test to detect Leptospira-specific antibodies . This test was developed by National Institute of Health , Ministry of Public Health of Thailand and has been used in Thailand since 2001 [16] . These three tests are widely used for the diagnosis of leptospirosis in Thailand . Field evaluations in endemic areas and in a clinical study in Thailand indicated that their performance was characterized by low sensitivities during acute-phase illness [10] , [11] , [13] . The data are shown in table 2 . The costs of these tests were similar . Prevalence estimates of leptospirosis in different settings varied from 7–36 . 9% [3] , [8] , [9] , [11] . The prevalence of leptospirosis in our latest concurrent clinical study was 26% [11] , which we used as a baseline in this study . We considered any major organ dysfunction such as acute renal failure , hypotension , acute respiratory failure [17] associated with leptospirosis to be a proxy for severe disease . Although mild leptospirosis could be self-limiting , complications of untreated leptospirosis occurred between 10–28% [11] , [18]–[20] . We used a baseline estimated of 10% . We used data from our concurrent clinical study [11] and findings from the literature on treatment options and their outcomes [21] . Among patients who presented with acute undifferentiated fever suspected of being leptospirosis , duration of fever and rate of complications in patients without leptospirosis or scrub typhus that did not receive doxycycline therapy were unknown , and were estimated to be similar to patients who had leptospirosis but did not receive doxycycline .
Not providing antibiotic treatment to any patient yielded the longest average duration of fever of 5 . 35 days ( 95%CI 5 . 32–5 . 39 ) , and the worst cure rate , but avoided all antibiotic side effects . Empirical antibiotic treatment yielded the shortest duration of fever , averaging 2 . 24 days ( 95%CI 2 . 23–2 . 25 ) , the highest cure rate at the end of the first week , and avoided any complications; it also , however , led to the highest rate of antibiotic side effects ( Table 5 ) . The application of lateral flow , MCAT or latex tests yielded average durations of fever of 4 . 66 ( 95%CI 4 . 70–4 . 89 ) , 4 . 83 ( 95%CI 4 . 80–4 . 87 ) and 4 . 30 ( 95%CI 4 . 25–4 . 34 ) , days respectively . Between 5–9% of patients developed complications , dependent on the exact test sensitivity and specificity . The baseline strategy of no-antibiotic- treatment incurred an average cost of 13 . 3 USD per patient . With no initial diagnosis or treatment costs , this expenditure is due entirely to patients that develop complications , not having received initial treatment . Empirical treatment of all patients incurred far lower expenditure than the other strategies , with an average cost per patient of 2 . 7 USD . This is comprised mostly ( 74% ) of the initial treatment cost , with the remaining expenditure due to the treatment of side-effects . The diagnostic tests all incurred similar costs , between15 . 3 and 17 . 2 USD . These costs were largely driven by the treatment of severe complication ( approximately 60% ) , with the tests themselves being the second largest component ( approximately 35% ) . When comparing the costs and productivity gains of the different strategies , empirical treatment clearly dominates the rest , being both the cheapest and most effective option . When compared to the no-antibiotic-treatment option , use of all three diagnostic tests provides some productivity gains , but also incurs higher costs . Accounting for the differences in costs and productivity , only the latex test had a BCR above one ( 2 . 68 ) , while the BCRs for the lateral flow and MCAT tests were 0 . 71 and 0 . 75 , respectively ( Table 6 ) . We performed sensitivity analyses to examine the effect of varying parameter values used in the analysis as specified in tables 2 and 3 and using a leptospirosis prevalence of 10% and 35%; this did not significantly alter results therefore these are not presented in detail . With lower or higher prevalence , empirical treatment remained the least costly and most effective strategy . Increased test sensitivity and specificity to 95% had little impact on overall costs and outcomes ( Table 5 ) . Increase test accuracy in the high leptospirosis prevalence area has much more impact when compared with the lower leptospirosis prevalence area , in terms of cost per patient ( 14 . 93 vs . 17 . 22 USD ) , duration of fever ( 4 . 14 vs . 4 . 88 days ) , percentage of patients with fever at day 7 ( 4 . 14% vs . 4 . 88% ) and percentage of patients with complications ( 5 . 8% vs . 7 . 7% ) . Varying the costs of the tests and of doxycycline within a reasonable range did not alter the results in favor of either of the tests; in fact doxycycline would have to cost over 29 USD , well beyond its current price , for the latex test to become a more efficient strategy .
Leptospirosis has become an important public health problem worldwide [22] . Much emphasis has been placed on the development of improved serologic tests that use whole cell Leptospira antigen preparations . Commercial whole –based assays are available in rapid formats amenable for ‘point- of- care’ use . Field evaluations indicate that these assays are characterized by low sensitivities ( 39–72% ) during acute – phase illness [10] , [22] . Lateral flow , MCAT , and latex tests are widely used assays for the diagnosis of leptospiorsis in Thailand . Because of their relatively high costs and low sensitivities , use of these tests for the initial management of acute leptospirosis was inferior to empirical treatment , and only the latex test was cost-effective when compared to the no-antibiotic- treatment option . Empirical treatment with doxycycline was found to be the most cost-effective strategy , being both cheap and effective in treating uncomplicated leptospirosis and other causes of febrile illness . Results from our concurrent clinical study showed that acute undifferentiated fever , i . e . acute fever without an obvious focus of infection , is the most common clinical presentation of both leptospirosis and scrub typhus [11] . Antibiotic treatment with either doxycycline or azithromycin shortened the duration of fever in both these and other illnesses . The use of a diagnostic test therefore implies that where a test provides a negative result , true or false , the patient would be denied a potentially effective treatment . Treatment failure occurred in 2% of intended-to- treat patients . The limitation of doxycycline empirical therapy was nausea/vomiting which occurred in about 24 . 3% of doxycycline treated patients and severe adverse events ( rash and severe vomit ) developed in 2 out of 145 treated patients [11] . Other epidemiological studies showed that “systemic infection” such as dengue infection , rickettsial infection , brucellosis , Q fever , CMV or EBV infection was common causes of acute fever syndrome , and in many cases fever disappeared without specific diagnosis being established [23] . Doxycycline was the most common empirical antimicrobial therapy for this syndrome and all episodes resolved without further complications . Shorter duration of fever among those patients who received treatment was also observed [23] . A primary limitation of this analysis is that it did not account for the potential harm to society that results in the overuse of doxycycline , which could eventually lead to increased bacterial resistance , reduced future effectiveness , and increased drug costs . Leptospiral resistance to doxycycline has not been described . More studies to clarify the link between individual antibiotic use and emerging community resistance are needed . If policy makers wish to prioritize the reduction in antibiotic use with the use of a diagnostic test , this study shows that the use of the latex test can be considered cost-effective . The data used in this analysis concerning doxycycline treatment was determined from our prospective clinical study [11] . The data used for the no-antibiotic-treatment group on the other hand was limited by its reliance on published data and expert opinion . The sensitivity analysis however showed that variation in these parameters , within reason , did not have a significant impact on results . Lastly , the costs for direct medical expenses was based on charges , rather than actual economic costs; based on familiarity with the Thai healthcare system we assume that actual costs are higher , therefore the cost-effectiveness of the intervention is likely to be slightly higher than that found in the analysis . In summary leptospirosis is a common cause of acute undifferentiated fever in rural areas where limited resources are available . Results of this study and other clinical studies [11] , [23] show that empirical treatment with doxycycline would be the most cost-effective option for these patients , as this strategy was also beneficial for patients with other diseases which clinically mimic leptospirosis such as scrub typhus . It should be noted that this strategy applies only to adult patients with acute fever suspected of mild leptospirosis . Patients with potentially more serious diseases should be treated more aggressively than implied by this analysis . | Symptoms and signs of leptospirosis are non-specific . A number of diagnostic tests for leptospirosis are available . We compared the cost-benefit of 5 management strategies: 1 ) no patients tested or given antibiotic treatment; 2 ) all patients given empirical doxycycline treatment; patients given doxycycline when a patient is tested positive for leptospirosis using: 3 ) lateral flow; 4 ) MCAT; 5 ) latex test . Outcomes were measured in duration of fever which is then converted to productivity losses to capture the full economic costs . Empirical doxycycline treatment was found to be the most efficient strategy , being both the least costly alternative and the one that resulted in the lowest average duration of fever . The significantly higher relative cost of using a diagnostic test as compared with presumptive treatment , and the limited sensitivity of all the diagnostic tests implied that only the latex test could be considered cost-effective when compared with the no-antibiotic-treatment option , and that all three tests were still inferior to empirical treatment . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/neglected",
"tropical",
"diseases"
] | 2010 | Strategies for Diagnosis and Treatment of Suspected Leptospirosis: A Cost-Benefit Analysis |
Visceral Leishmaniasis ( VL ) has spread to many urban centers worldwide . Dogs are considered the main reservoir of VL , because canine cases often precede the occurrence of human cases . Detection and euthanasia of serologically positive dogs is one of the primary VL control measures utilized in some countries , including Brazil . Using accurate diagnostic tests can minimize one undesirable consequence of this measure , culling false-positive dogs , and reduce the maintenance of false-negative dogs in endemic areas . In December 2011 , the Brazilian Ministry of Health replaced the ELISA ( EIE CVL ) screening method and Indirect Immunofluorescence Test ( IFI CVL ) confirmatory method with a new protocol using the rapid DPP CVL screening test and EIE CVL confirmatory test . A study of diagnostic accuracy of these two protocols was done by comparing their performance using serum samples collected from a random sample of 780 dogs in an endemic area of VL . All samples were evaluated by culture and real time PCR; 766 out of the 780 dogs were tested using the previous protocol ( IFI CVL + EIE CVL ) and all 780 were tested using the current protocol ( DPP CVL + EIE CVL ) . Performances of both diagnostic protocols were evaluated using a latent class variable as the gold standard . The current protocol had a higher specificity ( 0 . 98 vs . 0 . 95 ) and PPV ( 0 . 83 vs . 0 . 70 ) than the previous protocol , although sensitivity of these two protocols was similar ( 0 . 73 ) . When tested using sera from asymptomatic animals , the current protocol had a much higher PPV ( 0 . 63 vs . 0 . 40 ) than the previous protocol ( although the sensitivity of either protocol was the same , 0 . 71 ) . Considering a range of theoretical CVL prevalences , the projected PPVs were higher for the current protocol than for the previous protocol for each theoretical prevalence value . The findings presented herein show that the current protocol performed better than previous protocol primarily by reducing false-positive results .
Visceral leishmaniasis ( VL ) is a major public health problem worldwide . This disease in Brazil and Europe is caused by the protozoan parasite Leishmania infantum , which is transmitted to humans by the bite of sandflies from the genus Lutzomyia [1] . Dogs are considered the main reservoir of urban VL since: i ) these animals harbor high parasitism in skin that offers a high capacity of parasite transmission to sandflies , ii ) humans and dogs coexist in close proximity and iii ) canine cases generally precede the occurrence of VL in humans [1–4] . The identification and euthanasia of serologically positive dogs is one of the primary VL control strategies recommended by the governments of some countries , such as Brazil . The use of accurate diagnostic tests for canine VL ( CVL ) can reduce failures on VL control program by minimizing maintenance of false-negative animals and culling of false-positive dogs that impact on euthanasia controversial measure , subsequently , decreasing dog owners’ compliance and society disagreement . More accurate tests could also reduce the number of false-negative dogs that are maintained in endemic areas [4] . CVL is typically diagnosed by parasitological , serological and molecular tests . In December 2011 , the program of the Brazilian Ministry of Health for monitoring and control of leishmaniasis replaced the enzyme-linked immunosorbent assay ( EIE CVL ) screening method and the indirect immunofluorescence assay ( IFI CVL ) confirmatory test with a new serodiagnostic protocol for CVL composed of the Dual Path Platform ( DPP CVL ) screening test and the EIE CVL confirmatory test [5] . The evaluation of sensitivity and specificity revealed low values for previous protocol that detects infection by determining seropositivity in dogs . This low performance is probably due to undesirable preservation of blood samples normally collected onto filter papers . This simple procedure for sample collection is performed easily and facilitates sample storage and transportation . However , it often submits the biological specimens to stress conditions that might damage samples and lead to unreliable test results [2 , 6–8] . Additionally , the low sensitivity and specificity offered by the old protocol can be explained by further reasons: i ) both screening EIE CVL and confirmatory IFI CVL tests have been performed using blood samples that were collected in endemic areas and then sent to reference laboratories , where the tests were performed , ii ) EIE CVL and IFI CVL tests are time-consuming techniques , whereas IFI CVL has an additional difficulty to be standardized and interpreted depending on the ability of the observer to detect the antigen-antibody reaction by fluorescence microscope . This may lead to misinterpretation of the results and may compromise IFI CVL reproducibility in different laboratories . DPP CVL is a rapid test based on a multi-epitope , recombinant chimeric protein ( rK28 ) resulted from fusion of L . infantum genes: k9 , single repeat units of k39 and k26 [9] that has been adopted as the screening method in a new protocol established by the Brazilian government . DPP CVL rapid test is an immunochromatographic assay that offers several advantages: i ) rK28 was proven to provide very high levels of sensitivity and specificity for canine VL [9] , ii ) DPP CVL has a great potential for facilitating faster decision , since it is a point-of-care screening test that gives result within 15 minutes , iii ) DPP CVL in association with the confirmatory test EIE CVL give results within 15 days , in comparison to previous protocol ( EIE CVL + IFI CVL ) that results were only liberated after a lengthy time interval that varied from one to two months . Thus , the incorporation of this rapid test into the current protocol accelerates the implementation of the control measures in endemic areas . In addition , this procedure uses only small blood samples and does not require specialized equipment and supplies [10] . The use of tests presenting low accuracy has serious epidemiological consequences: false-negative dogs are undetected thereby maintaining the parasite life cycle in endemic areas , and detection of false-positive dogs results in excessive dog culling . The lack of a reliable gold standard test for CVL hinders the assessment of diagnostic protocol performance and can result in misinterpretation of diagnostic test accuracy [11–16] . Indeed , although the common used gold-standard , culturing of L . infantum , is highly specific , its low sensitivity [17] hampers the evaluation of other diagnostic techniques . In light of this limitation , latent class analysis ( LCA ) has been shown to be a valuable alternative to the classical validation approach of using parasitological methods as gold standards [18 , 19] . LCA is based on the theory that the observed results of different imperfect tests for the same disease are influenced by a latent common variable that cannot be directly measured , but can reflect accurately the true disease status . Previous studies employing LCA have accurately assessed serological [20–24] and molecular [12 , 25] diagnostic methods . Despite the advantages of DPP CVL [10 , 14 , 26 , 27] for CVL diagnosis , few studies have assessed its performance [28 , 29] . To the best of our knowledge , the present study is an initial attempt designed to compare the accuracy of the current ( DPP CVL and EIE CVL ) and previous protocol ( EIE CVL and IFI CVL ) for CVL diagnosis employing a latent class variable as the reference standard . Serum samples were obtained during a cross-sectional study performed in an endemic area for VL in Brazil .
All experimental procedures involving dogs were carried out according to the Brazilian Federal Law on Animal Experimentation ( Law no . 11794 ) , the guidelines for animal research established by the Oswaldo Cruz Foundation ( FIOCRUZ ) and the Brazilian Ministry of Health Manual for the Surveillance and Control of VL [4] . The Institutional Review Board approved the present study for Animal Experimentation ( CEUA , protocol no . 015/2009 ) . Dog owners who agreed to participate in the study signed a Free , Prior and Informed Consent ( FPIC ) form . A cross-sectional study was conducted in the municipality of Camaçari , located in the State of Bahia in Northeastern Brazil . Using district sketches of households throughout 36 districts in Camaçari obtained from the Zoonosis Control Center , a sample of domiciled dogs was randomly selected , during the years of 2011 and 2012 . The sample size was calculated using Epi Info 3 . 5 . 1 ( The Centers for Disease Control and Prevention—CDC , USA ) based on estimates of the canine population ( 15 , 820 dogs ) derived from an anti-rabies vaccination campaign and an expected CVL prevalence of 20% ( 5% margin of error , 95% confidence interval ) . Dogs were classified as asymptomatic or symptomatic based on the presence or absence of the following clinical signs: emaciation , alopecia , anemia , conjunctivitis , dehydration , dermatitis , erosion , ulcerations , lymphadenopathy , and onychogryphosis . They were classified as asymptomatic when presented 0 until 3 signs or symptomatic when presented more than 3 signs . Blood and splenic aspirate samples were obtained for CVL diagnosis from each dog at the same time . Blood was collected by venipuncture in sterile tubes to obtain serum . All serum samples were stored at -20°C until serological testing . Splenic aspirate samples were obtained using a puncture technique previously described by Barrouin-Melo and collaborators ( 2006 ) [30] , and modified by Solcà and collaborators ( 2014 ) for ultrasound-guided collection . All 780 splenic samples were evaluated by culture and real time PCR; 766 out of the 780 serum samples were tested using the previous protocol ( IFI CVL + EIE CVL ) and all 780 were tested using the current protocol ( DPP CVL + EIE CVL ) ( S1 Fig ) . Splenic aspirate samples were cultivated in Novy-Mac Neal-Nicolle ( NNN ) medium supplemented with 20% FBS ( Fetal Bovine Serum , Gibco BRL , New York , USA ) and 100 μg/mL of gentamicin . The cultures were maintained at 24°C for four weeks and examined weekly for the presence of parasites [31] . All serological diagnostic test kits for CVL ( DPP CVL , EIE CVL and IFI CVL Bio-Manguinhos ) were used in accordance with manufacturer’s recommendations . DNA was extracted from splenic aspirate samples using DNeasy Blood & Tissue kit from Qiagen ( Hilden , Germany ) , in accordance with manufacturer’s recommendations . DNA concentrations were determined using a digital spectrophotometer ( Nanodrop—ND-1000 Thermo Scientific , Wilmington , USA ) , then aliquoted at a concentration of 30 ng/μL and stored at -20°C until real time PCR amplification . DNA extracted from splenic aspirate samples was amplified using real time PCR technique , in accordance with the protocol established by Francino and collaborators ( 2006 ) [32] and modified by Solcà and collaborators ( 2014 ) . Control samples were added in all of the real time PCR experiments . As positive controls were used splenic aspirate samples from two dogs that had previously been identified in an endemic area as positive for Leishmania infection and as negative controls were employed splenic aspirates of two healthy dogs from the municipality of Pelotas , Rio Grande do Sul , Brazil , an area non-endemic for CVL . All test readers executing and reading the index tests had prior training and great experience in CVL diagnosis . All diagnostic testing was carried out under blinded conditions , which means that test readers interpreted the results obtained from each diagnostic technique for a given sample without knowledge of the other tests’ results . The interpretation of the results using the previous and current diagnostic protocols classified dogs as positive when both tests ( screening and confirmatory ) presented positive results . Epi Info 3 . 5 . 1 ( The Centers for Disease Control and Prevention—CDC , Atlanta , USA ) and STATA 12 . 0 ( StataCorp LP , Texas , USA ) software programs were used to analyze results . LCA was performed to define a latent class variable to evaluate the accuracy of the diagnostic tests and employed as previously described in Solcà and collaborators ( 2014 ) . Latent variable modeling used the results of the following diagnostic techniques as indictor variables: serological ( EIE CVL , DPP CVL and IFI CVL Bio-Manguinhos ) , parasitological ( culture of splenic samples ) , and molecular ( real time PCR of splenic aspirate ) tests . We chose a two-class latent class model based on goodness of fit criteria , such as the Akaike information criterion ( AIC ) and Bayes information criterion ( BIC ) . We also used the Lo-Mendel-Rubin test and the entropy for model evaluation [33] . MPlus version 5 software was used to implement LCA [34] . The performance of the diagnostic tests and protocols was estimated using the latent class variable as the reference standard . Diagnostic performance was calculated in 2 x 2 contingency tables of positive and negative test results , using the command diagt in Stata . We determined specificity , positive predictive values ( PPV ) , negative predictive values ( NPV ) and diagnostic accuracy with 95% exact binomial confidence intervals ( CI ) . Diagnostic accuracy was calculated as the number of true positive + number of true negative/total number of tested serum samples . Differences among diagnostic protocols regarding their performance ( sensitivity and specificity ) were assessed using McNemar chi-square test ( p-value < 0 . 05 ) , for all dogs and for two categories of disease status based on symptomatology . The number of animals considered as false negative and false positive was also calculated for each of the diagnostic techniques evaluated , considering as true positive those dogs that were positive according to the latent class variable .
From April 2011 until July 2012 , 780 dogs pure and mixed-breed with estimated ages from 1 to 10 years old , were enrolled in the study . According to the presence of clinical signs of CVL , 47 . 8% dogs were asymptomatics and 54 . 2% symptomatics . Five diagnostic tests were used to determine the proportion that tested positive in this random population . The IFI CVL yielded the highest percentage of positivity ( 36% ) , whereas the splenic aspirate culture yielded the lowest percentage of positivity ( 13 . 1% ) . Among the remaining tests , the EIE CVL , real time PCR and DPP CVL tests were positive in 24 . 9% , 22 . 4% and 16 . 9% of the dogs , respectively ( Fig 1 and Table 1 ) . Using LCA , 14 . 1% of the 780 dogs were classified as positive ( Table 1 ) . Evaluation of LCA entropy showed that a high accuracy in the classification of dogs by LCA was achieved , with value of 0 . 97 . A posteriori average probabilities that dogs were properly classified in the latent classes "positive" and "negative" were , respectively , 95% and 99% . Moreover , the test of Lo-Mendel-Rubin indicated that the model with two classes produced better results than that with only one class ( p < 0 . 01 ) . These results are supported by the analysis of AIC and BIC ( AIC = 3025 . 996 , BIC = 3077 . 249 ) . The real time PCR and culture techniques yielded the highest sensitivities , 0 . 97 and 0 . 90 , respectively , when the latent class variable served as the reference standard . Among the three serological tests , IFI CVL and DPP CVL had the highest sensitivity ( 0 . 86 ) and EIE CVL ( 0 . 79 ) ( Table 2 ) . Regarding specificity , culture was found to be the most specific ( 1 . 00 ) , followed by DPP CVL ( 0 . 94 ) , then real time PCR ( 0 . 90 ) , EIE CVL ( 0 . 84 ) and IFI CVL ( 0 . 73 ) . When the latent class variable was considered as the reference test , the PPV of culture was 1 . 00 . Among the other four techniques , DPP CVL had the highest PPV ( 0 . 71 ) , followed by real time PCR ( 0 . 61 ) , EIE CVL ( 0 . 45 ) and IFI CVL ( 0 . 37 ) . Likewise , among serological tests , DPP CVL ( 0 . 98 ) , followed by IFI CVL ( 0 . 97 ) and EIE CVL ( 0 . 96 ) showed the highest NPV ( Table 2 ) . The measures of diagnostic accuracy of the current diagnostic protocol were then compared to those of the previous protocol ( Table 3 ) . Both protocols had equally high sensitivity ( >0 . 72; McNemar’s chi-square test , p = 0 . 051600 ) and NPV ( 0 . 96 ) , whereas the new protocol consistently had a higher specificity ( >0 . 97 , p = 0 . 0078 ) and PPV ( >0 . 83 ) . The diagnostic accuracy was higher when current diagnostic protocol was compared to the previous protocol ( 0 . 94 vs . 0 . 92 ) . Comparing the performance of current protocol ( DPP CVL + EIE CVL ) to that of DPP CVL alone revealed that sensitivity showed higher value for DPP CVL ( 0 . 86 ) than that for the current protocol ( 0 . 73 ) , although PPV showed a slight lower value for DPP CVL ( 0 . 71 ) compared to PPV for the current protocol ( 0 . 83 ) . When the dogs were categorized according to the presence of clinical signs of CVL ( Table 3 ) , the sensitivity of both diagnostic protocols was similar in asymptomatic and symptomatic dogs . However , in symptomatic dogs , the new protocol had higher specificity and PPV ( 0 . 97 and 0 . 88 , respectively ) than the previous protocol ( 0 . 94 and 0 . 79 , respectively ) . In addition , in asymptomatic dogs , the PPV of the current protocol was significantly higher , by 22 . 5% , than that of the previous protocol ( p = 0 . 0078 ) . Also , difference was observed in diagnostic accuracy of protocols when they were used in symptomatic dogs ( 0 . 91 vs . 0 . 89 ) and asymptomatic dogs ( 0 . 97 vs . 0 . 95 ) . To generalize the better performance of current protocol to other settings , the PPV and NPV were calculated for the current and previous protocol accordingly to different theoretical values of CVL prevalence ( Table 4 ) . For each estimated prevalence value , the current protocol was estimated to yield higher PPVs , ranging from 0 . 23 to 0 . 99 , whereas the projected PPVs for the previous protocol ranged from 0 . 13 to 0 . 98 . Regarding NPV , both protocols yielded similar projected values , ranging from 0 . 47 to 1 . 00 .
The present study primarily demonstrated that the DPP CVL + EIE CVL protocol , in comparison with the EIE CVL + IFI CVL protocol , performed better for the serodiagnosis of CVL . The adoption of this new protocol offered several advantages , due to inclusion of the rapid DPP CVL screening test , which can be performed easily and quickly and does not require specialized equipment and personnel [27 , 28] . Several authors previously discussed that the lack of a perfect gold standard test for CVL hampers the evaluation of diagnostic tests for CVL [12 , 13] . Previous studies have proven that LCA is effective for evaluating diagnostic tests’ performance [12 , 20 , 21 , 35–38] . Herein , using a latent class variable as the reference standard , we were able to comprehensively compare two protocols for serodiagnosis of CVL using serum samples collected from 780 randomly selected dogs from an endemic area of VL . The use of LCA had an additional advantage: we were able to evaluate the performance of both real time PCR and culture . Very few studies have evaluated the performance of real time PCR , and most studies that evaluated the performance of CVL diagnostic techniques used culture as the gold standard [10 , 14 , 26 , 28] . Using LCA , we found that real time PCR and culture were the most sensitive techniques . Among the serological tests evaluated , DPP CVL had the best performance . Although IFI CVL had the highest sensitivity , it was the least specific , as previously described by de Santis and collaborators ( 2013 ) and Laurenti and collaborators ( 2014 ) . Regarding the performance measure that has epidemiological relevance , the PPV , the DPP CVL had the highest PPV among the serological tests evaluated as previously described by da Silva and collaborators ( 2013 ) . In addition to evaluating each individual test for CVL diagnosis , we compared the performance of previous and current protocols employed in Brazil . The usefulness of protocols was evaluated by determining PPVs and NPVs of each protocol . The better individual performance of the DPP CVL was reflected in the 13% higher PPV of the current protocol for CVL detection compared to the previous protocol . Both protocols yielded a NPV of 0 . 96 , suggesting that when these protocols have negative results it is highly probable that serum are from dogs that are actually uninfected . By contrast , the current diagnostic protocol provides a greater PPV ( 0 . 83 ) than that of the previous protocol ( 0 . 70 ) , indicating that the current protocol provides a greater level of assertiveness in diagnosing positive dogs . Although the new protocol showed a higher specificity and PPV than the previous one , the sensitivity is still limited ( around 0 . 73 ) in both protocols , meaning that the maintenance of false-negative dogs in endemic areas still represents a public health concern and more efforts should be done to try to find out better protocols or new antigens to reduce the maintenance of infected dogs in areas of zoonotic transmission . Considering questions rose about the wisdom to diagnose CVL using DPP CVL + EIE CVL instead of DPP CVL alone , the comparison of performances showed that a higher sensitivity value ( 0 . 86 ) and lower PPV ( 0 . 71 ) for DPP CVL compared to DPP CVL + EIE CVL ( 0 . 73 and 0 . 83 , respectively ) that might cause detection of false positive dogs . Mostly to avoid this , a confirmatory test , EIE CVL , has been associated to DPP CVL in the current protocol . ” When current protocol is applied for diagnosing asymptomatic and symptomatic dogs , it showed similar performance for sensitivity ( 0 . 71 , 0 . 73 ) and specificity ( 0 . 98 , 0 . 97 ) , respectively . While , the level of NPV ( 0 . 99 ) was greater , the level of PPV ( 0 . 63 ) was much lower for asymptomatic dogs in comparison to NPV ( 0 . 92 ) and PPV ( 0 . 88 ) for symptomatic animals . In accordance to this results , Otranto and collaborators ( 2009 ) showed that recently exposed or newly infected dogs might not be detected by serological tests , since these false-negative animals do not seroconvert soon after infection or they may develop a cellular type of immune response that are not detected using serological tests . In addition to this difficulty , no appropriate gold standard for Leishmania infection detection in asymptomatic dogs was established , highlighting the necessity for the development of new tests to improve diagnosis of asymptomatic dog . Across a range of plausible prevalence , the theoretical expectation for PPV varied among 0 . 13 to 0 . 98 for previous protocol , and 0 . 23 to 0 . 99 for current protocol . PPV and NPV of a diagnostic test are known to be influenced by the prevalence of a given disease in a population . Thus , as disease becomes more prevalent the probability of subjects to test positive in diagnostic tests will be higher among sick individuals . In the present study , the analysis using different theoretical prevalence revealed that the current protocol has high performance irrespective of disease prevalence . In accordance , higher PPVs provided by DPP CVL + EIE CVL for diagnosing CVL have additional advantages since in endemic countries , regardless of the prevalence of CVL , the current protocol compared to previous one would better discriminate truly uninfected dogs from those that have risky to be infected . In summary , our findings show that the current protocol for diagnosis of CVL implemented in Brazil has an excellent accuracy ( 0 . 91 for symptomatic dogs and 0 . 97 for asymptomatic ) , due to its greater specificity values and PPV . Because of the simplicity of test procedures and rapidity of results , the data presented herein strongly support the idea that the introduction of DPP CVL into the diagnostic CVL protocol contribute to improve CVL diagnosis that can have consequent effects that impact positively on disease control . | Visceral Leishmaniasis ( VL ) is a major public health problem . Its control is based on detection and culling of positive dogs , treatment of human cases and vector control . Canine cases often precede the occurrence of human cases; hence , disease control in dogs is important . Use of accurate diagnostic tests is required to avoid culling false-positive dogs and to minimize the number of false-negative dogs that are maintained in endemic areas . In December 2011 , the Brazilian Ministry of Health changed the diagnostic protocol for canine VL ( CVL ) . In the present study , the accuracy of this current protocol was compared to the previous one using serum samples of 780 dogs from an endemic area of VL . The findings revealed that the current protocol performed better than the previous protocol primarily by reducing false-positive results . Considering different theoretical prevalence values , the current protocol misdiagnosed fewer dogs than the previous one . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2016 | The Rapid Test Based on Leishmania infantum Chimeric rK28 Protein Improves the Diagnosis of Canine Visceral Leishmaniasis by Reducing the Detection of False-Positive Dogs |
Parasitic nematodes impose a debilitating health and economic burden across much of the world . Nematode resistance to anthelmintic drugs threatens parasite control efforts in both human and veterinary medicine . Despite this threat , the genetic landscape of potential resistance mechanisms to these critical drugs remains largely unexplored . Here , we exploit natural variation in the model nematodes Caenorhabditis elegans and Caenorhabditis briggsae to discover quantitative trait loci ( QTL ) that control sensitivity to benzimidazoles widely used in human and animal medicine . High-throughput phenotyping of albendazole , fenbendazole , mebendazole , and thiabendazole responses in panels of recombinant lines led to the discovery of over 15 QTL in C . elegans and four QTL in C . briggsae associated with divergent responses to these anthelmintics . Many of these QTL are conserved across benzimidazole derivatives , but others show drug and dose specificity . We used near-isogenic lines to recapitulate and narrow the C . elegans albendazole QTL of largest effect and identified candidate variants correlated with the resistance phenotype . These QTL do not overlap with known benzimidazole target resistance genes from parasitic nematodes and present specific new leads for the discovery of novel mechanisms of nematode benzimidazole resistance . Analyses of orthologous genes reveal conservation of candidate benzimidazole resistance genes in medically important parasitic nematodes . These data provide a basis for extending these approaches to other anthelmintic drug classes and a pathway towards validating new markers for anthelmintic resistance that can be deployed to improve parasite disease control .
Parasitic nematodes impose a debilitating health and economic burden across much of the developing world , conservatively resulting in the loss of 14 million disability-adjusted life years per annum [1] . This disease burden is partly curtailed by mass drug administration ( MDA ) programs that depend on the continued efficacy of a limited portfolio of anthelmintic drugs . Benzimidazoles are an indispensable component of this limited chemotherapeutic arsenal , and three benzimidazole derivatives are considered ‘Essential Medicines’ by the World Health Organization . In veterinary medicine , aggressive benzimidazole chemotherapy has generated geographically widespread benzimidazole resistance [2] . The ongoing expansion of anthelmintic coverage in humans threatens a similar outcome [3 , 4] . Reduced cure rates suggestive of benzimidazole resistance have been reported for human soil-transmitted helminths , including the etiological agents of hookworm infection , trichuriasis , and ascariasis [5–9] . The sustainability of chemotherapy-based parasite control is jeopardized by our deficient knowledge of potential mechanisms of anthelmintic resistance and a paucity of molecular markers to detect and slow the spread of resistance alleles in parasite populations . Traditional approaches to identify anthelmintic resistance markers rely on surveying candidate genes for polymorphisms [10] . Known mechanisms of nematode benzimidazole resistance have been limited to variants in the drug target beta-tubulin [10–13] . However , genetic differences in beta-tubulin genes do not explain all interspecific and intraspecific variation observed in benzimidazole efficacy [14] or in responses to different benzimidazole derivatives [15 , 16] . A more complete understanding of potential pathways to parasite benzimidazole resistance is necessary to help discover loci that account for drug response differences not entirely explained by polymorphisms in beta-tubulin genes . Additionally , exploring differences among benzimidazoles derivatives could be useful in helping optimize their use . Although C . elegans has been historically indispensable to the discovery of mechanisms of action for benzimidazoles and other anthelmintics [17] , genetic variation within C . elegans has only recently been exploited to study phenotypic variation in anthelmintic responses [18] . We used standing genetic variation and high-throughput quantitative phenotyping in two experimentally tractable Caenorhabditis species , C . elegans and C . briggsae , to identify genomic loci that control susceptibility to four clinically relevant benzimidazoles . Our findings reveal both conserved and drug-specific loci in each species that contribute to the effects of benzimidazoles on animal offspring production and growth rate . The genetic architectures of benzimidazole sensitivity and the specific genomic loci identified in this work provide new leads to identify genetic markers and molecular mechanisms that govern anthelmintic resistance in parasitic nematodes . We expect that translation of these leads will help to improve the detection and management of drug resistance in parasite populations .
Strains were propagated for four generations to reduce transgenerational effects of starvation and bleach-synchronized before transfer to 96-well growth plates ( ∼ 1 embryo/μl in K medium ) . Hatched L1 larvae were fed HB101 bacterial lysate ( 5 mg/ml ) and incubated for 48 hours at 20°C . L4 larvae were sorted into 96-well drug plates ( three animals/well ) using the COPAS BIOSORT large particle sorter ( Union Biometrica ) . Drug plates contained anthelmintics dissolved in K medium at the desired final concentrations along with 1% DMSO , 10 mg/ml HB101 bacterial lysate , and 31 . 25 μM kanamycin . These cultures were incubated for 96 hours at 20°C to allow development to the adult stage and the maturation of deposited embryos . Animals were fed a solution of 1 mg/ml bacterial lysate and 0 . 01 μM red fluorescent microspheres ( Polysciences , cat . 19507-5 ) for five minutes prior to scoring . Animals were immobilized with 50 mM sodium azide , and the COPAS BIOSORT large particle sorter was used to measure a range of animal fitness traits including length , pharyngeal pumping ( red fluorescence ) , and brood size [19 , 20] . Raw phenotype data collected from the COPAS BIOSORT large particle sorter were processed with the R package easysorter [21] . The function read_data ( ) was used to distinguish animals from bubbles using a support vector machine ( SVM ) . The functions remove_contamination ( ) and sumplate ( ) were used to mask contaminated wells and to calculate summary statistics across measured parameters . Parameters included time-of-flight ( animal length ) , extinction ( optical density ) , fluorescence ( pharyngeal pumping ) , and total object count ( brood size ) . Summary statistics included the mean and quantiles ( 10th , 25th , 50th , 75th , and 90th ) for each of these parameters . Brood size was normalized to the sorted number of animals per well ( n ) , and fluorescence was normalized to animal length ( TOF ) . The regress ( assay = TRUE ) function was used to fit a linear model to account for differences in assays carried out on different days . Outliers were defined as observations that fall outside the IQR by at least twice the IQR and that do not group with at least 5% of the observations . Outliers were removed using the bamf_prune ( ) function and regress ( assay = FALSE ) was used to fit a linear model ( phenotype ∼ control phenotype ) to calculate drug effects with respect to control ( DMSO solvent ) conditions . The dose-dependent phenotypic effects of benzimidazoles were assayed in technical quadruplicate across four genetically diverged C . elegans ( N2 , CB4856 , DL238 , and JU258 ) and C . briggsae ( AF16 , HK104 , VT847 , and ED3035 ) strains . Phenotypes were measured using the high-throughput assay [19] and trait generation pipeline described above . Drug concentrations for linkage mapping experiments were selected based on broad-sense heritability calculations for traits of interest and with the goal of maximizing differences in sublethal drug effects between the parental strains used to generate recombinant lines ( C . elegans: N2 and CB4856; C . briggsae: AF16 and HK104 ) . We did not calculate LC50 and EC50 values based on dose response data given that this assay is generational and we are simultaneously measuring the effects of prolonged drug exposure on both the seeded animals and their progeny . We could consider brood size the parameter most closely linked to lethality , however , this trait likely conflates suppression of fecundity ( sterility in the adults ) and true drug-induced lethality ( i . e . , “killing” of eggs and young larvae ) . The dose response data are replicated within assay and not performed in large block replicates . Therefore , absolute trait differences among the strains are not definitive data as compared to the highly replicated parent and near-isogenic line ( NIL ) values discussed below . Benzimidazole exposure phenotypes were measured for a population of 292 unique C . elegans recombinant inbred advanced intercross lines ( RIAILs ) resulting from an advanced intercross of N2 and CB4856 [19 , 22] , as well as 153 unique C . briggsae recombinant inbred lines ( RILs ) created using AF16 and HK104 [23] . These phenotypic data were collected and processed as described above . R/qtl [24] was used to carry out marker regression on 1454 C . elegans markers and 1031 C . briggsae markers . QTL were detected by calculating logarithm of odds ( LOD ) scores for each marker and each trait as −n ( ln ( 1 − r2 ) /2ln ( 10 ) ) , where r is the Pearson correlation coefficient between RIAIL genotypes at the marker and phenotype trait values [25] . Significance thresholds for QTL detection were calculated using 1000 permutations and a genome-wide error rate of 0 . 05 . The marker with the maximal LOD score exceeding significance was retained as the peak QTL marker for each of three mapping iterations . QTL confidence intervals were defined by a 1 . 5 LOD drop from peak QTL markers . The reproducibility of traits across divergent individuals and/or in parent-offspring relationships is often referred to as heritability . Broad-sense heritability is calculated as the amount of phenotypic variance that is caused by genetic differences across a population . We collected repeated measures of parental and recombinant strain phenotypes in independent assays and then fitted a linear mixed model to calculate broad-sense heritability , as described previously [26] . Near-isogenic lines ( NILs ) were generated by backcrossing N2xCB4856 RIAILs to either parental strain for six generations , followed by six generations of selfing to homozygose the genome . Primers were optimized to genotype N2xCB4856 insertion-deletion variants immediately flanking introgression regions of interest . These lines were whole-genome sequenced at low coverage to confirm genotypes at all loci . NIL reagents , primers , and PCR conditions are detailed in S1 Methods . Existing C . elegans mutants ( alg-4 ( tm1184 ) ; alg-3 ( tm1155 ) , ergo-1 ( tm1860 ) , prg-1 ( n4357 ) , ben-1 ( tm234 ) ) were propagated alongside N2 and CB4856 to assay albendazole-response phenotypes . The prg-1 mutant strain was backcrossed to N2 for 10 generations . Independent prg-1 deletion strains were generated by CRISPR/Cas9-mediated gene editing using Cas9 ribonucleoprotein [27] . A co-CRISPR strategy targeting dpy-10 was used to improve efficiency of screening for edits [28] . All Cas9 reagents were purchased through IDT ( Skokie , IL ) . Alt-R tracrRNA ( IDT , 1072532 ) , dpy-10 crRNA , and each prg-1 crRNA ( oECA2002 and oECA2003 ) were combined and incubated at 95°C for five minutes . Cas9 nuclease ( IDT , 1074181 ) was added , and the mix was incubated at room temperature for five minutes . Finally , the dpy-10 repair template was added , and the final volume was brought to 5 μL with nuclease-free water . The final concentrations are as follows: tracrRNA 13 . 6 μM , dpy-10 crRNA 4 μM , each prg-1 crRNA 9 . 6 μM , Cas9 23 . 8 μM , and dpy-10 repair construct 1 . 34 μM . The mix was centrifuged at maximum speed for five minutes , mouth pipetted into a pulled injection needle ( World Precision Instruments , 1B100F-4 ) , and injected into gravid adult N2 and CB4856 animals . Each injected animal was placed onto an individual 6 cm NGM plate approximately 18 hours post-injection . Rol F1 animals were placed onto individual 6 cm NGM plates when they reached the L4 stage , or when the Rol phenotype was apparent , and allowed to lay embryos . These F1 animals were then screened for large deletion events with PCR using primers oECA2004 and oECA2042 . Non-Dpy , non-Rol progeny from edited F1 animals were propagated until homozygous and verified with Sanger sequencing . Primers and other reagents used to genotype backcross progeny and to generate CRISPR alleles are outlined in S1 Methods . Reagents , primers , and PCR conditions are detailed in S1 Methods . NIL and mutant phenotyping assays were carried out with the high-throughput pipeline described above with at least two independent biological replicates carried out for each strain panel . Phenotype data are shown as Tukey box plots . Analyses were performed using R by one or two-tailed t-test ( for two groups and specific hypotheses about direction of effect ) or one-way ANOVA with Tukey’s multiple comparison test ( for more than two groups ) . P-values less than 0 . 05 were considered significant . P-values for all statistical tests are provided in S9 Table . C . elegans variants distinguishing N2 and CB4856 [29] were used to annotate QTL intervals with respect to existing gene annotations . C . briggsae variants distinguish AF16 and HK104 as well as their estimated functional consequences were produced from genomic sequence and annotation data using SnpEff [30] . For all C . elegans QTL identified in this study , parasite orthologs of C . elegans genes with variants predicted to be of ‘moderate’ or ‘high’ impact were extracted from Wormbase ( WS255 ) using custom Python scripts . Caenorhabditis orthologs from the clade IV gastro-intestinal parasite Stronygloides ratti and the clade III human filarial parasite Brugia malayi were extracted from WormBase [31] . Orthologs are reported across all QTL for completeness , but S8 Table provides orthologs specific to each QTL mapped . C . elegans strains N2 and CB4856 were bleach-synchronized and grown at 20°C for isolation of total RNA from young adult animals ( 60 hours post-embryo plating ) using a liquid N2 freeze-cracking protocol with TRIzol ( Life Technologies ) . RNA was collected from four independent biological replicates per strain . Sample RNA concentration and quality were assessed via Agilent Bioanalyzer . mRNA libraries were prepared from RNA samples using the TruSeq Stranded mRNA Library Prep Kit with oligo-dT selection ( Illumina ) . All samples were sequenced using the Illumina HiSeq 2500 platform with a single-end 50 bp read setting ( University of Chicago Genomics Facility ) and demultiplexed for downstream analyses . Mean per-sample yield was 10 . 0 Gb for mRNA-seq samples . Reads were adapter and quality trimmed using Trimmomatic [32] . HiSAT2 and StringTie [33] were used to align reads to the N2 reference genome ( WormBase . org [31] release WS255 ) and to produce raw and TPM ( transcripts per million ) read counts for annotated genes , respectively . DESeq2 [34] was used to identify differentially expressed genes . The complete RNA-seq pipeline was implemented with Nextflow [35] and is publicly available through GitHub ( www . github . com/AndersenLab/BZRNA-seq-nf ) .
We examined natural variation in C . elegans responses to four benzimidazoles ( albendazole , fenbendazole , mebendazole , and thiabendazole ) that are widely used in human and veterinary medicine . Dose responses were performed on a set of four genetically diverged strains using a flow-based , large-particle analysis device ( COPAS BIOSORT , Union Biometrica ) for high-throughput quantification of anthelmintic effects on animal fitness traits , including offspring production and growth rate . This platform enables trait measurements at unprecedented scales in a metazoan system [19] ( S1 Fig ) . In brief , strains are synchronized and dispensed into 96-well microtiter plates containing anthelmintic drugs or no drug ( DMSO ) . Conditions are strictly controlled for temperature , humidity , food source , and culture mixing . The COPAS BIOSORT measures the length , optical density , and fluorescence ( green , yellow , and red ) of every nematode from populations grown in 96-well microtiter plates . This high-throughput platform enables the measurement of animal size , fecundity , and feeding behavior . As C . elegans grows , animals get longer [36] , so length measurements are a proxy for developmental stage and growth rate . Fecundity is assessed by counting the number of offspring produced by a defined number of parent animals in each well . Feeding rate is quantified by exposing animals to fluorescent microspheres ( Fluoresbrite Fluorescent Microspheres , Polysciences Inc . ) , which are the same size as bacterial food , and then measuring fluorescence after a defined period of time . Anthelmintic drugs reduce growth rate , offspring production , and muscle activity , so these traits are directly relevant to anthelmintic mechanisms of action in parasitic nematodes [37 , 38] . Assay measurements with longer animals , more offspring , and/or more fluorescence indicate that that strain is more resistant to an anthelmintic than strains with shorter animals , fewer offspring , and/or less fluorescence . All drugs showed dose-dependent effects for at least one of these major phenotypic categories ( S2 Fig ) . Concentrations that exhibited high broad-sense heritability and strain-specific differences were chosen for quantitative genetic mappings . These dose response data were used to quickly assess doses for mapping potential , but due to lack of biological replication , do not support robust and statistically meaningful claims about trait differences between individual strains . Single concentrations of albendazole ( 12 . 5 μM ) and mebendazole ( 20 μM ) were selected . To assess the potential effects of dose on the genetic architecture of drug sensitivity , two concentrations of fenbendazole ( 15 μM and 30 μM ) and thiabendazole ( 62 . 5 μM and 125 μM ) were selected . These sublethal concentrations fall within the range of previous studies in C . elegans [11 , 39] and likely correspond to pharmacologically relevant drug accumulation levels in parasitic nematodes . This conclusion is supported by observations that the C . elegans cuticle is generally less permissive and that exogenous drug concentrations required to elicit effects are often orders of magnitude higher in this model system [40–42] . From our dose response assays of four genetically diverse C . elegans strains , we found that the laboratory strain from Bristol , England ( strain N2 ) and a wild strain from Hawaii , U . S . A . ( strain CB4856 ) differed in responses to benzimidazoles . We can use the natural differences between these two strains to identify the variants that contribute to divergent anthelmintic responses . Previously , these two strains were crossed to create a collection of recombinant inbred advanced intercross lines ( RIAILs ) to facilitate linkage mapping approaches [19 , 22] . Using 292 RIAILs , we identified 15 quantitative trait loci ( QTL ) that each explain greater than 5% of trait variation in benzimidazole susceptibility ( Fig 1A , S1 Table ) . Many of these QTL span multiple animal fitness traits ( S3 Fig ) . The trait groupings can be observed by the correlation structure of measured parameters and are robust across summary statistics ( mean , median , 75th quantile , and 90th quantile ) for these parameters ( S4 Fig ) . Benzimidazole sensitivity involved the contributions of multiple loci for many drug-trait combinations . Within this complex trait landscape , we identified QTL that are common but also some that are unique across drugs and doses . Drug-specific QTL provide new leads to explain observed differences in in vitro bioactivity [43 , 44] and clinical efficacy [45–48] among benzimidazole derivatives . Although mammalian host factors are likely a major source of variation in clinical efficacy among benzimidazoles , drug-specific QTL may help explain variation resulting from factors that affect drug bioactivity in the nematode context ( e . g . , uptake , metabolism , and potency ) . By contrast , dose-specific QTL reveal the engagement of different genetic determinants of benzimidazole response as a function of drug exposure . We discovered a major QTL associated with the effects of albendazole on animal size traits . This QTL is localized to a 441 kb interval on chromosome IV ( 15 . 47–15 . 91 Mb ) and explains 36% of albendazole-induced variation in animal length among the strains ( Fig 1B ) . This major-effect albendazole QTL extends to pharyngeal pumping and is also associated with small differences in brood size ( S3 Fig ) . An overlapping QTL associated with animal size and pumping behavior was identified for 30 μM fenbendazole and 20 μM mebendazole . A distinct size and pumping-associated QTL was mapped for 20 μM mebendazole on the left arm of chromosome V that does not overlap with loci identified for other benzimidazole derivatives . Another highly significant QTL on chromosome V ( 14 . 15–14 . 72 Mb ) explains 29% of thiabendazole ( 125 μM ) -induced variation in brood size between the strains . This QTL extends to length and pharyngeal pumping traits for both 62 . 5 μM and 125 μM thiabendazole , and partly overlaps with a QTL associated with differences in pumping behavior in 15 μM fenbendazole . A number of unique loci with smaller effect sizes were mapped for fenbendazole sensitivity at 15 and 30 μM , with only one overlapping QTL apparent across the two drug concentrations . Plots of phenotype as a function of peak QTL marker genotype ( Fig 1C ) show that the direction of effect varies across drugs and traits , indicating that both N2 and CB4856 strains possess alleles contributing to both benzimidazole sensitivity and resistance . The albendazole QTL confidence interval contains 17 protein-coding genes with coding variants and falls within the large Piwi-interacting RNA ( piRNA ) cluster on chromosome IV ( 13 . 5–17 . 2 Mbs ) [49] . In the linkage mapping experiment , recombinant strains with the N2 genotype at this chromosome IV QTL locus ( 15 . 47–15 . 91 Mb ) were more resistant to incubation in 12 . 5 μM albendazole than those strains with the CB4856 genotype , with CB4856 animals exhibiting a significantly greater decrease in length ( Fig 1C , top ) . To narrow this interval to a smaller region , we measured the albendazole response phenotypes of near-isogenic lines ( NILs ) with QTL genomic regions derived from either the N2 or CB4856 strains introgressed into the opposite genetic background ( Fig 2A ) . NILs with the N2 genotype spanning the 15 . 57–15 . 65 Mb interval of chromosome IV exhibited greater resistance to albendazole compared to the parental CB4856 strain and , conversely , a NIL with the CB4856 genotype in this region was more sensitive to albendazole when compared to the parental N2 strain ( Fig 2A , bottom ) . These results are consistent with the QTL direction of effect , and we expect that the variant ( s ) of largest effect fall within this narrowed interval . Significant differences among NILs within the resistance and sensitive groupings suggest that additional variants and epistatic interactions within the complete QTL interval influence albendazole susceptibility . The NIL interval was annotated with known variants distinguishing N2 and CB4856 and their estimated functional consequences ( Fig 2B , S2 and S3 Tables ) . The presence of large numbers of annotated piRNAs added to the complexity of analyzing genome variation in both the complete QTL interval and the NIL-narrowed interval . We considered protein-coding and non-coding RNA ( ncRNA ) variation as potentially causal to the albendazole-resistance phenotype . Only three proteins in the narrowed interval are predicted to have altered functions as a result of single nucleotide variants ( SNVs ) . However , the plausibility of these specific candidate genes was dampened by a number of factors . Y105C5A . 508 is curated as a short and likely pseudogenic transcript , which shares no homology with proteins in species with available sequence data . Additionally , expression of this transcript was barely detectable via RNA-seq in either the N2 or CB4856 strains ( S4 Table ) . The gene pqn-79 has a predicted coding variant ( Thr194Ala ) but belongs to a highly redundant gene family that has > 99 . 5% nucleotide sequence identity with at least three other homologs . None of these homologs exhibit coding variation or differences in expression across strains ( S4 Table ) , but the possibility of a dominant-negative or gene dosage effect cannot be excluded . The gene Y105C5A . 8 codes for a protein of unknown function that is predicted to contain a splice-donor variant , but we found no differences in overall expression ( S4 Table ) or splice form abundance ( S5 Fig ) between N2 and CB4856 for this gene . We therefore originally hypothesized that albendazole resistance is more likely a function of variation in the non-coding RNA ( ncRNA ) complement , specifically the piRNA-encoding genes . 1 , 684 SNVs occur in known piRNAs in the complete QTL interval , and 276 of these fall within the NIL-narrowed interval ( S3 Table ) . To test this hypothesis , we generated NILs encompassing the broader piRNA-enriched region on chromosome IV ( 13 . 5–17 . 2 Mbs ) . These NIL strains robustly recapitulated the QTL direction of effect ( S6A Fig ) . Next , we analyzed the responses of small RNA pathway mutants to albendazole in an effort to perturb the large number of diverse piRNAs . 21U-RNAs/piRNAs are regulated by the Piwi Argonaute PRG-1 [50] . We hypothesized that albendazole sensitivity should be dependent on prg-1 function . We tested mutants in all three Argonaute genes that encode proteins that interact with primary small RNAs immediately upstream of WAGO-associated 22G RNA generation ( ergo-1 , alg-4; alg-3 , and prg-1 ) [51] . ERGO-1 and ALG-3/4 engage 26G RNAs , and PRG-1 coordinates the processing of piRNAs . Of the mutations in these genes , a back-crossed prg-1 N2 strain conferred greater albendazole sensitivity compared to N2 ( S6B Fig ) . Concerned with the relative sickness of this mutant strain in control ( DMSO ) conditions , we generated and tested two independent prg-1 loss-of-function alleles in both the N2 and CB4865 backgrounds using CRISPR/Cas9 genome editing . These new prg-1 knock-out strains were normal in DMSO and not significantly different in albendazole from the wild-type parents in the expected direction of effect across three replicate assays ( S7 Fig ) . Although we have not identified the causal genetic locus , it is possible that the factors underlying the albendazole resistance phenotype are a mixture of coding or regulatory variants that interact and affect gene function . Numerous expression QTL ( eQTL ) have also previously been mapped to the major-effect albendazole QTL region , potentially connecting candidate variants to local ( 3 cis eQTL ) or distant ( 78 trans eQTL ) regulatory targets that affect drug sensitivity [52] ( S5 Table ) . To compare benzimidazole-resistance loci across nematode species , we examined variation in the responses of Caenorhabditis briggsae strains to the same set of benzimidazole compounds that we studied in C . elegans . Dose responses and heritability calculations ( S8 and S9 Figs ) were used to select concentrations for linkage mapping experiments . Single concentrations of albendazole ( 25 μM ) , fenbendazole ( 30 μM ) , mebendazole ( 50 μM ) , and thiabendazole ( 40 μM ) were chosen . Linkage mapping was carried out with a collection of 153 recombinant inbred lines ( RILs ) created using the parental strains AF16 and HK104 [23] , which led to the discovery of four QTL for the tested drugs ( Fig 3A , S10 Fig and S6 Table ) . The C . briggsae QTL of largest effect is found on the left arm of chromosome IV ( 2 . 56–3 . 45 Mb ) and explains approximately 18% of fenbendazole-induced variation in fecundity between the parental strains ( Fig 3B ) . QTL of smaller effect were identified on the center of chromosomes V for thiabendazole and on the left arms of chromosomes V and X for albendazole . No QTL were identified for mebendazole across the examined set of fitness traits . Strikingly , the major fenbendazole QTL falls within the primary C . briggsae piRNA cluster ( Chr IV: 0–6 . 9 Mb ) . Despite tens of millions of years of evolutionary distance [53] , the most significant C . elegans and C . briggsae benzimdazole QTL were found to occur in genomic regions that are syntenic between species [54] ( Fig 4A ) . The C . briggsae fenbendazole QTL region contains 84 protein-coding genes with variants ( S7 Table ) . However , no orthologs were identified across species that both contained variants and fell within these specific QTL confidence intervals . This result suggests that different gene ( s ) and mechanisms account for the drug effects mapped to these syntenic loci across species . piRNA-encoding genes that densely cover this locus could potentially underlie the fenbendazole-resistance phenotype in C . briggsae , but no piRNAs are shared between C . elegans and C . briggsae . To explore the potential translation of these loci to parasites , we looked at putative conservation of candidate C . elegans resistance genes ( s ) in parasitic nematodes . Specifically , we examined conservation of protein-coding genes that fall within benzimidazole QTL confidence intervals and contain predicted functional variants . Substantial fractions of these candidate genes have identifiable preliminary orthologs in representative parasites from other clades , although conservation varies across individual QTL ( S8 Table ) . Approximately 40% of the 5 , 434 candidate resistance genes across all identified QTL have orthologs in the clade III human filarial parasite Brugia malayi and the clade IV model gastro-intestinal parasite Strongyloides ratti , and 31% are conserved across all three species ( Fig 4B and S8 Table ) . It is reasonably likely that gene ( s ) validated in this system as modulators of benzimidazole response have counterparts in parasite genomes . The narrowing of these QTL in Caenorhabditis species and the validation of the independent effects of genes on phenotypes has the potential to discover novel benzimidazole modes of action and resistance . It will be necessary to carry out more comprehensive parasite ortholog identification and analyses once QTL have been narrowed to smaller numbers of candidate genes . A number of considerations can potentially hamper the translation of these data to parasitic nematodes . The linkage mapping approach is limited to sampling genetic variation between two parental strains . Genome-wide association ( GWA ) studies that explore species-wide variation in anthelmintic response would improve the potential scope of variant discovery and the prospects of translating results to parasite species . This study is focused on standing genetic variation and therefore does not speak to de novo mutations that might be selected through drug pressure in field populations of parasites . Many of the QTL identified span large confidence intervals and explain little variation , making them difficult to systematically narrow . Although it is reasonable to expect conservation of many underlying mechanisms that account for drug sensitivity , it is difficult to precisely map phenotypes and effect sizes between free-living and parasitic nematodes . Despite these caveats , we report a number of QTL associated with smaller confidence intervals and more significant effects . The systematic narrowing of these QTL is likely to lead to the discovery of causal genes with an appreciable likelihood of conservation in parasitic nematodes . Ultimately , anthelmintic resistance gene variants validated in the genetically tractable Caenorhabditis species can be evaluated with respect to effect size and mutation type ( e . g . , gain or loss-of-function ) . Although efforts to narrow QTL may lead to the discovery of variants with modest drug effects , it is conceivable that small effects can additively build to significant resistance in parasites and that other potential variation at the same loci can produce much larger effects . For genes with parasite orthologs , the effects of gene loss-of-function on parasite drug response can be assayed using targeted genetic perturbation techniques ( e . g . , RNA interference ) established in various helminth species . The expansion of the helminth genetic toolkit to CRISPR/Cas9 genome editing [56 , 57] will pave the way for more precise mappings of mutations and effects . These Caenorhabditis data complement efforts to improve the resolution of parasite population genomics data [58] , as well as efforts to carry out genetics studies in helminth species where feasible [59 , 60] . More broadly , we expect that these approaches and data can hasten better models and markers for the development of anthelmintic resistance in human and animal parasite control . | The treatment of roundworm ( nematode ) infections in both humans and animals relies on a small number of anti-parasitic drugs . Resistance to these drugs has appeared in veterinary parasite populations and is a growing concern in human medicine . A better understanding of the genetic basis for parasite drug resistance can be used to help maintain the effectiveness of anti-parasitic drugs and to slow or to prevent the spread of drug resistance in parasite populations . This goal is hampered by the experimental intractability of nematode parasites . Here , we use non-parasitic model nematodes to systematically explore responses to the critical benzimidazole class of anti-parasitic compounds . Using a quantitative genetics approach , we discovered unique genomic intervals that control drug effects , and we identified differences in the effects of these intervals across compounds and doses . We were able to narrow a major-effect genomic region associated with albendazole resistance and to establish that candidate genes discovered in our genetic mappings are largely conserved in important human and animal parasites . This work provides new leads for understanding parasite drug resistance and contributes a powerful template that can be extended to other anti-parasitic drug classes . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"and",
"discussion"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"chemical",
"compounds",
"quantitative",
"trait",
"loci",
"caenorhabditis",
"animals",
"organic",
"compounds",
"parasitology",
"animal",
"models",
"caenorhabditis",
"elegans",
"model",
"organisms",
"experimental",
... | 2018 | Discovery of genomic intervals that underlie nematode responses to benzimidazoles |
Plants respond to different stresses by inducing or repressing transcription of partially overlapping sets of genes . In Arabidopsis , the PHR1 transcription factor ( TF ) has an important role in the control of phosphate ( Pi ) starvation stress responses . Using transcriptomic analysis of Pi starvation in phr1 , and phr1 phr1-like ( phl1 ) mutants and in wild type plants , we show that PHR1 in conjunction with PHL1 controls most transcriptional activation and repression responses to phosphate starvation , regardless of the Pi starvation specificity of these responses . Induced genes are enriched in PHR1 binding sequences ( P1BS ) in their promoters , whereas repressed genes do not show such enrichment , suggesting that PHR1 ( -like ) control of transcriptional repression responses is indirect . In agreement with this , transcriptomic analysis of a transgenic plant expressing PHR1 fused to the hormone ligand domain of the glucocorticoid receptor showed that PHR1 direct targets ( i . e . , displaying altered expression after GR:PHR1 activation by dexamethasone in the presence of cycloheximide ) corresponded largely to Pi starvation-induced genes that are highly enriched in P1BS . A minimal promoter containing a multimerised P1BS recapitulates Pi starvation-specific responsiveness . Likewise , mutation of P1BS in the promoter of two Pi starvation-responsive genes impaired their responsiveness to Pi starvation , but not to other stress types . Phylogenetic footprinting confirmed the importance of P1BS and PHR1 in Pi starvation responsiveness and indicated that P1BS acts in concert with other cis motifs . All together , our data show that PHR1 and PHL1 are partially redundant TF acting as central integrators of Pi starvation responses , both specific and generic . In addition , they indicate that transcriptional repression responses are an integral part of adaptive responses to stress .
Plants have evolved adaptive responses to cope with growth under a variety of stress conditions . These responses involve changes that are specific to particular types of stress or shared by different stress types . A specific response to phosphate ( Pi ) starvation , for example , is increased Pi uptake capacity from the soil , whereas the induction of anthocyanin accumulation and acceleration of senescence are shared responses to many different kinds of stress [1]–[3] . In line with the overlap among physiological and developmental responses to different stress types , transcriptional responses overlap as well [4] , [5] . An important question regarding transcriptional responses to stress is how specific and shared responses are regulated - are they controlled by the same regulatory systems or are there generic stress response regulators ? A second question is the biological significance of transcriptional repression in stress responses . Is it mostly an integral part of the adaptive system or is it mainly an indirect consequence of plant malfunction due to stress ? We addressed these two questions , focussing on the Pi starvation stress response as a model in Arabidopsis . The importance of transcriptional control in the regulation of Pi starvation responses has already been established . The expression of a large number of genes is altered in response to Pi starvation ( between 900 and 3000 , depending on the study ) [6]–[11] . The transcription factor ( TF ) PHR1 is a key regulatory component of Pi starvation responses in Arabidopsis [12]; PHR1 binds to an imperfect palindromic motif present in the promoters of many Pi starvation-induced genes . Loss of function mutation of PHR1 affects several Pi starvation responses , including alteration of root to shoot growth ratio , anthocyanin accumulation , and the expression of several Pi starvation-induced genes . Nonetheless , the extent of the role of PHR1 in Pi starvation responses has yet to be established . PHR1 is part of a family of 15 genes in Arabidopsis ( MYB-CC family ) . Some functional redundancy among family members has been suggested , based on the fact that the phr1 effect on some Pi starvation-responsive genes is only partial [12] . In addition to PHR1 , members of bHLH , WRKY , Zinc finger and R2R3 MYB families of TF are involved in the control of Pi starvation responses , although their exact positions in the signalling pathway have not been established [13]–[17] . Whereas PHR1 is weakly transcriptionally responsive to Pi starvation , these other TF genes are highly responsive to Pi stress , suggesting that they act downstream of PHR1 . Additional mechanisms other than TF operate to regulate Pi starvation signalling . These include sumoylation [18] , degradation by the proteosome , which probably involves the E2 ubiquitin conjugase-related enzyme ( PHO2 ) [19]–[21] , and control of Pi uptake efficiency via a phosphate transporter traffic facilitator ( PHF1 ) [22] , as well as several miRNA and antagonists ( IPS1 and related genes ) of miRNA MiR399 , which controls PHO2 activity [20] , [23]–[27] . There is also a Pi starvation-induced gene family that encodes nuclear SPX domain-containing proteins , which affects responsiveness of several Pi starvation-induced genes through an unknown mechanism [28] , [29] . MiR399 , IPS1 and PHF1 are all under the control of PHR1 , which itself is sumoylated in vitro by SIZ1 , further strengthening the central role of PHR1 in the control of Pi starvation responses [12] , [18] , [20] , [22] . Here we performed a physiological and transcriptomic analysis of Pi starvation responses in plants with altered PHR1 ( -like ) activity , comparing mutants of phr1 , phr1-like1 ( phl1 ) and phr1 phl1 , and PHR1-overexpressing transgenic lines . Results showed that PHR1 and PHL1 are partially redundant and have a central role in the control of physiological and molecular responses to Pi starvation , independent of whether these responses are specific to Pi starvation stress . They also indicate that a large proportion of the transcriptional repression responses to Pi starvation are part of the adaptive response to this stress , and that their control by PHR1 ( -like ) is indirect . We also show the importance of the PHR1 binding sequence ( P1BS ) as an integrating cis-regulatory motif associated with genes that are highly induced by Pi starvation .
PHR1 mutants show distinct degrees of impairment of different Pi starvation responses , as evident in expression analyses of a set of Pi starvation-responsive genes [12] . Incomplete impairment of these responses could reflect partial gene redundancy , as PHR1 belongs to a transcription factor family with 15 close members ( Figure 1A and Figure S1 ) . It is also possible that more than a single regulatory system controls Pi starvation responses . To study the relationships between these possibilities , we searched for T-DNA mutations at PHR1-related genes in public databases; the two phylogenetically most closely related Arabidopsis genes for which a mutant was available were At5g29000 and At5g06800 . We selected At5g29000 , which we term PHR1-LIKE1 ( PHL1 ) for further analysis , as it displayed a higher degree of amino acid identity with PHR1 , and the T-DNA mutation disrupted the coding region of PHL1 mRNA ( Figure S1 and Figure S2A ) . We examined whether expression of PHR1 and PHL1 overlapped . Northern analysis showed that PHL1 expression overlapped with that of PHR1 in both shoots and roots under any Pi regime ( Figure S2B ) . This observation was confirmed by the large overlap in PHR1 and PHL1 expression at different developmental stages , according to GENEVESTIGATOR gene expression data ( https://www . genevestigator . com ) [30] ( Figure S2C ) . After generating a homozygous double mutant phr1 phl1 , functional redundancy between PHR1 and PHL1 was shown by northern analysis ( Figure 1B ) . Whereas the effect of the phl1 mutation on Pi starvation responsiveness was barely detectable , we observed a synergistic effect of phr1 and phl1 mutations for expression of all genes examined . To be noted is the limited effect of these mutations on expression of Pi starvation induced genes in plants grown under a high Pi regimen , as shown by northern analysis and also using quantitative reverse transcription PCR ( Q-RT-PCR ) ( Figure 1B and Figure S3 ) . For comparative purposes , we produced transgenic plants with the phr1 background , overexpressing PHR1 fused to the rat glucocorticoid receptor domain ( GR:PHR1 ) to allow dexamethasone ( DEX ) -inducible control of its activity [31] . Northern analysis showed that three independent lines overexpressing the GR:PHR1 fusion had DEX-dependent PHR1 activity ( Figure S4 ) . The effect of GR:PHR1 overexpression on gene expression was detected even when plants were grown under Pi sufficient conditions ( Figure S4 ) . These results are in agreement with previous reports [32] , [33] , and indicate that PHR1 overexpression can override , at least to some extent , the negative regulatory control that occurs at much more limited PHR1 levels in wild type plants . Physiological tests were performed on wild type , phr1 , phl1 and phr1 phl1 mutants , and transgenic plants overexpressing GR:PHR1 ( Figure 2 ) . In accordance with previous results [12] , Pi accumulation in plants grown under Pi sufficient conditions was reduced in the phr1 mutant ( compared to wild type plants; Figure 2A ) . The phl1 mutant had slightly , but significantly reduced Pi levels , and a further decrease in Pi accumulation was observed in the phr1 phl1 double mutant , indicating partial functional redundancy between these two MYB-CC family genes . Conversely , Pi accumulation in DEX-treated GR:PHR1-overexpressing plants ( OxGR:PHR1 ) was greatly increased with respect to that in wild type plants ( Figure 2A ) . After Pi starvation , anthocyanins accumulate in leaves and stems of wild type plants; much less anthocyanin accumulated in the phr1 mutant ( Figure 2A ) . The effect on anthocyanin accumulation was negligible for phl1 , however , and did not differ significantly between phr1 and the phr1 phl1 double mutant ( Figure 2A ) . In contrast , anthocyanin accumulation was enhanced in OxGR:PHR1 plants . Following Pi starvation , wild type plants show an increase in root to shoot growth ratio; this increase was significantly reduced in phr1 mutants , whereas the phl1 mutation had a negligible effect alone or in combination with phr1 . In DEX-treated phr1 GR:PHR1-overexpressing plants , the root to shoot growth ratio was similar to that of wild type plants in Pi starvation conditions ( Figure 2A ) . The effect of phl1 , and in particular , of phr1 and phr1 phl1 mutations on senescence and silique formation was evident on plants grown in Pi starvation conditions , as these plants died before flowering ( Figure 2A and 2B ) . In contrast , DEX-treated OxGR:PHR1 plants showed slightly accelerated flowering and higher silique production . These findings concur with the idea that cell death in the mutants reflects a lack of correct protection against the stress inherent in Pi starvation , and that increased PHR1 activity results in increased reproductive success in these stress conditions . The effect on root hair length was quite obvious when plants where grown in Pi-lacking medium in vertical plates; the phr1 mutation affected root hair length , which was enhanced when combined with the phl1 mutation ( Figure 2C ) . Given the partial functional redundancy between PHR1 and PHL1 , as shown by our analyses of phr1 and phl1 single and double mutants , we examined whether these two proteins had similar DNA binding properties and whether they were able to heterodimerise . For these studies , we used two N-terminally truncated versions of each protein obtained by in vitro translation , since a previous study with PHR1 showed that in vitro-translated N-terminally truncated PHR1 protein had similar DNA binding specificity but higher affinity than the full length protein [12] . The two deletions removed 99 or 198 N-terminal amino acids of PHR1 and 103 or 210 residues of PHL1 ( Figure 3A ) . Electrophoretic mobility shift assays ( EMSA ) indicated that both PHL1 versions could interact with P1BS sequences , the prototype PHR1 binding site ( Figure 3B ) . EMSA with the two cotranslated truncated PHL1 versions showed the appearance of a band of intermediate mobility , in addition to those corresponding to the medium and short versions of truncated PHL1; this was indicative of the self-dimerisation properties of PHL1 [34] , as also observed for PHR1 [12] . Intermediate mobility bands were also observed when the medium size PHL1 version was cotranslated with the short PHR1 version , indicating that they can form heterodimers ( Figure 3C ) . The ability of PHL1 and PHR1 to heterodimerise was confirmed by identification of PHL1 as a PHR1-interacting protein in a yeast two-hybrid assay ( Figure S5 ) . To examine the correspondence between the effects of phr1 and combined phr1 phl1 mutations on physiological responses and transcriptional responses to Pi starvation , we performed transcriptomic analysis in wild type , as well as single phr1 and double phr1 phl1 mutants . For these assays , wild type plants were germinated and grown for 7 days in Pi-sufficient and -starvation conditions , and mutant plants were grown in Pi starvation conditions . The use of long-term stress treatment for the analysis was aimed to identify the long-term effects of these mutations . For transcriptomic analyses , we collected RNA separately from shoots and roots in three independent replicates obtained over a two-month interval . A total of 1873 and 704 genes were upregulated , and 1795 and 326 downregulated in Pi-starved shoots and roots , respectively ( cut-off values 2-fold , false discovery rate ( FDR ) <0 . 05; Table 1 and Table S1 ) . The effect of the phr1 mutation , particularly when combined with phl1 , on the expression of these Pi starvation-responsive genes was striking ( Table 1 ) . Of the genes whose expression was induced two-fold or more in the wild type plants in response to Pi starvation , 68% and 47% showed at least two-fold reduced expression in the shoots and roots , respectively , of the Pi-starved phr1 phl1 double mutant compared to wild type in the same conditions . In contrast , only 1 . 4% and 2% of Pi starvation-induced genes in shoots and roots showed increased expression in the Pi-starved double mutant . These numbers are even more extreme if only genes induced four-fold or more in wild type plants are analysed , or if the cut-off values in the mutant versus wild type comparison are relaxed ( 1 . 5-fold , FDR<0 . 1 ) ( Table 1 ) . For example , >80% and 60% of the Pi starvation-induced genes in shoots and roots , respectively , show reduced expression in the Pi-starved phr1 phl1 double mutant using cut-off values of 1 . 5-fold , FDR<0 . 1 . The situation is similar for the repressed genes , as 70% and 46% of genes repressed in shoot and roots , respectively , of wild type plants grown in Pi starvation conditions showed higher expression in the Pi-starved double mutant . These data underline the central regulatory role of PHR1 ( -like ) genes in the transcriptional control of Pi starvation responses . Reciprocally , phr1 and phl1 mutations mostly affect expression of Pi responsive genes ( Figure S6 ) . For example , Pi starved shoots of the double mutant display reduced expression relative to wild type levels of almost 90% of highly ( >4× ) Pi starvation induced genes , while this proportion falls below 2% for non Pi starvation responsive ( or Pi starvation repressed ) genes ( Figure S6 ) . To measure the extent of functional redundancy between PHR1 and PHL1 , we examined the Pi starvation-responsive genes whose expression was greatly altered in the double mutant compared to phr1 . Only a small proportion of Pi starvation-responsive genes showed more than two-fold altered expression in the phr1 phl1 double mutant compared to phr1 ( 200 Pi starvation-induced and 82 Pi starvation-repressed genes displayed more than two-fold reduced and increased expression , respectively , in phr1 phl1 vs . phr1; Table S2 ) . Of the genes showing altered expression in phr1 phl1 compared to phr1 , only 30% did not show altered expression in Pi-starved single phr1 mutant vs . Pi-starved wild type plants , indicating a large functional overlap between phr1 and phl1 . We used MAPMAN ontology tools to obtain an overview of Pi starvation-responsive genes involved in metabolism and regulation ( http://mapman . gabipd . org/web/guest/home ) [35] ( Figure S7 ) . Pi starvation had a broad effect on genes involved in all aspects of metabolism . In particular , induced genes were greatly enriched in secondary metabolism biosynthetic genes , especially those of phenylpropanoids ( Figure S7A ) . There was also an increase in biosynthetic genes of sulpholipids and galactolipids , which replace phospholipids under Pi-limiting conditions as previously reported [36]–[38] , and of tetrapyrroles . Pi starvation also had a large effect on transcriptional repression of genes involved in light reactions of photosynthesis and in photorespiration ( Figure S7A ) . These are likely protective responses as they would reduce the potential generation of reactive oxygen species . Pi starvation-triggered changes in the transcription of regulatory components showed notable effects on genes encoding transcription factors , components of protein degradation machinery , hormone biosynthesis and signalling , and calcium-based regulation , with induction more prominent than repression ( Figure S7B ) . Our findings are qualitatively similar to those in previous reports [9] , [10] . We compared our set of Pi starvation-induced and -repressed genes with the sets of genes responsive to different types of stress or hormones available at the GENEVESTIGATOR database [30] ( Table S3 ) . In most cases , there were significant overlaps between the set of genes responsive to Pi starvation and those responsive to other types of stress and , as reported , there were also many significant overlaps with hormone-responsive gene sets [39] . To infer whether control of shared genes , i . e . , responsive to Pi starvation and other stresses , could occur through independent regulatory systems ( involving different stress type-specific TF ) or , alternatively , could in part use common regulatory components , we examined the representation of TF genes in the sets of shared genes that respond to Pi starvation and to other stress types . In most cases , we found that TF genes were equally over-represented relative to non-TF genes in the sets of shared genes ( Table S3 ) ; this favours the idea that transcriptional control of genes that respond to two stress types in part uses common regulatory components . Two exceptions corresponded to hydrogen peroxide treatment and low nitrate growth conditions , whose induced genes are significantly enriched in Pi starvation-induced genes; enrichment was much weaker or non-existent for Pi starvation-induced TF , however , raising the possibility that in these cases , part of the shared response is controlled by independent TF . Finally , we studied the effect of Pi starvation on general stress response ( GSR ) genes . Two independent studies recently identified sets of general stress-induced genes [4] , [5] . There is considerable overrepresentation of these genes in our set of Pi starvation genes induced two-fold or more ( >44% vs . a predicted 9% ) ( Table 2 ) . A large proportion of these general stress-induced genes responsive to Pi starvation show reduced expression in the Pi-starved phr1 phl1 double mutant ( ∼70% ) , indicating that general stress responses associated to Pi starvation are controlled by PHR1 ( -like ) TF . To examine direct targets of PHR1 , we followed the strategy originally described by Galaktionov and Sablowski [40] , [41] , which is based on the use of a transgenic phr1 mutant plants expressing the GR:PHR1 fusion ( OxGR:PHR1 phr1 ) , whose activity is postranslationally controlled by DEX . Gene expression analysis following DEX-mediated PHR1 activation and the concomitant inhibition of translation with cycloheximide ( CHX ) , which prevents PHR1 effects on the expression of secondary targets , will inform on PHR1 direct targets . For this study , OxGR:PHR1 phr1 and phr1 plants were grown in complete ( +Pi ) liquid medium for 7 days , then transferred for 2 days to phosphate-lacking ( −Pi ) medium . Plants were then supplemented with 5 µM DEX and 10 µM CHX , and incubated for 6 h before harvest . Total RNA was isolated from 3 independent samples of OxGR:PHR1 phr1 and phr1 plants and transcriptomic analysis was performed . Using standard cut-off values ( two-fold , FDR<0 . 05 ) , 319 and 21 genes showed increased or decreased expression in OxGR:PHR1 phr1 vs . phr1 mutant plants , respectively . A considerable overlap was found between the set of genes with increased expression in CHX-treated OxGR:PHR1 phr1 plants with the set of Pi starvation-induced genes ( 210 out of 319 ) , whereas there was almost no overlap between genes with reduced expression in CHX-treated OxGR:PHR1 phr1 plants and Pi starvation-repressed genes ( 1 gene; Table 3 and Table S4 ) . This finding indicates that PHR1 is a bona fide transcriptional activator and that PHR1 control of Pi starvation-repressed genes is indirect . To substantiate the conclusion that PHR1 control of Pi starvation-repressed genes is indirect , we tested for P1BS in different parts of the Pi starvation-responsive genes and in PHR1 direct targets . Direct targets were enriched in P1BS sequences in all parts of the gene compared to average Arabidopsis genes . As a result , only 3% of PHR1 direct targets did not have a P1BS site in the region encompassing 3 kb of the promoter region to 3 kb downstream , compared to 17% for average Arabidopsis genes . Enrichment was especially high in the 1 kb proximal promoter region and even higher in the 5′UTR . Although the 3′UTR of direct targets was only weakly enriched in P1BS sequences , P1BS was significantly enriched in the 3′UTR of the whole set of Pi starvation-induced genes ( Figure 4 and Table S5 ) . We next tested whether specificity of Pi starvation inducibility correlated with P1BS content in the promoter . We examined the average number of other stresses in which Pi starvation-induced genes are also induced relative to the presence of none , one , or more than one P1BS in the 1 kb proximal promoter region , 5′UTR , 3′UTR , introns or 1 kb proximal downstream region , or in any combination of these , in which the set of Pi starvation-induced genes and/or PHR1 direct targets showed a significantly higher P1BS levels compared to average Arabidopsis genes ( Figure 4A and Table S5 ) . P1BS content in the proximal promoter region , the 5′UTR or introns was associated with slightly higher specificity of Pi starvation-responsiveness; however , the difference in specificity was insufficient to ascribe specificity to the class of genes containing P1BS ( Figure 4B and Figure S8A ) . We examined whether genes with P1BS in their promoters , 5′UTR , 3′UTR , introns and the 1 kb downstream region were induced at a higher level by Pi starvation . Analysis of P1BS representation relative to the x-fold induction showed a striking correlation between inducibility and P1BS content in the 1 kb proximal promoter region , whereas P1BS content in other gene regions showed no correlation with inducibility ( Figure 4C and Figure S8B ) . To confirm the importance of P1BS as key cis-regulatory motifs in Pi starvation signalling , we performed two types of experiments: i ) evaluation of the effect of P1BS mutation on Pi starvation-responsive genes and ii ) analysis of Pi starvation responsiveness mediated by a minimal promoter containing multimerised P1BS . For the first experiment , we selected the promoters of two genes , IPS1 , a highly specific Pi starvation-induced gene [42] , and RNS1 , which is also responsive to wounding stress [43] . A 1 kb DNA fragment containing the promoter proximal region up to the first initiation codon in the transcribed region was obtained for each gene by PCR amplification of genomic DNA . We also prepared mutants in which the P1BS sites of each gene were impaired . For the IPS1 promoter , which has two P1BS , we obtained single mutants of each P1BS and a double mutant of both . Transgenic plants harbouring these promoters or their mutant versions fused to the coding region of GUS in the pBI101 binary vector [44] were obtained ( Figure 5A ) . In the case of wild type IPS1 constructs ( IPS1:GUS ) , 10 of 10 transgenic plants examined showed Pi starvation-induced GUS activity . Mutation of P1BS-2 had no effect on Pi starvation responsiveness ( 9 of 10 transgenic plants showed Pi starvation-induced GUS activity ) , whereas mutation of P1BS-1 abolished Pi starvation responsiveness ( 10 of 10 plants had no GUS activity; see example in Figure 5B ) . For wild type RNS1 , 9 of 10 transgenic plants displayed Pi starvation-induced GUS activity , whereas P1BS impairment resulted in no Pi starvation-induced GUS activity ( Figure 5C ) . In the case of RNS1 , we also examined responsiveness to wounding . Both the wild type RNS1 promoter and the mutant promoter impaired in P1BS showed similar wounding-induced GUS activity ( Figure 5C ) . We examined whether phr1 and/or phl1 mutations affected RNS1 expression . Northern analysis indicated that mutation of PHR1 and PHL1 , while impairing Pi starvation responsiveness , had no effect on the RNS1 wounding response ( Figure 5D ) . These results point to a critical role for P1BS in Pi starvation responsiveness and , in the context of non-specific Pi starvation-responsive genes , indicate that PHR1 ( -like ) and P1BS are not necessarily required for responsiveness to stresses other than Pi starvation . In addition , it is evident from the case of IPS1 that not all P1BS motifs in a promoter are equally relevant for Pi starvation responsiveness . Other architectural determinants such as nucleosome positioning and P1BS organisation with respect to additional cis motifs might determine P1BS function . To analyse the capacity of P1BS to mediate Pi starvation responsiveness , we fused four tandem copies of P1BS to the −46 minimal 35S promoter from CaMV ( 4xP1BS:GUS ) [45] . Transgenic plants harbouring this construct were fully responsive to Pi starvation ( 9 of 10 independent lines ) . We chose one of these lines to study the specificity of Pi starvation responsiveness and the effect of known agonists ( sucrose ) [46]–[48] or antagonists ( cytokinins and arsenate ) [39] , [46] , [49] . As in the case of IPS1:GUS , the 4xP1BS:GUS construct was highly responsive to Pi starvation , but not to other types of stress ( nitrogen , potassium and sulphur starvation , and salt and osmotic stress ) ; in addition , it was responsive to the stimulatory effects of sucrose and the repressing effect of arsenate and cytokinins ( Figure 6A and 6B ) . Systemic repression is a characteristic type of control in nutrient physiology; it stands for the fact that most responses to nutritional deficiency are determined by shoot nutritional status rather than by the local nutrient concentration in the vicinity of the root system [50] . To evaluate whether systemic repression is signalled through P1BS , we used a split root assay in which part of the root system of Pi-starved plants was placed in Pi-lacking medium and the other part in Pi-rich medium . GUS activity was not detected in the Pi-lacking parts of the roots in the split root assay ( Figure 6C ) . These results define P1BS and , consequently , PHR1 ( -like ) TF as central integrators in Pi starvation signalling ( Figure 6D ) . To examine whether P1BS sequences are sufficient in the context of a natural promoter to mediate Pi starvation responsiveness , we performed phylogenetic footprinting analysis to search for conserved cis-regulatory regions that could be relevant in the control of gene expression . For this analysis , we examined the promoter of the highly specific Pi starvation- responsive IPS1 gene . Using oligo-adapted PCR amplification with a conserved region of IPS1 , we amplified fragments containing the promoter region of orthologous genes from four different Brassicaceae species ( Figure S9 ) . Sequence alignment showed two highly conserved regions , spanning from nt −626 to −527 and from nt −280 to −109 from the first ATG in the transcribed region of IPS1 ( Figure S9 ) . As this analysis did not provide sufficient resolution to identify cis-regulatory motifs , we included At4 in the alignment , as it is also responsive to Pi starvation and is the most closely related IPS1 homologue in Arabidopsis [25]; we thus delimited the candidates for cis-regulatory sequences to six short motifs ( motifs A to E and P1BS1; Figure 7 and Figure S9 ) . Further inspection of additional members of the family in Arabidopsis and other species such as tomato , medicago , maize and poplar showed that two of these six conserved motifs were also conserved outside the Brassicaceae family ( P1BS1 and B motifs; Figure 7A ) . Fusion of the region encompassing motifs A-P1BS-B to the −46 minimal 35S promoter ( A-P1BS-B:GUS ) showed that this region is sufficient to mediate Pi starvation responsiveness of a GUS reporter gene ( Figure 7B and 7C ) . Mutational analyses indicated that whereas impairment of motif A had no effect , mutation of motif B abolished Pi starvation responsiveness and resulted in weak constitutive expression of the mutant gene ( Figure 7B and 7C ) . These results indicate that motif B acts in concert with P1BS to mediate Pi starvation responsiveness . We also analysed whether motif B could drive Pi starvation inducibility , as is the case of P1BS . An artificial gene containing four tandem copies of motif B fused to the -46 minimal 35S promoter from CaMV and the GUS coding region ( 4xB:GUS ) did not show any GUS activity , even in Pi starvation conditions ( 10 of 10 independent transgenic lines; Figure 7B and 7C ) . To evaluate a possible informatic approach to predict the relevance of motif B and not of other conserved motifs in the context of Pi starvation responsiveness , we examined whether the conserved motifs A and B were overrepresented in the promoters of our set of Pi starvation-responsive genes . Neither of these conserved motifs were significantly overrepresented; nonetheless , we found a clear overrepresentation of motif B in combination with P1BS when the distance between the two motifs was restricted to 25 nt ( 26 observed vs . 12 . 4 predicted , p<0 . 0002; Table S6 ) . These data , including the result using the artificial 4xB:GUS gene , strongly suggest that the role of motif B in Pi starvation responsiveness is subsumed to that of P1BS , and that it is likely that PHR1 ( -like ) proteins interact with a yet to be identified TF that interacts with motif B , directly or via a co-adaptor protein .
The partial functional redundancy between PHR1 and PHL1 is indicated by the additive or synergistic effects of mutations in these two genes on most of the traits examined , including transcriptional responsiveness to Pi starvation . Redundancy probably involves additional members of the MYC-CC family , since mutation of both PHR1 and PHL1 does not fully abolish Pi starvation responses . For instance , IPS1 is still weakly responsive in the phr1 phl1 double mutant , and we demonstrate that mutation of a P1BS site in IPS1 completely abolishes the Pi starvation response of this promoter ( Figure 1B and Figure 5B ) . In line with this partial functional redundancy , PHR1 and PHL1 have similar DNA binding specificity and can heterodimerise . The use of plants with different PHR1 ( -like ) activity levels ( phr1 and phl1 single mutants , phr1 phl1 double mutant , and PHR1-overexpressing plants ) confirmed the essential role of PHR1 and PHL1 in the control of intracellular Pi concentrations and anthocyanin accumulation [12] , as well as in other aspects of the response , such as root hair length , silique formation and senescence ( Figure 2 ) . The observed effect of phr1 and phl1 mutations on Pi levels of plants grown under a Pi rich regimen contrasts with the limited effect of these mutations on expression of Pi starvation induced genes in plants grown under these conditions . This could reflect a partial compensation of a lower amount of PHR1 ( -like ) protein in these mutants with a higher activity of the remaining PHR1 ( -like ) protein ( likely encoded by PHR1-related genes ) , as the level of Pi , which inhibits PHR1 ( like ) , in mutants grown in +Pi medium is lower than in the wild type . The observed effect of the phr1 mutation on root hair formation supports a previous finding in rice , in which overexpression of a rice PHR1 homologue was shown to affect root hair length and density [33] . Given that root hair response is dependent on local Pi concentration in the root surroundings rather that on shoot Pi concentration [51] , our data indicate that PHR1 also controls at least part of local Pi-dependent responses . The results also show the importance of a proper response to nutrient stress for reproductive success , which is enhanced in PHR1-overexpresssing plants ( Figure 2 ) . Our transcriptomic analyses reveal the large quantitative dimension of the Pi starvation transcriptional response and the central regulatory role of PHR1 and PHL1 . A total of 4170 genes , representing 18 . 5% of the genes analysed , displayed Pi starvation responsiveness ( Table 1 ) ; of these , 75% of induced and 65% of repressed genes showed decreased and increased expression , respectively , in the Pi-starved phr1 phl1 double mutant ( Table 1 ) , indicative of reduced responsiveness in the mutant lines . There is no precedent for a small number of related TF controlling a complex stress response to such a large extent , although a quantitatively similar role was described for two Snf1-related kinases , KIN10 and KIN11 , that act as central integrators in sugar/energy depletion responses [52] . As for physiological and developmental responses , many Pi starvation-responsive genes are also responsive to other stresses , yet their responsiveness to Pi starvation is compromised in the phr1 and phr1 phl1 mutants . Non-specific molecular responses can thus be controlled by stress type-specific regulatory systems . A paradigmatic example of this is represented by general stress response ( GSR ) genes; these genes , identified in two independent studies , have been ascribed to an independent regulatory system [4] , [5] . Nonetheless , a large proportion of Pi starvation-responsive GSR genes are controlled by PHR1 ( -like ) ( Table 2 ) . One way to reconcile the existence of an independent regulatory system for GSR genes and the observation that they are controlled by PHR1 ( -like ) TF is that PHR1 ( -like ) TF exert their regulatory role on these genes by acting on the GSR regulatory system . Also noteworthy is the finding that 65% of the genes repressed by Pi starvation are more highly expressed in the phr1 phl1 double mutant than in wild type after Pi starvation ( Table 1 ) . This indicates that a large proportion of the transcriptional repression response is also an integral part of the adaptive response , since it is evident that the phr1 phl1 double mutant is more sensitive to Pi starvation , as it cannot mount a correct response ( Figure 2B ) . Here we show that PHR1 ( -like ) regulation of Pi starvation-responsive genes involves both direct and indirect control . Direct control is essentially exerted on induced genes containing the P1BS ( GNATATNC ) sequence [12] , whereas transcriptional repression is essentially indirect . Indeed , it can be noted that genes identified as direct targets ( in which activation is independent of protein translation ) are highly enriched in Pi starvation-induced genes containing P1BS sequences in different parts of the gene , particularly in the promoter proximal region and even to a higher extent in the 5′UTR . This indicates that PHR1 acts most prominently as a transcriptional activator , and that control of transcriptional repression is mostly , if not completely indirect ( e . g . , via activation of a transcriptional repressor ) . A large proportion of the Pi starvation-induced genes ( more than 70% ) are also probably controlled indirectly by PHR1 , since only about 30% of Pi starvation-induced genes have a P1BS motif in their promoter proximal region , 5′UTR or 3′UTR , where P1BS content is significantly higher than in an average Arabidopsis gene . Another finding is the strong association between P1BS content in the promoter and the degree of Pi starvation inducibility ( Figure 4 ) . It is interesting that although other regions are also P1BS-enriched , particularly the 5′UTR , but also the 3′UTR , introns and 1 kb proximal downstream region of Pi starvation-responsive genes , P1BS content in these regions does not correlate with inducibility . This suggests that the role of P1BS differs qualitatively in these regions compared to its role in the promoter . The correlation between P1BS content in the promoter and gene inducibility is not strict , however; for IPS1 , we show that one of the P1BS motifs in its promoter is in fact dispensable for Pi starvation responsiveness . In any case , the higher P1BS content of highly upregulated genes suggests that bioinformatic searches for stimulus-specific cis-regulatory motifs will be more efficient if performed in highly responsive genes . Taken together , these observations suggest a simple evolutionary path to construct a complex adaptive response to a specific stress type , under the control of a central regulatory system . Our data are in agreement with a central regulator that controls pre-existing , shared genetic networks by acting on the regulators of those networks , as it is probably the case of GSR genes , rather than on each individual gene . In line with this idea , we found that in most cases , TF genes and non-TF genes are equally over-represented in the sets of genes responsive to Pi starvation and to any other type of stress ( Table S3 ) ; we would predict under-representation of TF genes if shared genes were exclusively controlled by independent stress type-specific regulators . Genes for which the transcription rate obtained via this indirect route was insufficient , as could be the case of RNS1 ( see below ) , might have been recruited under the direct control of the central regulator , similar to the situation in Pi starvation-specific networks . Here we demonstrate the key importance of P1BS in Pi starvation gene inducibility , reinforcing the importance of PHR1 . In addition to the fact that P1BS is overrepresented in phosphate starvation-induced genes , as shown here and elsewhere [9] , [11] , P1BS is highly conserved in a Pi starvation-responsive gene ( IPS1 ) . Mutation of critical P1BS motifs in promoters of Pi-responsive genes abolishes Pi starvation responsiveness in dicots and monocots ( Figure 5 ) [53] , and a minimal promoter containing four tandem copies of P1BS is specifically responsive to Pi starvation ( Figure 6 ) . The fact that a minimal promoter containing P1BS specifically responds to Pi starvation allowed us to examine the effect of several modulators of the Pi starvation response , and to show that this element can recapitulate Pi starvation-specific responsiveness ( Figure 6 ) ; this includes the effect of all the best known modulators of this response , such as sugars , cytokinins , arsenate and long distance systemic repression [11] , [46]–[50] . These data qualifys P1BS and , consequently , PHR1 ( -like ) TF as central integrators of the Pi starvation response ( Figure 6D ) . By analysing the function of a promoter responsive to Pi starvation and wounding ( RNS1 ) [43] , we show that the P1BS motif is necessary only for Pi starvation responsiveness and not for responsiveness to other types of stress . Conversely , mutation of PHR1 and PHL1 affect only RNS1 responsiveness to Pi starvation and not to other stress types ( Figure 5 ) . Independent multisignal responsiveness can thus also be attained through independent cis motifs in the promoter . Although our data indicate the importance of P1BS as a Pi starvation response cis motif , we also show that P1BS function is dependent on sequence context , and that P1BS alone is insufficient to drive Pi starvation responsiveness in the context of a natural promoter such as that of IPS1 ( Figure 5 and Figure 7 ) . Indeed , our phylogenetic and mutational analysis of IPS1 identified a second motif , motif B ( GAWGATNC ) , necessary for correct Pi starvation responsiveness of IPS1 . The conditional overrepresentation of motif B , dependent on the presence of P1BS ( Table S6 ) , and the finding that motif B is unable to drive Pi starvation responsiveness strengthens the idea that PHR1 and P1BS represent a central integrator module in Pi starvation responsiveness . The results of this study show that PHR1 and a functionally related member of its family comprise a central integrator system for the Pi starvation response . Pi limitation is a common condition in many natural soils , which implies that selective pressure against this stress has been very strong throughout evolution , underlining the adaptive value of this simple regulatory system of such a complex response . A consequence of our finding that a single TF family largely controls a stress response is that transcriptionally overlapping programs in response to different stress types can ultimately be controlled by independent regulatory systems . Such systems act indirectly , using ( pre-existing ) shared regulatory components in many targets , and directly on the remaining small proportion of target genes on average highly enriched in P1BS . The finding that the 5′UTR of PHR1 primary targets and of Pi starvation induced genes shows the highest overrepresentation in P1BS sequences , raises the possibility of an important role of this region in transcriptional control , in addition to its most commonly associated role in translational control . The fact that a large proportion of the transcriptionally repressed genes are controlled by PHR1 ( -like ) TF indicates that transcriptional repression is an integral part of the Pi starvation response , and not merely a consequence of plant malfunction under stress .
All Arabidopsis thaliana plants used in this study , including mutants and transgenic plants , were on the Columbia ( Col-0 ) background . phl1 was obtained from the SAIL collection ( SAIL_731_B09 ) [54] . Growth conditions and the complete Johnson medium containing 2 mM Pi ( KH2PO4 ) and 2% sucrose were as described [22] , [55] . For specific experiments , the concentration of Pi , sucrose , kinetine or arsenate ( NaH2AsO4·7H2O ) is indicated . Anthocyanin was extracted from rosettes of plants grown on Pi-lacking medium supplemented with 5 µM DEX for 12 days . Anthocyanin content was measured as described [56] . The method of Ames [57] was used to determine the cellular phosphate content of seedlings grown on complete medium for 12 days ( supplemented with 5 µM DEX when specified ) . Mean values were compared using Student's t-test . Plants were transformed by the vacuum infiltration method [58] . Routine molecular work was performed as described [12] , [59] , except where indicated . Sequences of primers used for PCR amplification and construction of genomic DNA/cDNA fragments are given in Table S7 . A NcoI-SpeI fragment containing the ORF of PHR1 was amplified by PCR from the PHR1 cDNA [12] purified and digested with NcoI and SpeI . This fragment was introduced into the binary vector pBHAGR , which contains the CaMV 35S promoter , the 3xHA epitope and a fragment of the rat glucocorticoid receptor ( GR ) cDNA encoding the receptor-binding domain , generating the recombinant expression cassette 35S:HA:GR:PHR1 ( pBHAGRPHR1 ) . The pBHAGR vector was generated introducing a BamHI-NcoI cDNA fragment codifying for the 277 carboxy-terminal amino acids of the rat glucocorticoid receptor [60] into a binary vector pBHA kindly supplied by Dr . F . Parcy ( Institut National de la Recherche Agronomique , Grenoble , France ) . The HindIII-BamHI 1kb fragment containing the IPS1 promoter and the XbaI-blunt 1kb fragment containing the RNS1 promoter were amplified by PCR . The mutated promoter sequences were generated as overlapping PCR products using semi-complementary primers with the mutated sequences . The PCR products were purified , digested with HindIII-BamHI ( IPS1 ) or XbaI ( RNS1 ) and inserted between HindIII-BamHI or XbaI-SmaI sites into the pBI101 vector [44] . The four tandem copies of P1BS ( 4xP1BS:GUS ) , the B motif ( 4xB:GUS ) constructs , and the 42bp IPS1 promoter fragment ( A-P1BS-B:GUS ) were generated by annealing semi-complementary primers , resulting in DNA fragments with HindIII and XbaI overhangs . The over-hanged DNA fragments were inserted between HindIII and XbaI sites at the 5′ end of a minimal 35S promoter in the pTi0046 plasmid . The pTi0046 plasmid contains a −46bp truncated version of the CaMV 35S promoter [45] into the BamHI site of the pBI101 vector . Quantitative PCR ( Q-RT-PCR ) was performed on three independent biological samples as described [61] . The pairs of primers used are described in Table S7 . PHR1 and PHL1 deletion derivatives were generated by in vitro translation ( or cotranslation in the dimerization experiments ) using the TnT T7 Quick System for PCR DNA ( Promega ) , as described [62] . PCR and labeling of promoter fragments and oligonucleotides , DNA binding reactions and EMSA were performed as described [63] . A cDNA fragment corresponding to a deletion derivative of PHR1 , Δ-PHR1 , encompassing amino acid residues 208–362 , that lacks transactivation domain was cloned into the pGBKT7 ( Gal4 DNA binding domain , BD; Clontech ) . We used this to screen a whole seedling cDNA library prepared in the pGADT7 vector ( Gal4 activation domain , AD , Clontech ) to detect PHR1-interacting proteins . One of these was Δ-PHL1 ( lacking amino acids 1–60 , details to be described elsewhere ) . To confirm protein interactions , the plasmids were cotransformed in Saccharomyces cerevisiae AH 109 cells following standard heat-shock protocols [64] . Successfully transformed colonies were identified on yeast synthetic drop-out lacking Leu and Trp; these colonies were resuspended in 30 mM NaCl and transferred to the same media plus β-gal or to selective media lacking Ade , His , Leu and Trp . Plates were incubated ( 30°C , 2–4 days ) . The empty vector pGADT7 was also cotransformed with the pGBKT7-DPHR1 construct as a negative control . Transcriptomic analyses were performed using the Affymetrix ATH1 platform . For the phosphate starvation response analysis , wt , phr1 and phr1 phl1 plants were grown for 7 days in complete ( +Pi ) or phosphate-lacking ( −Pi ) solid media , and roots and shoots were processed separately . For PHR1 direct target analysis , complete OxGR:PHR1 phr1 and phr1 plants were grown for 7 days in +Pi liquid media , then for 2 days in −Pi liquid media and harvested after 6 h treatment with 5 µM DEX and 10 µM CHX . In each experiment , RNA was isolated from three independent biological samples using the RNeasy plant mini kit ( Quiagen ) . Biotin-labeled cRNA was synthesized using One-Cycle target labelling and control reagents ( Affymetrix , Santa Clara , CA ) and fragmented into 35–200 bases in length . Three replicates for each condition were hybridized independently to the Arabidopsis ATH1 Genome array following manufacturer's recommendations ( Affymetrix ) . Each microarray was washed and stained with streptavidin-phycoerythrin and scanned at 2 . 5 µm resolution in a GeneChip Scanner 3000 7G System ( Affymetrix ) . Data analyses were performed using GeneChip Operating Software ( GCOS ) and analyzed using the affylmGUIR package [65] . Robust Muti-array Analysis ( RMA ) algorithm was used for background correction , normalization and expression levels summarization [66] . Differential expression analysis was performed with the Bayes t-statistics from the linear models for Microarray data ( limma ) . P-values were corrected for multiple-testing using the Benjamini-Hochberg method ( False Discovery Rate ) [67] . Except where indicated , genes were considered to be differentially expressed if corrected P values were <0 . 05 , and only genes with a signal log ratio more than one or less than minus one were considered for further analysis . For transcriptome comparisons we used microarray data for different treatments/stresses available in the GENEVESTIGATOR database ( https://www . genevestigator . com ) [30] . The two-fold up- and down-regulated genes were identified by the Meta-Analyser tool included in this platform . Transcription factor genes for transcriptomic analysis were identified in the RARTF Database ( http://rarge . psc . riken . jp/rartf/ ) [68] . Average expression signals for the Pi starvation treatment were expressed relative to those in complete media , converted to a log2 scale and imported into the MapMan software , which showed values in colour scale diagrams ( http://mapman . gabipd . org/web/guest/home ) [35] . The promoters regions of gene orthologs were obtained using commercially available GenomeWalker technology ( Clontech ) , following manufacturer's recommendations . Sequences of interest were obtained by two rounds of PCR amplification using as template the adapter-ligated genomic DNA from different Brassicaceae species , an IPS1-specific primer and the adaptor primer . Primary PCR was performed with seven cycles of 25 sec at 94°C and 4 min at 72°C , followed by 32 cycles of 25 sec at 94°C and 4 min at 67°C , with a final extension of 4 min at 67°C . Secondary PCR was performed using a 1∶50 dilution of the primary reaction product as a template and similar PCR cycling parameters , with 5 and 22 cycles of the first and second steps , respectively . PCR products were cloned into the pCRII-TOPO TA system ( Invitrogen ) . Sequences were aligned using DiAlign ( http://www . genomatix . de/cgi-bin/dialign/dialign . pl ) [69] . The analysis was performed on the Phylogeny . fr platform ( www . phylogeny . fr ) [70] . Sequences were aligned with MUSCLE ( v3 . 7 ) configured for highest accuracy . After alignment , ambiguous regions were removed with Gblocks ( v0 . 91b ) . The phylogenetic tree was reconstructed using PhyML program ( v3 . 0 aLRT ) . The default substitution model was selected assuming an estimated proportion of invariant sites ( of 0 . 021 ) and 4 gamma-distributed rate categories to account for rate heterogeneity across sites . The gamma shape parameter was estimated directly from the data ( gamma = 1 . 044 ) . Reliability for internal branch was assessed using the bootstrapping method ( 100 bootstrap replicates ) . The tree was represented withTreeDyn ( v198 . 3 ) . Arabidopsis Genome Initiative locus identifiers for the genes mentioned in this article are At5g29000 ( PHL1 ) , At4g28610 ( PHR1 ) , At5g43350 ( PHT1;1 ) , At2g02990 ( RNS1 ) , At3g09922 ( IPS1 ) , At5g20150 ( SPX1 ) , At5g03545 ( At4 ) , At4g33030 ( SQD1 ) and At3g17790 ( ACP5 ) . The GenBank accession numbers for the sequences of the proximal promoter region of IPS1 orthologs are GQ184774 ( Descurainia sophia , DsIPS1 ) , GQ184775 ( Arabis auriculata , AaIPS1 ) , GQ184776 ( Brassica intermedia , BiIPS1 ) and GQ184777 ( Lepidium campestre , LcIPS1 ) . The GEO accession number for the array experiments are GSE16722 and GSE20955 . | As sessile organisms , plants are often exposed to stress conditions , and have evolved adaptive responses to protect themselves from different types of stress . Some responses are stress type-specific whereas others are common to different stress types . Understanding how these responses are controlled is crucial for rational improvement of stress tolerance , a limiting factor in crop productivity . Here we examined the physiological and molecular responses to phosphate starvation and found that a single transcription factor family , represented by PHOSPHATE STARVATION RESPONSE REGULATOR 1 ( PHR1 ) , has a central role in the control of specific and shared phosphate starvation stress responses . In consonance with the importance of PHR1 , we found that the PHR1-binding sequence , present in most PHR1 direct targets , is a crucial cis motif for Pi starvation responsiveness . An artificial promoter controlled by PHR1 recapitulates responsiveness to Pi starvation and to modulators of this response , qualifying PHR1 family members as central integrators in Pi starvation signalling . This central integrator system also controls most transcriptional repression responses to Pi starvation , indicating that they are an integral part of the adaptive response , and not a consequence of plant malfunction due to stress . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"plant",
"biology/plant-environment",
"interactions",
"genetics",
"and",
"genomics/plant",
"genetics",
"and",
"gene",
"expression",
"plant",
"biology/plant",
"biochemistry",
"and",
"physiology"
] | 2010 | A Central Regulatory System Largely Controls Transcriptional Activation and Repression Responses to Phosphate Starvation in Arabidopsis |
Interspecies transmission of prions is a well-established phenomenon , both experimentally and under field conditions . Upon passage through new hosts , prion strains have proven their capacity to change their properties and this is a source of strain diversity which needs to be considered when assessing the potential risks associated with consumption of prion contaminated protein sources . Rabbits were considered for decades to be a prion resistant species until proven otherwise recently . To determine the extent of rabbit susceptibility to prions and to assess the effects of passage of different prion strains through this species a transgenic mouse model overexpressing rabbit PrPC was developed ( TgRab ) . Intracerebral challenges with prion strains originating from a variety of species including field isolates ( ovine SSBP/1 scrapie , Nor98- scrapie; cattle BSE , BSE-L and cervid CWD ) , experimental murine strains ( ME7 and RML ) and experimentally obtained ruminant ( sheepBSE ) and rabbit ( de novo NZW ) strains were performed . On first passage TgRab were susceptible to the majority of prions ( Cattle BSE , SheepBSE , BSE-L , de novo NZW , ME7 and RML ) tested with the exception of SSBP/1 scrapie , CWD and Nor98 scrapie . Furthermore , TgRab were capable of propagating strain-specific features such as differences in incubation periods , histological brain lesions , abnormal prion ( PrPd ) deposition profiles and proteinase-K ( PK ) resistant western blotting band patterns . Our results confirm previous studies proving that rabbits are not resistant to prion infection and show for the first time that rabbits are susceptible to PrPd originating in a number of other species . This should be taken into account when choosing protein sources to feed rabbits .
Prions are protein based , genome devoid , infectious agents causing Transmissible Spongiform Encephalopathies ( TSEs ) , a group of diseases classified as transmissible protein misfolding disorders [1 , 2] . Prions show a remarkable ability for interspecies transmission . Initially , a species barrier was defined , but extensive field and experimental evidence has been published proving that interspecies prion transmission is not an isolated phenomenon [3–6] . Interspecies transmission of prions has resulted in the generation of significant prion strain diversity and its incidence has been documented worldwide [3 , 4 , 7–10] . The existence of prion diseases has been documented for centuries with the earliest reports of scrapie cases dating back to 1732 [11] . In the last seven decades prions were also reported in other animal species , usually in the form of outbreaks , which somehow involved human intervention . Namely classical bovine spongiform encephalopathy ( BSE-C ) [12] , feline spongiform encephalopathy ( FSE ) [13] and transmissible mink encephalopathy ( TME ) [14] . Humans can also be included in the list of TSE susceptible species due to the Fore tribe from Papua New Guinea suffering from Kuru [15] or the relatively newly created variant Creutzfeldt-Jakob disease ( vCJD ) [3] . Cervidae is another family of animals currently affected by a , yet uncontrolled , epizooty: chronic wasting disease ( CWD ) [16] . Although classical animal prion disease strains , as opposed to the so called atypical prion disease strains [17–20] , have been documented for at least three centuries [11] , sporadic spontaneous generation of atypical prions has probably existed for as long as susceptible species have been present in large enough numbers for the spontaneous event to occur . Currently there is no evidence to suggest that any mammalian species cannot undergo a spontaneous disease-linked prion protein misfolding event [21] as long as there are sufficient numbers of individuals with the necessary lifespan . Although the mechanisms of interspecies prion transmission remain unknown , in vitro and in vivo studies have shown that species particularly susceptible to certain prion strains can actually be resistant to others which originated in the same or different species [9 , 21–28] . The ability of prions to adapt to new species and even generate new strains with pathobiological properties different from the original one is not an isolated phenomenon [9 , 27 , 29 , 30] . Therefore new prion strains may arise with the ability to infect new species previously considered resistant . Normal cellular prion protein ( PrPC ) is a host encoded protein , particularly abundant in nerve cells , which when misfolded is believed to acquire pathological properties leading to TSE neurodegenerative disease [1] . Several studies argue that certain species specific amino acid sequences of PrPC may render some species less susceptible to TSE [31–33] due to them being less prone to misfolding . This , along with absence of experimental evidence or TSE field cases described , led to belief that dogs , horses and rabbits ( leporids ) were more resistant to prion infection than other mammalian species [34 , 35] . Leporids have been the most intensely studied , both in vivo and in vitro , of the presumed prion disease resistant species . This is probably because rabbits are consumed by humans and also due to their comparatively small size and long lifespan which facilitates their use as experimental animals . Our group has proved recently , in contrast with the last three decades of reports , that rabbits are susceptible to prion diseases . Using protein misfolding cyclic amplification ( PMCA ) , inocula where generated in vitro which were infectious and transmissible in this species [23] and more recent studies have proven that rabbit PrPC has a misfolding ability comparable to other species as BSE prions have been shown to retain their in vivo strain properties after misfolding rabbit PrPC [36] . Houdebine’s group studied whether the genetic background of rabbits was responsible for their apparent prion resistance generated transgenic rabbits expressing ovine PrPC . Upon inoculation with scrapie , these rabbits succumbed to prion disease further proving that leporids are not resistant to prion disease [37] . In the present paper we report an extensive evaluation of the susceptibility of TgRab mice to a variety of prion strains by means of in vivo experiments . A transgenic murine model has been generated ad hoc for this purpose which overexpresses the leporid PRNP on a mouse Prnp-null background . This model , denoted TgRab , has already been shown to correlate well with the rabbit model [23] . Our results show the susceptibility of rabbits has been vastly underestimated previously and that they behave similarly to other species whose vulnerability and/or resistance to prion disease also varies depending upon the prion disease strain encountered .
Even though the actual rabbit model would be more suitable for this purpose there are several significant limitations ( size , cage space in biocontainment conditions , lifespan , expression levels , and budget required ) that are easily overcome by using a transgenic mouse model and such models have been of great use within the field of prion research . Based on our previous experience a new mouse line was generated by pronuclear injection of a construct consisting of the moPrP promoter and the rabbit PrP sequence . From a total of seven positive animals identified from the 83 pups obtained , five animal founders transmitted the transgene to their progeny . After backcrossing to a line that did not express endogenous PrP ( STOCK-Prnptm2Edin ) , expression levels of the transgene were analyzed by western blot . Two out of five transgenic lines expressed PrP at higher levels than the endogenous gene . However , only hemizygous line 58 showed a consistent expression pattern of 5x-6x that of the endogenous rabbit prion protein level and 10x-12x that of the endogenous mouse prion protein level ( S2 Fig ) . This line was selected for further studies . The low expressing lines were discarded since PrPC expression levels were lower than those found in WT rabbits and this would predictably diminish their susceptibility to prions . In previous experiments normal rabbit brain homogenate was seeded in vitro with different prion strains before applying serial automated PMCA ( saPMCA ) to determine the ability of rabbit PrPC to be converted by different PrPres conformations . The results of some of these experiments have been reported previously such as seeding with cattle BSE which generated BSE-RaPrPres [36] . Additional prion isolates were included in the present work , which successfully misfolded rabbit PrPC in vitro including SSBP/1 sheep scrapie , ME7 and RML murine adapted scrapie strains and CWD . The following rabbit adapted strains were generated respectively: SSBP/1-RaPrPres , ME7-RaPrPres , RML-RaPrPres and CWD-RaPrPres ( S1 Fig ) [23] . Spontaneously misfolded PrPres was also obtained from unseeded normal rabbit brain homogenates and named de novo RaPrPres . This spontaneous strain has been demonstrated to be infectious to rabbits [23] . Despite saPMCA not being a quantitative method , rabbit PrPC appeared to be quite susceptible to misfolding since all seeds tested were able to generate PK-resistant RaPrPres by or before round 7 and the unseeded homogenate produced RaPrPres by round 13 [23] . All in vitro-derived RaPrPres products were easily amplified further in vitro . The western blotting migration pattern of the obtained RaPrPres , particularly the unglycosylated band , was similar to the strains of origin used in the bovine and ovine strains tested . Accordingly , the following isolates were selected for in vivo challenge: BSE-C , SSBP/1 , ME7 , RML and CWD . Additionally , we included L-type atypical BSE ( BSE-L ) and Nor98 Atypical scrapie and the PMCA obtained de novo RaPrPres . The rationale for including the latter , in vitro generated , PrPres was that it was able to infect the natural host i . e . rabbits , our species of study [23] , de novo RaPrPres was the only positive control available . Finally , de novo NZW prions ( obtained from rabbits infected with de novo RaPrPres ) were also inoculated into this transgenic mouse model and even though most of these results have already been published [23] , some of them are discussed in the present paper . As reported for rabbits showing very long incubation times ( 766 dpi ) and a 33% attack rate ( 1/3 ) [23] , TgRab mice were also susceptible to de novo RaPrPres with a low attack rate ( 1/11 ) and a rather long incubation period of 604 dpi . However , upon inoculation with rabbit in vivo-adapted de novo RaPrPres ( de novo NZW ) the TgRab mice developed a 100% attack rate ( 8/8 ) with a shortened incubation period of 256 ( ±5 ) dpi ( Table 1 and Fig 1 ) . The same rate was obtained by Chianini et al . [23] in rabbits inoculated ( in second passage ) with this prion strain ( Table 1 ) . In vivo experimental challenges in rabbits and TgRab mice have shown a good correlation making the transgenic mouse model overexpressing rabbit PrPC a valid model to study rabbit prion susceptibility . Even though rabbits had been considered resistant to prion infection until recently [23] , TgRab mice could be infected with a number of the prions tested . Prions originating from BSE , i . e . cattle BSE-C and sheep BSE-C , were both infectious ( Table 2 ) . Cattle BSE-C showed an attack rate of 44 . 4% with an incubation period of 551 ( ±10 ) dpi . Interestingly Sheep BSE-C showed a 100% attack rate and a significantly shortened incubation period of 368 ( ±12 ) dpi ( P = 0 . 0069 , Mann-Whitney test ) without previous adaptation to rabbit compared to cattle BSE-C ( Fig 1 ) . This supports , once again , the idea that after passage through sheep BSE-C shows enhanced virulence [29] . The picture with scrapie-originating prion isolates was quite different . SSBP/1 prions were not able to infect TgRab ( mice survived for longer than 750 dpi ) . Two other murine adapted classical scrapie prion sources were tested , ME7 and RML , and both strains readily infected TgRab mice with attack rates of 50% and 70% and incubation periods of 360 ( ±41 ) and 371 ( ±6 ) dpi , respectively ( Fig 1 ) . Therefore , prion strains originating from classical scrapie were transmissible to TgRab mice but only after being adapted previously to rodents . This situation is similar to that found with CWD which will infect hamsters readily after passage through ferrets [9] . The new TgRab model was further characterized by testing its susceptibility to atypical prion strains using the more frequent isolates for each species , BASE ( BSE-L , cattle ) and Nor89 ( sheep ) . TgRab mice were resistant to infection on first passage with atypical scrapie prions ( living up to 775 dpi ) ( Fig 1 ) with one exception: a single animal ( euthanized at 742 dpi ) showed a positive result for PrPd by ELISA but was negative when examined by western blotting and IHC and showed no TSE related spongiform change . A second passage is ongoing to determine if this animal was truly infected . A 27% attack rate was present in the group inoculated with BSE-L with a mean incubation period of 280 ( ±26 ) dpi , a similar rate to that of cattle BSE-C but with a much shorter incubation period ( the number of positive animals per group was too low to assess statistical significance ) . As mentioned before , in vitro adaptation of CWD prion to rabbit PrPc was successful which indicated a potential susceptibility of rabbits in vivo . However , TgRab mice inoculated with CWD did not show any indication of a TSE on first passage , living up to 825 dpi ( Fig 1 and Table 2 ) . A second passage is ongoing to confirm these results . Biochemical and neuropathological characterization of the brains of the inoculated mice strongly suggests that TgRab mice are not only susceptible to multiple prion strains but are also able to maintain their strain features . Western blotting analysis of TgRab brain homogenates after protease K digestion revealed the characteristic three-band pattern with a predominant diglycosylated band and a 19-20kDa unglycosylated band in mice inoculated with BSE-derived strains ( Fig 2 ) . The brains of mice inoculated with RML showed a typical predominance of the monoglycosylated band and a 21kDa unglycosylated band . As shown in the 12B2 antibody developed membrane only ME7 and RML inoculated mice fully maintained the N-terminus specific epitope after PK digestion ( Fig 2 ) . The migration pattern of the bands from mice inoculated with de novo strains , both with the in vitro generated ( de novo-RaPrPres ) and the one obtained from NZW rabbits ( de novo NZW ) , was constant and showed a similar pattern to BSE-C even though , as shown later , the immunohistochemical features differed completely . No bands were observed in western blots of brains of mice inoculated with SSBP/1 , Atypical scrapie nor CWD or in any of the negative controls . Spongiform change and PrPd distribution throughout the brain was semi-quantitatively assessed in histological sections of the inoculated brains of TgRab mice ( Figs 3 and 4 ) . Classical BSE-derived strains , namely BSE-C and sheep BSE , yielded very similarly shaped curves characterized by a strong involvement of the medulla oblongata , mesencephalon and thalamus but sparing of the hypothalamus . Involvement of the cortices and hippocampus was less intense but present , particularly at the deeper layers of the parietal cortex , involving the corpus callosum and sometimes extending to the oriens layer of the hippocampal formation . This pattern is equivalent to the one observed for BSE-C in the botg110 mouse model previously published by our group [36] . Mice inoculated with BSE-L , in contrast , showed a widespread involvement of the neocortex and less so in the diencephalon , mesencephalon and medulla oblongata in accordance with the brain PrPd distribution observed in natural and experimental cases of BSE-L in cattle [18 , 38] . The type of PrPd deposits seen by immunohistochemistry was also distinct in all mice inoculated with classical BSE-derived strains and consisted of amyloid-like rounded plaques , often confluent , which were readily visible on HE stained sections and positively stained in sections subjected to immunohistochemistry for PrPd ( Fig 4A ) . BSE-L inoculated mice lacked plaque type deposits and showed a very different punctate immunolabelling pattern in the neuropil and perikarya ( Fig 4A ) . This was consistent with the pattern obtained in tgBov mice when inoculated with BSE-L . The scrapie-derived strains RML and ME7 showed PrPd deposits with a tropism for the diencephalon , including a consistent involvement of the hypothalamus ( distinct from BSE strains ) , the mesencephalon and the medulla oblongata and also showed tropisms for the neocortex and cerebellar cortex . The PrPd type , on immunohistochemistry , was distinguishable from that of BSE infected mice , as it was comprised of a fine punctate pattern in the neuropil and perikarya ( Fig 4 ) . The lesion and PrPd distribution of the rabbit-obtained de novo NZW strain showed a tropism confined to the diencephalon , including a consistent involvement of the hypothalamus , mesencephalon and medulla oblongata while sparing the cortices and hippocampus . The PrPd type , on immunohistochemistry , consisted of a fine punctate pattern in the neuropil and perikarya resembling that observed in ME7 and RML infected mice . The data presented validate the TgRab model to study rabbit susceptibility to prion strains . However , the TgRab line 058 , chosen because it was the transgenic line showing the highest PrPC expression levels , also showed a spontaneous phenotype secondary to PrPC overexpression , as described by Westaway and coworkers [39] , which needs to be taken into account when evaluating the results of any given experiment . Similar changes have been observed previously in other useful transgenic models overexpressing PrPc [40 , 41] . In this phenotype , between 300 and 400 days , the majority ( over 80% ) of hemizygous mice ( 5x-6x PrPc expression compared to normal levels; S2 Fig ) developed gait abnormalities in the hindquarters that progressed slowly to complete hind-limb paralysis and atrophy of muscles ( S3 Fig ) . The animals were able to feed , drink and groom normally and when it was not the case , as with any infected animal that reached the end point criteria , they were humanely euthanatized . See death time points for control groups in Fig 1 . The same clinical presentation , but with enhanced severity , appeared in mice homozygous ( 10x-12x PrPc expression levels ) for the transgene as early as 60 days of age . Microscopically , the skeletal muscle tissue showed irregular diameter of the muscle fibers along with the presence of anguloid fibers , centralization of nuclei and substitution by adipose tissue proliferation in the endomysium ( S3E Fig and Fig 3 ) . These changes are compatible with neurogenic atrophy . Lesions were observed also in the central nervous system and consisted of an intense spongiform change in the white matter , particularly in the corpus callosum and internal capsule ( S4 Fig ) . The remaining brain parenchyma also showed diffuse moderate spongiosis , which was more evident in the diencephalon and brainstem and particularly intense in mice euthanized at older ages . Even though no PrPd was detected by western blotting or ELISA in any of the control animals , upon immunohistochemistry an intense PrPC background immunolabelling was present throughout the brain in agreement with the known overexpression of PrPC . Additionally , a more intense labelling was observed , that could be mistaken for PrPd signaling , which consisted of punctuate labelling around and within the cytoplasm of neurons , mainly located in the cortices but occasionally in the diencephalon and brain stem . Also , in the white matter , a punctate immunolabelling pattern was observed . Certain regions consistently showed strong immunolabelling of PrPC including the cochlear nucleus in the medulla oblongata and the cerebellar cortex where a diffuse labelling was observed in the molecular layer and intense labelling in the granular layer , depicting the synaptic glomeruli ( S4 Fig ) . Despite most of the animals displaying an overt overexpression phenotype , characterization allowed clear discrimination of this from bona fide prion infection in this model .
Brain homogenates ( 10−1 in PBS ) for use as seeds for PMCA or direct intracerebral inocula were prepared manually using a glazed mortar and pestle from brains of animals clinically affected by various TSE: BSE-C and BSE-L field cases supplied by the Laboratorio Central de Veterinaria ( Algete , Madrid , Spain ) , SSBP/1 and ME7 supplied by Animal Heath and Veterinary Laboratory Agency ( New Haw , Addlestone , Surrey , UK ) , CWD from the thalamus area of the brain of a female Mule deer , genotype 225SS , infected with CWD ( 04–22412 WSV2 EJW/JEJ ) , supplied by Department of Veterinary Sciences ( Laramie , WY , USA ) , RML supplied by Rocky Mountain Laboratories ( Hamilton , MT , USA ) and Sheep BSE supplied by Ecole Nationale Vétérinaire ( Toulouse , France ) . The atypical scrapie isolate was obtained from a field case diagnosed in PRIOCAT laboratory , CReSA-IRTA ( Barcelona , Spain ) . Rabbit spontaneous prions were those obtained in the rabbit bioassays conducted in the Moredun Research Institute , Scotland [23] . The in vitro prion replication and PrPres detection of amplified samples was performed as described previously with minor modifications [23 , 42] . Briefly , rabbit brains used for substrate were perfused using PBS + 5 mM EDTA and the blood-depleted brains were frozen immediately until required for preparing the 10% rabbit brain homogenates ( PBS + NaCl 1% + 1% Triton X-100 ) . 50–60 μl of 10% rabbit brain homogenate , either unseeded or seeded with the corresponding prion strain were loaded onto 0 . 2-ml PCR tubes and placed into a sonicating water bath at 37–38°C without shaking . Tubes were positioned on an adaptor placed on the plate holder of the sonicator ( model S-700MPX , QSonica , Newtown , CT , USA ) and subjected to incubation cycles of 30 min followed by a 20 s pulse of 150–220 watts sonication at 70–90% of amplitude . Serial rounds of PMCA consisted of 24-48h of standard PMCA followed by serial in vitro 1:10 passages in fresh 10% rabbit brain homogenate substrate . An equivalent number of unseeded ( 4–6 duplicates ) tubes containing the corresponding brain substrate were subjected to the same number of rounds of saPMCA in order to control cross-contamination and/or the generation of spontaneous PrPres . The detailed protocol for PMCA , including reagents , solutions and troubleshooting , has been published elsewhere [43] . PMCA treated samples were incubated with 85–200 μg/ml of protease K ( PK ) for 1 h at 42°C with shaking ( 450 rpm ) as described previously [44] . Digestion was stopped by adding electrophoresis Laemmli loading buffer and the samples were analyzed by Western blotting . After isolation by PCR amplification using 5’ CCGCCGTACGTCATCATGGCGCACCTCGGCTAC 3’ and 5’ GGGGCCGGCCTCATCCCACGATCAGGAAG 3’ as primers , the open reading frame ( ORF ) of the rabbit PRNP gene was cloned into the pGEM-T vector ( Promega ) . The rabbit-PrP ORF was excised from the cloning vector by using the restriction enzymes BsiWI ( Thermo Fisher Scientific Inc . ) and FseI ( New England Biolabs Ltd . ) and then inserted into a modified version of MoPrP . Xho vector [45] as described previously [46] , which was also digested with BstWI and FseI . This vector contains the murine PrP promoter and exon-1 , intron-1 , exon-2 and 3’ untranslated sequences . The transgene was excised using NotI and purified with Invisorb Spin DNA Extraction Kit ( Inviteck ) according to the manufacturer recommendations . Transgenic mouse founders were generated by microinjection of DNA into pronuclei following standard procedures [40] . DNA extracted from tail biopsies was analyzed by PCR using specific primers for the mouse exon 2 and 3’ untranslated sequences ( 5’ GAACTGAACCATTTCAACCGAG 3’ and 5’ AGAGCTACAGGTGGATAACC 3’ ) . Those which tested positive were bred to mice null for the mouse Prnp gene in order to avoid endogenous expression of mouse prion protein . Absence of the mouse endogenous Prnp was assessed using the following primers: 5’ ATGGCGAACCTTGGCTACTGGC 3’ and 5’ GATTATGGGTACCCCCTCCTTGG 3’ . The rabbit PrP expression levels of brain homogenates from transgenic mouse founders were determined by western blot using anti-PrP MAb L42 antibody ( RIDA-Biopharm , Darmstadt , Germany ) and compared with the PrP expression levels from NZW rabbit brain homogenates . Animals homozygous for the transgene showed a spontaneous clinical phenotype as early as 60 days old , resembling the one described in the results section , but more severe , requiring euthanasia at 60–120 days old . Due to this , hemizygous mice were maintained for subsequent studies . The international code to identify this transgenic mouse line is STOCK-Prnptm2Edin Tg ( moPrpn rabPrP ) 58Bps although throughout the paper they are referred to as TgRab mice . Mice of 42–56 days of age were intracerebrally inoculated under gaseous anesthesia ( Isoflurane ) through the right parietal bone . A 50 μl SGC precision syringe was used with a 25 G gauge needle and coupled to a repeatability adaptor fixed at 20 μl . A dose of buprenorphine was subcutaneously injected before recovery to consciousness to reduce post-inoculation pain . Mice were kept in a controlled environment at a room temperature of 22°C , 12 h light-darkness cycle and 60% relative humidity in HEPA filtered cages ( both air inflow and extraction ) in ventilated racks . The mice were fed ad libitum , observed daily and their clinical status assessed twice a week . The presence of ten different TSE-associated clinical signs [47] was scored . The experimental groups are listed in Table 2 . As the hemizygous mice had a slight spontaneous phenotype due to PrPC overexpression ( see Results ) , involving gait abnormalities , animals were euthanized following the end-point criteria ( body weight , measurable clinical signs , physical appearance , unprovoked behavior and response to external stimuli ) . Positive TSE diagnosis relied principally on the detection of PrPd ( either by immunohistochemistry and/or western blotting or ELISA ) and associated spongiform change in the brain parenchyma . All experiments involving animals were approved by the animal experimentation ethics committee of the Autonomous University of Barcelona ( Reference number: 585–3487 ) in agreement with Article 28 , sections a ) , b ) , c ) and d ) of the “Real Decreto 214/1997 de 30 de Julio” and the European Directive 86/609/CEE and the European council Guidelines included in the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes . When the clinical end-point criteria were reached , mice were euthanized by an overdose of pentobarbital administered intraperitoneally followed by decapitation . The brain was immediately extracted and placed into 10% phosphate buffered formalin . From each mouse a rostral section of the brain ( including olfactory bulbs and frontal cortex ) , a caudal fraction of the medulla oblongata and the whole spinal cord were kept frozen ( for biochemical studies and second passage ) . Transversal sections of the remaining brain tissue were performed at the level of the piriform cortex , optic chiasm and medulla oblongata . Samples were embedded in paraffin-wax after dehydration through increasing alcohol concentrations and xylene . Four micrometer sections were mounted on glass microscope slides which were stained with hematoxylin and eosin for morphological evaluation . Additional sections were mounted in 3-trietoxysilil-propilamine-coated glass microscope slides for immunohistochemistry . A pool of all frozen central nervous tissues samples was homogenized 1:10 ( W/V ) in PBS using closed tubes containing ceramic beads , placed in a ribolyzer ( Precess , Bio-Rad ) and subsequently analyzed either by western blotting , as described above , or by ELISA ( IDEXX , Herdcheck ) . The latter is a commercial ELISA based on the affinity of misfolded prions to an anionic substrate ( termed Seprion ) . A new threshold was defined to adapt to the higher densitometry readings obtained when working with samples with PrPc overexpression: only samples with a ratio spectrophotometry reading/cutoff over 5 were considered positive . Immunohistochemistry ( IHC ) against PrPd was performed as described previously [48] . Briefly , deparaffinized sections were subjected to epitope unmasking treatments: immersed in formic acid and boiled at low pH ( 6 . 15 ) in a pressure cooker and pre-treated with proteinase K . Endogenous peroxidases were blocked by immersion in a 3% H2O2 in methanol . Then , the sections were incubated overnight with anti-PrP MAb 6H4 primary antibody ( 1:2000 , Prionics AG ) and subsequently visualised using the DAKO Goat anti-mouse EnVision system ( Ref . K400111/0 ) and 3 , 3’diaminobenzidine as the chromagen substrate . As a background control , incubation with the primary antibody was omitted . Histological lesions ( i . e . spongiform change ) and PrPd immunolabelling were evaluated under a light microscope by a pathologist . A semi-quantitative approach was used to obtain comparable data from the different prions used to challenge mice . Spongiform lesion and PrPd immunolabelling were scored separately . A total of 15 different brain regions were chosen: piriform cortex ( Pfc ) , hippocampus ( H ) , frontal cortex ( Fc ) , parietal cortex ( Pc ) , temporal cortex ( Tc ) , occipital cortex ( Oc ) , thalamus ( T ) , hypothalamus ( HT ) , mesencephalon ( M ) , medulla oblongata ( Mobl ) , cerebellar nuclei ( Cm ) , cerebellar vermis ( Cv ) and cerebellar cortex ( Cc ) . Scores ranging from ( 0 ) absence of spongiosis or immunolabelling: ( 1 ) mild , ( 2 ) moderate , ( 3 ) intense and ( 4 ) maximum intensity of lesion or immunolabelling were assigned to each brain area studied ( Fig 3 ) . Each area was investigated globally as region for the scoring . Brain profiles were plotted as a function of the anatomical areas which were ordered along the X axis in an attempt to represent the caudo-rostral axis of the brain . This methodology was adapted from a previous study performed on BSE-infected cattle [49] . Graphs were plotted using Microsoft Office 2007 Excel software .
This is the first report of in vivo evidence suggesting that TgRab mice are susceptible to cross species transmission of prion strains . This not only reinforces that rabbits can no longer be considered TSE resistant , but also that there is a possibility they could act as a reservoir for other prion strains . As such , rabbits must be taken into account when determining the epidemiology of several TSE both in relation to the species of origin , especially sympatric ones , but also to potential zoonotic transmission . In previous studies we demonstrated that rabbits were able to propagate abnormal prions and that these were transmissible to other rabbits . However , this was only one prion strain which was generated de novo in an in vitro PMCA assay in rabbit brain homogenate ( a spontaneous rabbit prion strain ) and on first passage it had only a very limited attack rate [23] . This new mouse model , which responded in a comparable manner to rabbits when challenged with the same in vitro generated rabbit derived inoculum , has allowed us to evaluate the TgRab’s susceptibility to a number of actual field prions strains from a variety of different species . Although the use of rabbits would have been the most appropriate model there are strong , particularly budgetary , limitations due to the longer lifespan of rabbits and the need to use level 3 biosafety facilities . Thus , a transgenic mouse model overexpressing rabbit PrPC was designed to overcome these limitations and allow us to determine its susceptibility to different prion strains . No polymorphisms have been described in the PRNP rabbit gene , therefore several mouse transgenic lines were generated expressing rabbit PrPC at different expression levels . The line with the highest possible PrPC expression levels was selected to allow for easier prion propagation capacity but the overexpression was not so high as to generate a spontaneous phenotype at an early age which would preclude the attainment of infectivity/susceptibility data . The hemizygous TgRab line met these criteria with levels of PrPC between 5 to 6 times those present in rabbits . The use of transgenic mice overexpressing ovine PrPC to obtain the infectivity titer of specific prion isolates has been shown to be equivalent to titrations obtained through bioassay in the natural host [50] . Phenotyping of the newly developed prion transgenic model was essential , especially as the levels of PrPc expression present have not been shown to be problematic in other models [41 , 46] . Eighty percent of the TgRab mice presented with a late onset spontaneous neurological disease phenotype ( S3 Fig and S4 Fig ) which , fortunately , did not interfere in the interpretation of prion susceptibility results . This allowed us to work with a model that faithfully reproduced the behavior in rabbits with respect to their capability to propagate different prion strains . One cannot exclude the possibility that the presence of spontaneous disease might create a toxic environment in the brain which artificially enhances the transmission of certain strains . Therefore a thorough knowledge of the PrPC overexpression-related changes in uninfected controls was essential to identify the true prion disease status and validity of susceptibility . Lesion morphology and profiling within the brain and identification of specific PrPd deposition-types allowed unequivocal identification of infected animals ( either spontaneous or as a result of an inoculation ) . Further biochemical detection of the presence of PrPres by western blotting confirmed the ability of morphological techniques to identify an infected animal . Additionally , as PrPC overexpression may mask an incipient infection , second passages are required to confirm if rabbits are totally resistant to those prion isolates to which they initially appeared to be , such as SSBP/1 , atypical scrapie or CWD , and these experiments are ongoing . Once validated the TgRab model was used to evaluate which TSE strains were able to infect the model ( Table 2 ) . Previous attempts in rabbits had concluded they were resistant , probably due to incomplete studies and the strong barrier of rabbits to propagate prions [34] . Initially classical cattle BSE , the most relevant field strain , was tested and found to be infectious on first passage with a low attack rate ( 4/9 ) and relatively long incubation period ( 551dpi±10 ) . The strain properties observed in the infected TgRab mice ( western blotting , brain lesion and PrPd deposition profiles ) were typical of BSE and indistinguishable from those obtained in other BSE murine models [36] . Parallel bioassay studies were conducted with the BSE isolate previously amplified in vitro using rabbit normal brain homogenate as a substrate ( BSE-RaPrPres , this inoculum was characterised previously in a TgBov mouse model by our group [36] ) . These animals showed a 100% ( 12/12 ) attack rate and a shortened incubation period ( 396dpi ±12 vs 551dpi ±10 ) compared to the cattle BSE inoculated TgRab mice . This reduction already indicated that a transmission barrier between species had been overcome thanks to the in vitro adaptation of the cattle BSE-C to rabbit PrPC , a second passage was performed from that isolate which also showed a 100% attack rate ( 3/3 ) . Its incubation period was reduced to 322dpi ±12 ( mean ± s . e . m . ) indicating further host adaptation ( S5 Fig ) . SheepBSE , derived from BSE-C , infected TgRab mice with a 100% attack rate ( 9/9 ) , a relatively short incubation time ( 368±10 dpi ) and with lesion and PrPd brain profiles identical to those of BSE-C inoculated mice , suggesting that the same strain was being propagated through both isolates . This enhanced virulence of sheepBSE compared to BSE-C has been previously demonstrated in other experimental scenarios [29 , 51] . The results obtained with sheep scrapie differed completely as , in agreement with early experiments in rabbits [34] , none of the TgRab mice inoculated with SSBP/1 showed any evidence of a prion disease on first passage . However , this result does not preclude that , if further in vivo SSBP/1 passages were to be performed , the transmission barrier would be crossed . As in the case of BSE in the bank vole ( Myodes glareolus ) , where after an initial resistance a bank vole adapted BSE strain was obtained which was highly transmissible [52 , 53] . Conversely , ME7 and RML scrapie , both murine adapted sheep scrapie strains , infected TgRab mice on first passage and their incubation times , PrPres biochemical profiles , lesion profiles and PrPd deposition patterns were clearly distinguishable from cattle derived strains . Together these data are the first evidence that TgRab mice are not only able to propagate prions but they do it maintaining clearly the different distinguishing strain features ( Figs 1 , 3 and 4 ) which strongly suggests that rabbits may also . It is noteworthy that both ME7 and RML , which originated from serial passages of SSBP/1 in different rodents [54 , 55] , directly propagated in TgRab mice on first passage . Conversely , SSBP/1 did not infect TgRab mice on first passage . The murine adapted prion strains behaved differently to their parent strain , possibly because passage through rodents had selected for a strain capable of crossing the rodent species barriers . The situation is analogous to CWD which will infect hamsters after initial passage through ferrets [9] . In the present work , previous adaptation of scrapie to rodents , possibly resulting in a higher sequence identity in some specific and crucial PrP regions with rabbits compared to sheep , allowed rodent adapted scrapie prions to misfold rabbit PrPC . In previous studies ME7 did not infect rabbits after 4–5 years of incubation , with the exception of a single inconclusive case [23 , 34] . This result is difficult to extrapolate since we are discussing different species , of differing lifespans and with a species barrier between them . The PrPC overexpression in TgRab may have allowed ME7 to propagate more efficiently than in rabbits which suggests that if the original rabbit experiments had been performed over the maximum lifespan of rabbits ME7 may have propagated on first passage also . Once BSE in cattle has been virtually controlled , CWD in cervids is the animal prion disease with the most repercussions , at least in the North American continent . The uncertainty of its transmissibility to humans [56] and its unique ability to spread through the free ranging cervid population make its study highly relevant with respect to transmissibility to other species . Moreover CWD prions are known to be shed and are highly persistent in the environment . Rabbits are a sympatric species with cervids . Even though CWD has been shown to transmit on first passage to many species including sheep , cattle [57] , squirrel monkeys [58] , cats [59] , hamsters [60] , ferrets [9] , mink [61] , bank voles and deer mice ( Genus Peromyscus ) [62] its transmissibility efficiency is relatively low with very long incubation periods and low attack rates . For instance , wild type mice could not be readily infected , so tga20 mice overexpressing murine PrPC were required to prove susceptibility to CWD [63] or required a second passage [64] . Another example is the transmission of CWD to cats , which required an incubation period of longer than 4 years [59] . The present study showed CWD was not able to infect TgRab on first passage ( 0/12 ) . Further experiments are required to confirm the resistance of rabbits to CWD including a blind second passage and bioassays with CWD previously passaged in other species , especially rodents [9] . This will rule out an analogous situation as the one observed in this paper with sheep scrapie whereby SSBP/1 does not transmit to TgRab but murine passaged counterparts , ME7 and RML , do . With respect to the atypical prion strains of purported spontaneous origin [18 , 65 , 66] , BSE-L infected TgRab mice on first passage and , although the attack rate was low ( 3/11 ) , they had the shortest incubation period observed in this model so far ( 221dpi for the first animal to die , mean 280±26dpi ) . The lesion and PrPd deposition brain profiles differed considerably from those of BSE-C . None of the TgRab mice inoculated with atypical scrapie showed evidence of a TSE with the exception of one animal , euthanized at 742 dpi which , even though no histological lesions nor PrPd deposits were present suggestive of infection , it was positive by PrPd ELISA . This result could not be confirmed by western blotting . However , this ELISA detects PrPd through its affinity to an anionic ligand not due to its resistance to protease K so we cannot rule out this single mouse was positive . A second passage is ongoing which will determine the result . Initial in vitro experiments predicted that BSE as well as SSBP/1 and CWD isolates were able to missfold rabbit PrPC . However , a discrepancy was found with the bioassay results since neither SSBP/1 nor CWD infected TgRab mice on first passage . Several saPMCA rounds were needed in order to amplify the different isolates , varying in number depending of each strain . Thus , it is not surprising that on first passage some of the isolates do not transmit . Besides the PRNP sequence , another component of the transmission barrier is the genetic background in which each PrPC is contained . This has been demonstrated by infectivity studies showing BSE propagated more efficiently in RIII mice than C57/Black mice , two mice strains of the same species with the same PRNP gene [67] . Or when the genetic background ( i . e . passage through different inbred mouse lines ) determined not only the incubation period but also the propagation of two biochemically different BSE-derived strains [68] . For these reasons the belief that rabbits were resistant to prion infection was not only attributed to the rabbit PrPC sequence but also to its genetic background . To study whether the genetic background of rabbits was responsible for the apparent prion resistance , Houdebine’s group generated transgenic rabbits expressing an ovine PrPC which was known to easily misfold . Upon inoculation with ovine prion strains these rabbits succumbed to prion disease further proving that rabbits are not resistant to prions ( results published paired with this article ) and that the genetic background is not a limiting factor [37] . The differential susceptibility observed between actual rabbits and the transgenic model presented here can be explained by the higher PrPC expression levels of TgRab mice . Lower expression mouse lines would probably only be susceptible on first passage to strains previously adapted to rabbit PrPC as occurs with rabbits . It has taken more than three decades to finally dismiss the rabbit as a prion resistant species . We believe that the studies presented here confirm that in vitro studies are of great help in interpreting in vivo results , leave no room for misinterpretation , and that it can be ascertained that rabbits , and probably all other mammal species [21] , are susceptible to infection by specific prion strains . The prion strain and its species of origin determine the extent of susceptibility , but neither rabbit PRNP nor their genetic background suggest they are resistant to prion propagation . Unfortunately , as with other mammals , the exact molecular mechanisms governing the capricious choice of strains that can be propagated in a certain species is still unknown . In light of our results , especially susceptibility to spontaneous cattle prions ( BSE-L ) , the restrictions on rabbits being fed ruminant protein should be maintained sine die to minimize the chances of any prion strain transmitting to rabbits . | Prions , the infectious agents responsible for causing mad cow disease , amongst other diseases , can transmit from one species to another . For example , Bovine Spongiform Encephalopathy can transmit to humans resulting in invariably fatal variant Creutzfeldt-Jakob Disease . We wanted to study the susceptibility of rabbits as , until recently , they were considered a prion resistant species . Once proven otherwise , we wanted to know which particular prions rabbits were susceptible to . With this aim , a transgenic mouse was designed expressing the rabbit prion protein gene instead of the corresponding mouse gene to model the transmission barrier between rabbits and other species . The resultant mice where challenged with several field prion isolates including classical and atypical strains of Bovine Spongiform Encephalopathy , sheep Scrapie and cervid Chronic Wasting disease . The transgenic mice were susceptible to classical and atypical Bovine Spongiform Encephalopathy prions and also to mouse-adapted Scrapie prions . This information must be taken into account when assessing the risk of using ruminant derived protein as a protein source to feed rabbits . | [
"Abstract",
"Introduction",
"Results",
"Materials",
"and",
"Methods",
"Discussion"
] | [] | 2015 | Transgenic Mouse Bioassay: Evidence That Rabbits Are Susceptible to a Variety of Prion Isolates |
Fungi , such as Candida spp . , are commonly found on the skin and at mucosal surfaces . Yet , they rarely cause invasive infections in immunocompetent individuals , an observation reflecting the ability of our innate immune system to control potentially invasive microbes found at biological boundaries . Antimicrobial proteins and peptides are becoming increasingly recognized as important effectors of innate immunity . This is illustrated further by the present investigation , demonstrating a novel antifungal role of histidine-rich glycoprotein ( HRG ) , an abundant and multimodular plasma protein . HRG bound to Candida cells , and induced breaks in the cell walls of the organisms . Correspondingly , HRG preferentially lysed ergosterol-containing liposomes but not cholesterol-containing ones , indicating a specificity for fungal versus other types of eukaryotic membranes . Both antifungal and membrane-rupturing activities of HRG were enhanced at low pH , and mapped to the histidine-rich region of the protein . Ex vivo , HRG-containing plasma as well as fibrin clots exerted antifungal effects . In vivo , Hrg−/− mice were susceptible to infection by C . albicans , in contrast to wild-type mice , which were highly resistant to infection . The results demonstrate a key and previously unknown antifungal role of HRG in innate immunity .
The innate immune system , based on antimicrobial peptides ( AMP ) and proteins , provides a first line of defence against invading microbes [1]–[3] . At present , over 880 different AMPs have been identified in eukaryotes ( www . bbcm . univ . trieste . it/tossi/pag5 . htm ) . During recent years it has become increasingly evident that many AMPs , such as defensins and cathelicidins , are multifunctional , also mediating chemotaxis , apoptosis , and angiogenesis [4]–[6] . Conversely , molecules previously not considered as AMPs , including proinflammatory and chemotactic chemokines [7] , neuropeptides [8] , peptide hormones [9] , [10] , the anaphylatoxin peptide C3a [11] , [12] , growth factors [13] and kininogen-derived peptides [14]–[17] have recently been found to exert antibacterial activities . Histidine-rich glycoprotein ( HRG ) is a plasma protein which was first isolated in 1972 by Heimburger et al . [18] , [19] . The protein is present in human plasma at 1 . 5–2 µM , but the local concentration when HRG is released from activated platelets is likely to be higher [20]–[22] . It is a type 3 cystatin family protein [23] , along with α-2-HS-glycoprotein/fetuin-A , fetuin-B and kininogen , and is found in vertebrates as well as in some invertebrates . The structure contains two cystatin-like domains , a central histidine-rich region ( HRR ) with highly conserved GHHPH tandem repeats flanked by proline-rich regions , and a C-terminal region [20] . This modular structure of HRG facilitates multiple interactions , involving ligands such as heparin , plasminogen , fibrinogen , thrombospondin , heme , IgG , FcγR , and C1q . Due to its high content of histidine residues ( ∼13% ) , which are concentrated to the HRR , HRG can acquire a positive net charge either by incorporation of Zn2+ , or by protonation of histidine residues at acidic conditions [20] . In this context it has been proposed that HRG acts as a pH and Zn2+ sensor , providing a mechanism for regulating the various activities of HRG [24] . HRG has recently been ascribed antiangiogenic [25] effects in vitro , as well as antitumor [26] effects in vivo . Recent studies on Hrg−/− mice furthermore suggest that HRG plays a role as both an anticoagulant and an antifibrinolytic modifier , and may regulate platelet function in vivo [22] . Previous work has also demonstrated that HRG exert direct antibacterial activities in vitro which are dependent on Zn2+and pH [27] . However , as many cationic proteins and peptide sequences display antimicrobial properties in vitro , the ultimate role ( s ) of HRG in innate immunity in vivo still remained unresolved . During the course of our studies , we observed that HRG had a significant activity against Candida . Candida , an eukaryote , is present as a commensal at mucosal surfaces and on skin . Although it may cause life-threatening sepsis in immunocompromised individuals it seldom causes invasive disease in immunologically normal individuals [28] . We therefore speculated that HRG could constitute a natural defence against Candida infections . In the present study we show , using a combination of microbiological , biochemical , and biophysical methods , that HRG exerts a potent antifungal activity particularly at low pH , which is mediated via its HRR , and targets ergosterol-rich membrane structures such as those of Candida . In mouse infection models , HRG protects against systemic infection by Candida , indicating a previously undisclosed antifungal role of HRG in innate immunity .
In order to assess possible antifungal effects of HRG , we tested the activity of the protein against various Candida isolates . HRG was shown to be antifungal against C . parapsilosis at normal pH ( 10 mM Tris , pH 7 . 4 ) , and the activity was significantly increased in low pH buffer ( 10 mM MES , pH 5 . 5 ) ( Figure 1A ) . It is well-known that activities of AMPs and antimicrobial proteins are dependent of the microenvironment . For example , various chemokines , defensins , LL-37 as well as heparin binding protein are partly , or completely , antagonized by high salt conditions or the presence of plasma proteins in vitro [27] , [29] , [30] . Therefore , the influence of salt was tested . The results showed that HRG partially retained antifungal activity at physiological Cl− levels ( 0 . 1 M ) but only at low pH ( Figure 1B ) . The antifungal activity against C . parapsilosis was both time- and dose-dependent ( Figure 1C ) . In subsequent experiments various Candida strains ( C . parapsilosis , C . albicans , C . glabrata and C . krusei ) were incubated with HRG ( at 3 µM ) at neutral as well as low pH . Figure 1D demonstrates , in line with the above experiments , that HRG is particularly active at low pH . Thus , C . parapsilosis , C . albicans and C . krusei were all nearly completely killed by HRG at low pH , whereas C . glabrata exhibited a partial resistance at this concentration of HRG , the latter in analogy to C . glabrata displaying some resistance against histatin 5 [31] . Next , to investigate the binding of HRG to fungi , C . papapsilosis was incubated with HRG at low pH , washed , and analysed by immunoblotting . Since previous results indicated that the HRR of HRG , which binds heparin/heparan sulfate , mediates antibacterial effects [27] , heparin was added for competition of binding to Candida . Figure 1E shows that HRG was able to bind to the fungal cells and that the binding was partially inhibited by an excess of heparin . This finding is compatible with the observation that heparin completely blocks the antifungal effect of HRG ( Figure S1 ) . As demonstrated by flow cytometry , HRG bound to C . parapsilosis at neutral pH , and the binding was significantly increased at pH 5 . 5 ( Figure 1F ) , results compatible with the fungal killing assays ( Figure 1A ) . In summary , therefore , the results demonstrate that the antifungal actions of HRG were pH-dependent and likely mediated via the heparin-binding region of the protein . Many AMPs kill microbes by membrane lysis , while others may translocate through membranes and subsequently interact with intracellular targets , such as DNA and mitochondria , all eventually resulting in microbial killing [32] , [33] . Considering the antifungal effects and the binding to Candida cells , it was of interest to further study the possible mode of action for HRG on Candida . Electron microscopy demonstrated that HRG caused membrane breaks in Candida cells and release of cytoplasmic components ( Figure 2A ) , effects particularly noted at low pH , where significant extracellular material was detected . The effects were similar to those observed after treatment with the “classical” human AMP LL-37 ( Figure 2A ) . These data suggest that HRG acts on fungal membranes , however they do not demonstrate the exact mechanistic events , as secondary metabolic effects on fungi also may trigger death and membrane destabilization . Therefore , the impermeant dye FITC was used to assess permeabilisation . The results showed that HRG indeed was able to permeabilise Candida membranes ( Figure 2B ) . In line with previous antifungal and binding experiments ( see Figure 1A and 1F ) , the permeabilisation was most apparent at low pH . These results were further substantiated by the use of a liposome model to assess membrane permeabilisation . In correspondence with the effects of HRG on Candida , HRG caused liposome leakage . Compatible with the pH sensitivity observed for HRG , the molecule preferably disrupted ergosterol-containing liposomes at pH 6 . 0 when compared with pH 7 . 4 ( Figure 2C , left panel ) . Notably , ergosterol-containing liposomes , mimicking fungal membranes , were more sensitive than cholesterol-containing ones , mimicking mammalian membranes ( Figure 2C , right panel ) . These results are in agreement with numerous previous findings on the membrane-stabilizing effects of cholesterol [34] , as well as the findings that ergosterol induce less membrane stability in phospholipids than cholesterol [35] . At lower pH , protonation of histidine groups ( pKa for the isolated histidine group is approximately 6 . 5 ) , effectively increases the net charge density of HRG , thus the observed effects are compatible with findings previously reported for histidine-containing consensus peptides and histidine-rich endogenous peptides [36] , [37] . Also noteworthy is that HRG did not display any major conformational changes either at low pH , or in the presence of fungal mannan ( Figure 2D ) or ergosterol-containing phospholipid liposomes ( not shown ) . Hence , large-scale conformational changes appear not to be critical for the antifungal action of HRG . Taken together , the combination of electron microscopy , FITC-studies , and liposome data demonstrates that HRG acts at least in part through membrane disruption , although it is possible that additional intracellular effects of HRG may also contribute to fungal death . It is also notable that the observed effects were most marked and consistent at low pH . At neutral pH , binding ( Figure 1B ) , as well as permeabilization ( Figure 2A and 2B ) was less apparent and these observations reflected the diminished antifungal effects at pH 7 . 4 ( Figure 1A and 1D ) . In order to explore the structure-function relationships of epitopes of HRG , overlapping peptide sequences comprising 20mers ( Figure 3A and Table S1 ) were synthesized and screened , at both neutral and acidic pH , for antifungal activities against C . parapsilosis as well as C . albicans . The experiments identified several antifungal regions . In particular peptides no . 20–24 and 26 , spanning the HRR , displayed a significant antifungal activity against both Candida strains at low pH ( Figure 3B ) . There was a clear correlation with net charge ( at the respective pH ) of the various peptide regions and their observed antifungal activity ( Figure S2 ) . Although intuitively apparent ( Figure 3B ) , the analysis furthermore showed that peptides derived from the HRR were ( with the exception of the K and R-rich peptide no . 27 ) characterized by an increase in net charge at low pH ( Table S1 and Figure S2 ) . In order to further study the importance of the HRR we investigated the activity of recombinant HRG ( rHRG ) and a truncated version ( rHRG1-240 ) , lacking the HRR and C-terminal domain . In contrast to full-length rHRG , truncated rHRG ( 0 . 6 µM ) displayed no activity at pH 5 . 5 against Candida ( Figure 3C ) . Taken together , considering the well-known heparin binding capacity of HRR , its pH dependence , as well as the absence of antifungal activity of rHRG1-240 , it was logical to focus on the HRR of HRG in the subsequent studies of antifungal activity . The HRR contains 12 tandem repeats of five consensus sequences of amino acids , GHHPH [20] , a motif highly conserved among various vertebrate species [27] . To examine the activity of this sequence motif further , a 20-mer peptide ( GHHPH ) 4 [16] , [27] was chosen for further studies . Similar to intact HRG , GHH20 was antifungal against C . parapsilosis and C . albicans , particularly at low pH ( Figure 4A ) . As demonstrated by FACS analysis , Tetramethyl-6-Carboxyrhodamine ( TAMRA ) -labeled GHH20 peptide bound to C . parapsilosis , and in correspondence with the antifungal data , the binding was stronger at pH 5 . 5 when compared to neutral pH ( Figure 4B ) . As illustrated by fluorescence microscopy , TAMRA-labeled GHH20 showed a significant binding to Candida at pH 5 . 5 ( Figure 4C ) . As with the HRG holoprotein , heparin abolished the binding , reflecting the heparin-binding capacity of this region of the HRR [27] . Also in line with the above experiments on fungi , GHH20 preferably disrupted liposomes at pH 6 . 0 , with no significant activity at pH 7 . 4 ( Figure 4D ) . The GHH20 peptide caused liposome leakage within a few hundred seconds ( not shown ) , which contrasted to the significantly slower HRG-induced liposome leakage ( Figure 2B ) , likely a manifestation of the much higher molecular weight of the holoprotein . Again as with intact HRG , CD spectroscopy showed that GHH20 displayed no major conformational changes associated with the histidine protonation at low pH , nor on interaction with phospholipid liposomes or mannan ( not shown ) . Taken together , the GHH20 peptide showed similar characteristics as the holoprotein HRG with respect to activity , binding , and membrane permeabilisation . In order to investigate the functional relevance of the above in vitro activities , we first tested the role of HRG against fungi in relevant physiological “settings” ex vivo . Initial results showed that HRG was detected in blood fractions ( plasma , serum ) and in platelets , also in wound fluid from acute wounds , and chronic leg ulcers ( Figure 5A ) . The latter wound type is characterized by unregulated and excessive proteinase activity leading to degradation of many plasma proteins [38] , [39] . However , compared with plasma and serum HRG , the molecule was not fragmented in this chronic wound fluid fraction ( Figure 5A ) . The protein was also detected in fibrin clots ( Figure 5A ) but not present in seminal plasma . It is of note that the molecule migrated aberrantly in the used gel systems; relative 55–60 kDa in 8% gels ( Tris-Glycine ) and 45–50 kDa in 16 . 5 gels ( Tris-Tricine ) . Identical serum and plasma preparations of HRG were used in the two gel systems , and recombinant HRG showed the same anomalous migration ( not shown ) . In addition to its presence in plasma and other biological fluids , HRG occurs at significant levels in , and binds avidly to , fibrin clots [40] . Coagulation was initiated in normal and HRG-deficient human plasma in the presence of FITC-labeled HRG ( Figure 5B ) . FITC-labeled HRG bound to clots derived from HRG-deficient plasma , and notably , it appeared to be present at clot boundaries , suggesting that it may “coat” the clot surfaces . In clots from normal plasma , no staining was seen , indicative of an inhibition of binding of FITC-HRG by the excess of endogenous HRG ( ∼150 µg/ml ) . Clots , physiologically important “barriers” , formed during hemostasis and infection , could thus constitute a unique milieu with high levels of surface-immobilized HRG . Considering the above results we investigated whether the presence of HRG could reduce the growth of Candida in plasma . Firstly , the growth of C . parapsilosis was investigated in normal human plasma and in plasma depleted of HRG . The results showed that C . parapsilosis multiplied significantly faster in HRG-depleted human plasma ( Figure 5C ) . Analogous results on fungal growth were observed using plasma from mice deficient in HRG ( data not shown ) . It is of note that these results do not exclude the possibility that other antifungal mechanisms may be involved , such as those dependent of complement activation . Furthermore , although the total protein levels ( as determined by the Bradford method ) and contents ( as assessed by SDS-PAGE on 8% gels , not shown ) were the same in depleted plasma ( 51 . 0+/−1 . 2 g/l ) when compared with control plasma ( 51 . 7+/−3 . 3 g/l ) , it cannot be excluded that additional changes of low abundance proteins , induced by passage over Ni-NTA agarose could affect Candida growth . Nevertheless , the observation that similar results were obtained with the mice plasmas points at HRG as the main factor responsible for the partial growth inhibition noted . Furthermore , as demonstrated in Figure 5D , fibrin clots derived from plasma of HRG deficient mice were significantly more prone to infection by C . parapsilosis than clots from wild-type mice , and similar results were obtained with human plasma depleted of HRG when compared with normal plasma ( not shown ) . The observation that clots devoid of HRG showed detectable , although reduced , antifungal activity ( Figure 5D ) suggest the existence of other yet unidentified factors in clots also mediating fungal killing . Nevertheless , the results indicate that HRG contributes to antifungal activity under physiological conditions . To investigate the role of HRG during Candida infection in vivo , we designed a mouse model of intraperitoneal infection with C . albicans . After infection , the body weight of the mice was followed for three days ( Figure 6A ) . Hrg−/− mice showed a significantly increased weight loss at day 1 and 2 ( p = 0 . 02 ) when compared with wild type mice , and the wild type mice regained their initial weight after three days . Blood samples were collected from the animals 2 days post infection , and the fungal load in blood was determined ( Figure 6B ) . A significantly higher amount of Candida cells was detected in the blood of Hrg−/− mice when compared with wild type mice ( p = 0 . 032 ) , indicating that a systemic infection has developed in HRG-deficient mice . In a similar experiment , we determined the ability of the fungi to establish infection in target organs distant from the site of administration . The spleen and kidney were harvested 3 days after initiation of intraperitoneal infection and the fungal load was determined . The results showed significant differences between Hrg−/− mice and the wild type mice; one animal out of 10 in the control group showed fungal load in the spleens and kidneys compared with 8 out of 10 in the Hrg−/− group ( p = 0 . 009 ) ( Figure 6C ) . Histopathological examination of the kidney tissues from Hrg−/− mice showed dense neutrophil infiltrates and notably , Candida cells were visualised by PAS staining in the centre of these infiltrates ( Figure 6D ) . These results show a striking protective role for HRG against invasive Candida infection in vivo .
The key findings in our study are the identification of an antifungal activity of HRG in vivo together with the characterization of possible epitopes of HRG mediating this effect , as well as mechanistic data on HRG targeting of Candida membranes . The results have implications for our understanding of novel antifungal properties of HRG , and demonstrate that HRG constitutes a previously undisclosed natural and antimicrobial defence system . From a structural perspective , several lines of evidence indicate that the HRR is , at least to a significant extent , responsible for the HRG interaction with Candida membranes . Although the 3D structure of HRG has not yet been determined , modelling studies suggest that the HRR of HRG forms a polyproline ( II ) helical structure with numerous histidines . At physiological pH , HRG is net negatively charged ( pI 6 . 45 ) . However , due to its high content of histidine residues ( ∼13% ) , which are concentrated to the HRR , it can acquire a positive charge by protonation [20] , [41] , and this in turn likely facilitates the interactions between HRG and Candida . These results were substantiated by the finding that a region of HRG containing the motif sequence GHHPH , was antifungal , and that low pH enhanced this activity . The high conservation of this sequence among vertebrates likely reflects its importance for membrane interactions of HRG [27] . However , as evident in Figure 3B , there are also other antifungal regions in the protein , active irrespective of pH in the interval investigated , an observation compatible with the antifungal activity of HRG detected at neutral pH . It should be pointed out however , that the peptide data do not reflect the complex structure-activity relationships of the holoprotein . Although the CD experiments did not detect any major conformational changes upon interaction with liposomes or polysaccharides , it cannot be ruled out that conformational changes mediated by HRR interactions with intact fungal cells lead to the exposure of additional antimicrobial epitopes in the molecule . Nevertheless , a recombinant and truncated variant of HRG , lacking the histidine-rich and C-terminal domains , was not active against Candida , pointing to the HRR as an important , possibly the most important , effector of HRGs antifungal effects . Many histidine-rich AMPs are known , among these the clavanins [36] , histatins , and calprotectin [42] . We have previously shown that the antibacterial effects in vitro of various histidine-rich peptides , both consensus motifs and peptides derived from domain 5 of HMW kininogen [17] and from HRG [27] are enhanced at low pH or upon addition of Zn2+ . Others have reported that the antimicrobial activity of clavanins were substantially increased in low pH as compared with neutral pH [36] . Furthermore , the antimicrobial effect of histatin 5 is enhanced at low pH [43] , and histidine-rich variants of magainin , the LAH4-peptides , were recently shown to have increased antibacterial activity in low pH compared to neutral pH [37] . Taken together , the pH dependent activity of HRG is thus comparable to other histidine-rich proteins and peptides , and provides an additional link between pH sensitive AMPs and HRG . However , contrasting to histatins , which translocate through Candida membranes , bind mitochondria , and induce cell death by non-lytic ATP-release [44] , HRG acts directly on fungal membranes . Many AMPs are generated by proteolysis of larger , and non-antimicrobial holoproteins . For example , the cathelicidin LL-37 is released from hCAP18 , and other AMPs are proteolytically generated from complement factor C3 and high molecular weight kininogen [3] , [11] , [12] , [14]–[17] . Considering that intact HRG is antifungal , proteolysis of this molecule does not appear to be needed for activity . It is of note that like HRG , several antimicrobial proteins are antimicrobial per se , including bacterial permeability increasing protein , serprocodins such as proteinase 3 , elastase and heparin binding protein , as well as lactoferrin [27] , [45] . However , it is also described that antibacterial proteins , such as bacterial permeability increasing protein and lactoferrin , may give rise to peptides exerting antibacterial activities [46] , [47] . Likewise , it has been shown that HRG may be degraded by plasmin [48] , as well as in patients undergoing thrombolytic therapy [49] and bioactive fragments of HRG are involved in antiangiogenesis [26] , [41] . Thus , although a major fragmentation of HRG was not observed in this work , e . g . , in wound fluid and after binding to fibrin , it is likely that degradation of HRG may occur at sites of high proteolysis and plasmin activity . Indeed , the finding that the HRG-derived peptide GHH20 , as well as numerous other other 20mer peptides were antifungal , and as particularly noted for HRR-derived peptides , exhibiting a similar pH dependence as HRG , exemplifies that the holoprotein is not a prerequisite for antifungal action . Clearly , such possibilities need to be addressed in future studies . As previously mentioned , HRG is involved in various aspects of angiogenesis , coagulation , and fibrinolysis [20] , reflecting its interactions with ligands such as heparin , plasminogen , fibrinogen , and thrombospondin . Additionally , it acts as an opsonin by bridging FcyRI receptors on macrophages to DNA on apoptotic cells , stimulating phagocytosis [50] , and modulates the binding of IgG and immune complexes to FcγRI [50] . Considering these multiple roles , it is likely that HRG binding to microbial surfaces could induce additional “down-stream” effects , such as modulation of plasminogen activity and phagocytosis . The history of “classic” AMPs have shown that these molecules , initially believed to take part merely in direct microbial killing , have extended their roles into the ability to act as chemokines and to induce chemokine production leading to recruitment of leukocytes , promotion of wound healing , and an ability to modulate adaptive immunity [51] . Indeed , as interest in the in vivo functions of host defence peptides is increasing , it is important to consider the direct antimicrobial and immunomodulatory properties observed . Nevertheless , several findings in this study unequivocally demonstrate that HRG , like many AMPs , acts directly on microbes . Thus , in addition to the antifungal in vitro data , the enhanced fungal growth in HRG-deficient plasma , as well as the finding that Candida was detected at higher levels in blood of Hrg−/− animals , indicates a direct antifungal action of the molecule . It is also interesting to note that these HRG deficient animals have also been shown to be more susceptible to Streptococcus pyogenes infection ( Shannon et al , unpublished results ) . However , considering both AMP and HRG multifunctionality in vitro as well as in vivo , it may be envisaged that additional actions , resulting in the observed antifungal effects , will likely be revealed . All of these effects may be dependent on binding of HRG to microbes and subsequent interactions with cells ( e . g . , neutrophils and macrophages ) in different compartments ( e . g . , skin , internal organs , and blood ) . In this respect , the pH dependence of HRG is particularly interesting and relevant . It is well known that infection foci , including abscesses , are characterized by low pH levels reaching as low as pH 5 , due to increased anaerobic metabolism and lactate production , as well as leukocyte mediated oxidative burst and subsequent acidification [52] . The capacity of HRG to kill Candida at these pH levels and the corresponding increase in salt-resistance at low pH suggest that HRG could target infection foci , resulting in a physiologically relevant concentration and localization of antifungal activity . As previously mentioned , HRG's opsonising activity could hypothetically lead to enhanced phagocytosis . Although it remains to be investigated , such localisation of antifungal activity to endosomal compartments , where acidification could result in enhanced HRG-mediated killing of phagocytosed fungi , could serve as an effective way of eliminating invading Candida cells at sites of tissue inflammation without releasing potentially toxic microbial components . Again hypothetically , genetic deficiencies of HRG or acquired functional defects could provide interesting clues with respect to functional roles of HRG . In some patients , reduced levels of HRG are associated with a thrombophilic phenotype , indeed compatible with the phenotype observed in Hrg−/− mice , which had a shorter prothrombin time [22] . As these patients still have ∼20–50% of normal levels of HRG , the human phenotype of complete absence of HRG remains , however unknown . Although patients with low levels of HRG have not been reported to be more prone to infections , it must be remembered that examples from deficiencies of particular innate immune proteins , e . g . , complement and mannose-binding lectin , illustrate that even homozygous deficiency and a complete absence of a particular innate immune molecule may give rise to surprisingly mild symptoms . For example , patients with mannose-binding lectin deficiencies are normally not at risk of developing infections unless compromised by immune suppression or severe disease [53] . In this context , it is particularly interesting that antibodies against HRG have been detected in patients with antiphospholipid syndrome [54] , a disease associated with thrombodiathesis and systemic lupus erythematosus . Notably , the latter disease is associated with an increased risk for opportunistic infections , including Candida [55] . Taken together , and considering the role of HRG in innate immunity , it should be of interest to study potential associations between functional inactivation ( s ) or deficiencies of HRG as well as genetically determined differences , in relation to the occurrence of infections . During the last three decades , research on innate immune molecules has demonstrated the significance of the innate immune system for prevention of invasion by microbes at biological boundaries . Previous studies have emphasized that various molecules , such as “classic” AMPs , complement factors , and cytokines , bridge between innate and adaptive immunity . The present work adds another significant component to this family of molecules , the plasma protein HRG .
The peptides GHH20 ( GHHPHGHHPHGHHPHGHHPH ) and histatin 5 ( DSHAKRHHGYKRKFHEKHHSHRGPY ) were synthesized by Innovagen AB ( Lund , Sweden ) , and were of >95% purity . The purity and molecular weight was confirmed by MALDI-TOF MS analysis ( Voyager , Applied Biosystems ) . 20-mer synthetic peptides ( PEP-screen ) spanning the sequence of HRG ( Table 1 ) were obtained from Sigma-Genosys ( St Louis , MO ) . Polyclonal rabbit antibodies against GHH20 and TAMRA-labeled GHH20 were from Innovagen AB ( Lund , Sweden ) . HRG was FITC-labeled using the FluoroTag FITC Conjugation Kit ( Sigma , St Louis , MO ) . Human serum and plasma were collected from healthy volunteers . Sterile wound fluids were obtained from surgical drainages after mastectomy . The use of human wound fluid was approved by the Ethics Committee at Lund University ( LU 708-01 ) . Seminal plasma was collected at the Fertility Center at Malmö University Hospital , Sweden . The GenBank ( http://www . ncbi . nlm . nih . gov/Genbank/index . html ) accession number of human histidine-rich glycoprotein is NP_000403 . The fungi Candida parapsilosis BD 17837 and Candida albicans BD 1060 were clinical isolates . C . parapsilosis ATCC 90018 , C . albicans ATCC 90028 , Candida glabrata ATCC 90030 , and Candida krusei ATCC 6258 isolates were from the American Type Culture Collection ( ATCC , Rockville , MD ) . Serum HRG was purified using nickel-nitrilotriacetic acid ( Ni-NTA ) agarose as described before [27] . The concentration of the protein was determined using the Bradford method [56] . Recombinant His-tagged HRGP and truncated version of HRG ( HRG1-240 ) , containing amino acids 1-240 was produced and purified as previously described [26] , [27] . Plasma , serum , wound fluids , seminal plasma ( 1 µl ) , and platelets ( fluid from 1×103 cells , disrupted by freeze thawing ) were electrophoresed on 8% SDS-polyacrylamide ( SDS-PAGE ) gel or an 16 . 5% Tris-tricine gel and transferred to a nitrocellulose membrane ( Hybond-C , GE Healthcare BioSciences , Little Chalfont , UK ) [57] . The membrane was incubated in 3% skimmed milk in 10 mM Tris , 0 . 15 M NaCl , pH 7 . 4 for 1 h at room temperature , followed by incubation for 1 h with rabbit polyclonal antibodies against GHH20 ( diluted 1:1000 in the same buffer ) . The membrane was washed 3 times , and incubated again for 1 h with horseradish peroxidase-conjugated secondary swine anti rabbit antibodies diluted 1:1000 ( Dako , Carpinteria , CA ) . The image was developed using the ECL system ( Amersham Biosciences ) . Human plasma was subjected to a Ni-NTA agarose gel . The eluent ( plasma completely depleted of HRG ) was collected and used to form clots . Hrg−/− and C57BL/6 ( wild type ) mice [22] were used for preparation of fibrin clots from plasma of the respective animals . Plasma deficient of HRG and normal plasma were incubated with a total concentration of 10 mM Ca2+ in eppendorf tubes at 37°C over night . Clots were washed three times and then stored in 10 mM 2-Morpholinoethanesulfonic acid ( MES ) , pH 5 . 5 . Clots ( ∼0 . 04g ) were used in viable count experiments . To investigate the localization of HRG in fibrin clots , human plasma and HRG-deficient plasma were incubated with 10 µl FITC-labeled HRG ( 0 . 4 mg/ml ) and then processed as before in the presence of 10 mM Ca2+ over night . The clots were then washed in distilled water and mounted on slides using Dako mounting media ( Dako ) . C . parapsilosis , C . albicans , C . glabrata and C . krusei were grown to mid-logarithmic phase in Todd-Hewitt ( TH ) medium ( Becton and Dickinson , Maryland , USA ) at 27°C and washed in 10 mM Tris , pH 7 . 4 or 10 mM MES , pH 5 . 5 . For dose-response experiments , purified HRG or GHH20 ( 0 . 03–6 µM ) were incubated with 1×105 C . parapsilosis ATCC 90018 or C . albicans ATCC 90028 for 2 h at 37°C in 10 mM , Tris , pH 7 . 4 or in 10 mM MES-buffer , pH 5 . 5 , plated on Sabouraud dextrose broth ( Becton and Dickinson ) agar , and incubated 48 hours at 27°C , whereafter the number of cfu was determined . In order to investigate the antifungal activity of HRG in presence of salt , 6 µM HRG were incubated with 1×105 C . parapsilosis ATCC 90018 for 2 h at 37°C in 10 mM , MES , pH 5 . 5 containing 0 , 25 , 50 , 100 or 150 mM NaCl , plated and the number of cfu was determined . In kinetic experiments , 0 . 3 and 3 µM HRG were incubated with C . parapsilosis ATCC 90018 for 5 , 15 , 30 , 60 , or 120 minutes in 10 mM MES , pH 5 . 5 , plated and the number of cfu was determined . For determination of the effect of HRG on various Candida strains , HRG ( 3 µM ) was incubated with C . parapsilosis ATCC 90018 or BD 17837 , C . albicans ATCC 90028 or BD 1060 , C . glabrata ATCC 90030 or C . krusei ATCC 6258 in 10 mM Tris , pH 7 . 4 or 10 mM MES , pH 5 . 5 , plated and number of cfu determined . Truncated and full length recombinant HRG , 0 . 6 µM rHRG , or rHRG1-240 were incubated with C . parapsilosis ( 1×105 ) for two hours and then plated and number of cfu determined . To investigate the in vitro antifungal activity of HRG , normal or HRG-deficient fibrin clots ( ∼0 . 04g ) were incubated with C . parapsilosis ATCC 90018 for 2 h in 10 mM MES , pH 5 . 5 , plated and number of cfu were determined . For inhibition studies , 0 . 3 µM HRGP were incubated with C . parapsilosis ( 1×105 ) in 10 mM MES , pH 5 . 5 , in presence or absence of heparin ( 50 µg ) for two hours and then plated and number of cfu was determined . In all experiments , 100% survival was defined as total survival of fungi in the same buffer and under the same conditions in absence of peptide , protein , or clots . The p-values were determined using Kruskall-Wallis one-way ANOVA analysis . Radial diffusion assay ( RDA ) was performed essentially as described earlier [58] . C . parapsilosis ATCC 90018 and C . albicans ATCC 90028 were grown to midlogarithmic phase in TH-medium , and then washed with distilled water . 4×106 colony forming units was added to 5 ml of the underlay agarose gel ( 0 . 03% ( w/v ) trypticase soy broth ( TSB ) , 1% ( w/v ) low electroendosmosis type agarose ( Sigma ) , 0 . 02% ( v/v ) Tween 20 ( Sigma ) . The buffers used in the underlay gels were 10 mM Tris , pH 7 . 4 or 10 mM MES , pH 5 . 5 . The underlay gel was poured into an 85-mm Petri dish . After agarose solidification , wells of 4 mm in diameter were punched , and 6 µl of peptide solution was added to each well . Buffers were used as a negative control . Plates were incubated at 28°C for 3 h to allow diffusion of the peptides . The underlay gel was then covered with 5 ml of molten overlay . Antimicrobial activity of a peptide is visualized as a zone of clearance around each well after 18–24 h of incubation at 28°C . Peptides were tested in concentrations of 100 µM . C . parapsilosis ( 1×105 cfu ) were incubated with 0 . 6 µM HRG in 50 µl 10 mM MES , pH 5 . 5 , with or without heparin ( 50 µg/ml ) for 2 h at 37°C , centrifuged and the pellet was washed three times in 10 mM MES , pH 5 . 5 . The pellet and the supernatant were resuspended in SDS sample buffer , electrophoresed ( 8% SDS-PAGE ) , and then transferred to a nitrocellulose membrane . Western blotting was performed as above . C . parapsilosis ATCC 90018 fungi were grown in TH medium at 27°C to mid-logarithmic phase . The fungi were washed in 10 mM Tris , pH 7 . 4 , and resuspended in the same buffer . C . parapsilosis ( 2×106/ ml ) were incubated with 1 µl of TAMRA-labeled GHH20 ( 2 mg/ml ) in 10 mM MES , pH 5 . 5 , with or without heparin ( 50 µg/ml ) , left standing for 5 minutes on ice , and then washed twice in 10 mM Tris , pH 7 . 4 . Fungi were fixed with 4% paraformaldehyde by incubation on ice for 15 minutes and in room temperature for 45 minutes . The fungi were then applied onto Poly-L-lysine coated cover glass and after an incubation time of 30 minutes , finally mounted on slides using Dako mounting media ( Dako , Carpinteria , CA ) . In order to assess permeabilisation , C . albicans ATCC 90028 ( 2×106 cfu ) were incubated with HRG or LL-37 ( both at 10 µM ) in 10 mM Tris , pH 7 . 4 or 10 mM MES , pH 5 . 5 for 30 minutes at 37°C . Samples were transferred to Poly-L-lysine coated cover glass and incubated for 45 minutes at 37°C , washed and 2 µg of FITC were added in a volume of 200 µl , and incubated for 30 minutes at 30°C , washed and then fixed as above . Samples were visualized using a Nikon Eclipse TE300 inverted fluorescence microscope equipped with a Hamamatsu C4742-95 cooled CCD camera , a Plan Apochromat 100X objective and a high N . A . oil condenser . C . parapsilosis ATCC 90018 were grown in TH medium at 37°C to mid-logarithmic phase . The fungi were washed in 10 mM Tris , pH 7 . 4 or 10 mM MES , pH 5 . 5 , and resuspended in the same buffer . HRG or LL-37 ( 10 µM ) was incubated with C . parapsilosis ( 20×106 cfu ) for two hours in a total volume of 10 µl in Tris buffer , pH 7 . 4 or in MES buffer , pH 5 . 5 . Samples of C . parapsilosis fungi suspensions were adsorbed onto carbon-coated copper grids for 1 min , washed briefly on two drops of water , and negatively stained on two drops of 0 . 75 % uranyl formate . The grids were rendered hydrophilic by glow discharge at low pressure in air . Specimens were observed in a Jeol JEM 1230 electron microscope operated at 60 kV accelerating voltage . Images were recorded with a Gatan Multiscan 791 CCD camera . C . parapsilosis ATCC 90018 were grown in TH medium at 27°C to mid-logarithmic phase . The fungi were washed in 10 mM Tris , pH 7 . 4 or 10 mM MES , pH 5 . 5 and resuspended in the same buffer . C . parapsilosis ( 5×107 in a total volume of 0 . 5 ml ) were incubated with 10 µl of FITC-labeled HRG ( 0 . 4 mg/ml ) or 10 µl TAMRA-labeled GHH20 ( 2 mg/ml ) in 10 mM Tris , pH 7 . 4 or in 10 mM MES , pH 5 . 5 , let stand for 5 minutes on ice and then washed in 10 mM Tris , pH 7 . 4 . The cells were fixed with 4% paraformaldehyde by incubation on ice for 15 minutes and in room temperature for 45 minutes . Flow cytometry analysis was performed using a FACS-Calibur flow cytometry equipped with a 15 mW argon laser turned a 488 mm ( Becton-Dickinson , Franklin Lakes , NJ ) . The fungal population was selected by gating with appropriate settings of forward scatter ( FSC ) and sideward scatter ( SSC ) . The FL1 fluorescence channel ( λem = 530 nm ) was used to record the emitted fluorescence of FITC , and the FL3 fluorescence channel ( λem = 585 nm ) was used to record the emitted fluorescence of Texas red . Dry lipid films were prepared by dissolving dioleoylphosphatidylcholine ( 1 , 2-dioleoyl-sn-Glycero-3-phoshocholine , >99% purity , Avanti Polar Lipids , Alabaster , AL ) ( 60 mol% ) and either ergosterol or cholesterol ( both >99% purity , Sigma , St Louis , MO ) ( 40 mol% ) , and then removing the solvent by evaporation under vacuum overnight . Subsequently , buffer ( 10 mM Tris , pH 7 . 4 ) was added together with 0 . 1 M carboxyfluorescein ( CF ) ( Sigma , St Louis , MO ) . After hydration , the lipid mixture was subjected to eight freeze-thaw cycles consisting of freezing in liquid nitrogen and heating to 60°C . Unilamellar liposomes , of about Ø140 nm were generated by multiple extrusions through polycarbonate filters ( pore size 100 nm ) mounted in a LipoFast miniextruder ( Avestin , Ottawa , Canada ) at 22°C . Untrapped CF was then removed by two gel filtrations ( Sephadex G-50 ) at 22°C , with Tris buffer as eluent . CF release was determined by monitoring the emitted fluorescence at 520 nm from liposome dispersions ( 10 mM lipid in 10 mM Tris ) . An absolute leakage scale was obtained by disrupting the liposomes at the end of the experiment through addition of 0 . 8 mM Triton X100 ( Sigma , St Louis , MO ) , causing 100% release and dequenching of CF . Although calcein is frequently used for pH-dependent leakage studies , the high charge of this dye has been noted to influence its leakage behaviour in the presence of highly cationic peptides [59] . Instead , therefore , CF was used as a leakage marker at both pH 6 . 0 and 7 . 4 , however , avoiding pH-dependent fluorescence effects through neutralization prior to probing the limiting leakage in case of pH 6 . 0 leakage . Throughout , a SPEX-fluorolog 1650 0 . 22-m double spectrometer ( SPEX Industries , Edison , NJ ) was used for the liposome leakage assay . Measurements were performed at 37°C . The CD spectra of the peptides in solution were measured on a Jasco J-810 Spectropolarimeter ( Jasco , U . K . ) . Measurements were performed at 37°C in a 10 mm quartz cuvet under stirring and the effect on protein/peptide secondary structure monitored in the range 200–260 nm . The background value , detected at 250 nm , was subtracted , and signals from the bulk solution were corrected for . The secondary structure was monitored at a concentration of 0 . 25 µM of HRG in buffer , in the presence of liposomes ( lipid concentration 100 µM ) , and in the presence of mannan from Saccharomyces cerevisiae ( 0 . 02 wt%; Sigma-Aldrich , St . Luis , USA ) . C . parapsilosis ATCC 90018 were grown in TH medium at 27°C to mid-logarithmic phase . The fungi were washed in 10 mM MES , pH 5 . 5 and resuspended in the same buffer . C . parapsilosis ( 2×107 ) cfu in a total volume of 10 µl was added to 50 µl of human normal plasma or HRG-depleted plasma ( eluent from Ni-NTA agarose gel ) , and incubated for 0 , 4 , 8 or 18 hours at 27°C and then plated and number of cfu determined . The original knockout mice 129/B6-HRGtm1wja1 were crossed with C57BL/6 mice ( Taconic ) for 14 generations to obtain uniform genetic background . These HRG-deficient mouse strain was called B6-HRGtm1wja1 following ILAR ( Institute of Laboratory Animal Resources ) rules . Wildtype C57BL/6 control mice and C57BL/6 Hrg−/− mice ( 8–12 weeks , 27+/−4g ) were bred in the animal facility at Lund University . C57BL/6 Hrg−/− , lacks the translation start point of exon 1 of the Hrg gene [22] . Animals were housed under standard conditions of light and temperature and had free access to standard laboratory chow and water . In order to study Candida dissemination , C . albicans ATCC 90018 were grown to midlogarithmic phase , washed and diluted in PBS , pH 7 . 4 . Two hundred and fifty µl containing 1×109 cfu was injected intraperitoneally into C57BL/6 or C57BL/6 Hrg−/− mice , divided into weight and sex matched groups . The animals were sacrificed 48 hours post infection , and blood was collected by cardiac puncture . The number of cfu was determined by viable count . In order to study fungal dissemination to target organs , the mice were infected as previously described and three days later the spleen and kidney were harvested on ice . Representative animals were sacrificed three days post infection and the kidneys were removed into 4% formalin . The tissues were embedded in paraffin , sectioned and stained with Hematoxylin and eosin ( H&E ) and with Periodic acid-Schiff ( PAS ) . | It has been estimated that humans contain about 1 kg of microbes , an observation that reflects our coexistence with colonizing microbes such as bacteria and fungi . The fungal species Candida is present as a commensal at mucosal surfaces and on skin . Although it may cause life-threatening infections , such as sepsis , particularly in immunocompromised individuals , it seldom causes disease in normal individuals . In order to control our microbial flora , humans as well as virtually all life forms are armoured with various proteins and peptides that comprise integral parts of our innate immune system . Here we describe a new component in this system; histidine-rich glycoprotein ( HRG ) , an abundant plasma protein . We show , using a combination of microbiological , biochemical , and biophysical methods , that HRG exerts a potent antifungal activity , which is mediated via a histidine-rich region of the protein , and targets ergosterol-rich membrane structures such as those of Candida . HRG killed Candida both in plasma as well as when incorporated into fibrin clots . In mouse infection models , HRG was protective against systemic infection by Candida , indicating a novel antifungal role of HRG in innate immunity . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases/fungal",
"infections",
"immunology/innate",
"immunity",
"dermatology/skin",
"infections"
] | 2008 | Histidine-Rich Glycoprotein Protects from Systemic Candida Infection |
The splicing regulator Polypyrimidine Tract Binding Protein ( PTBP1 ) has four RNA binding domains that each binds a short pyrimidine element , allowing recognition of diverse pyrimidine-rich sequences . This variation makes it difficult to evaluate PTBP1 binding to particular sites based on sequence alone and thus to identify target RNAs . Conversely , transcriptome-wide binding assays such as CLIP identify many in vivo targets , but do not provide a quantitative assessment of binding and are informative only for the cells where the analysis is performed . A general method of predicting PTBP1 binding and possible targets in any cell type is needed . We developed computational models that predict the binding and splicing targets of PTBP1 . A Hidden Markov Model ( HMM ) , trained on CLIP-seq data , was used to score probable PTBP1 binding sites . Scores from this model are highly correlated ( ρ = −0 . 9 ) with experimentally determined dissociation constants . Notably , we find that the protein is not strictly pyrimidine specific , as interspersed Guanosine residues are well tolerated within PTBP1 binding sites . This model identifies many previously unrecognized PTBP1 binding sites , and can score PTBP1 binding across the transcriptome in the absence of CLIP data . Using this model to examine the placement of PTBP1 binding sites in controlling splicing , we trained a multinomial logistic model on sets of PTBP1 regulated and unregulated exons . Applying this model to rank exons across the mouse transcriptome identifies known PTBP1 targets and many new exons that were confirmed as PTBP1-repressed by RT-PCR and RNA-seq after PTBP1 depletion . We find that PTBP1 dependent exons are diverse in structure and do not all fit previous descriptions of the placement of PTBP1 binding sites . Our study uncovers new features of RNA recognition and splicing regulation by PTBP1 . This approach can be applied to other multi-RRM domain proteins to assess binding site degeneracy and multifactorial splicing regulation .
Alternative splicing of pre-mRNA commonly determines the protein output of mammalian genes , with most genes generating multiple mRNA and protein products [1] . A typical alternative exon is affected by multiple pre-mRNA binding proteins that may either enhance or repress splicing [2] . The expression and activity of these splicing regulatory proteins can vary with development , cell type , or cellular stimulus [3] . This complex combinatorial regulation can be seen in the conserved sequences within and surrounding alternative exons , which generally contain the binding sites for many different regulators . These sequences make up what is sometimes called the splicing code as they determine where and when the exon is spliced into an mRNA [4] , [5] , [6] , [7] . Such a code should allow the development of models that predict exon regulation based solely on the RNA binding affinity of the many regulatory proteins and their other interactions . However , this is not currently feasible , in part due to our incomplete understanding of RNA recognition by the splicing regulators and their mechanisms of action . Whole-transcriptome crosslinking methods for individual proteins in vivo are allowing the identification of large numbers of protein/RNA interaction sites [8] , [9] , [10] , [11] . These data can be overlapped with functional data on splicing to identify possible direct target exons for particular proteins [12] , [13] , [14] , [15] . However , there are limitations in the interpretation of these data . Crosslinking efficiency can vary between different proteins and between individual binding sites , making it difficult to relate the crosslinking signal to the actual binding affinity . These signals are also dependent on the expression of the bound RNA , and since these data are generated one tissue or cell type at a time it is not always feasible to extend the results from one setting to a new cell type or point in development . It would be extremely useful to be able to scan for binding affinity across the complete transcriptome and to predict exon targets in tissues that have not yet been subjected to experimental analysis . Splicing regulatory proteins commonly contain multiple RRM or other RNA binding domains , with each domain recognizing a short element of a few nucleotides [2] , [16] . Subtle variation in the optimal binding element of each domain and flexible peptide linkers between them allow for significant degeneracy within high affinity binding sites . Although the short sequence motifs that are common to a set of binding sites are readily identified , these likely constitute only a portion of a full high affinity site . To rank binding sites and assess their finer structures , we need an approach to search for clusters of these short motifs and to score for binding affinity . The Polypyrimidine tract binding protein 1 ( PTBP1 ) is a widely studied splicing regulatory protein [17] , [18] . PTBP1 is known to repress the splicing of a large number of exons by binding in their adjacent introns or within the exons themselves . PTBP1 is down regulated in differentiating neurons and muscle cells to allow inclusion of PTBP1 repressed exons during development of these tissues [19] , [20] , [21] . In neurons the loss of PTBP1 is accompanied by the up-regulation of the homologous protein PTBP2 [17] , [20] , [22] . PTBP2 has similar binding properties to PTBP1 and represses some of the same exons [23] . Other exons are more sensitive to PTBP1 than PTBP2 and are induced to splice when PTBP2 replaces PTBP1 in early neurons [24] . PTBP1 contains four RRM domains that recognize short pyrimidine elements [25] . Flexible linkers separate RRM domains one and two , and domains two and three . RRM domains three and four interact through a hydrophobic interface that position their RNA binding surfaces on opposite faces of the two-domain structure . This orientation requires that the RNA elements interacting with the structure be separated by an RNA loop [26] . The structure of each of the PTBP1 RRM domains has been solved in complex with the hexanucleotide , CUCUCU [25] . These structures show each domain binding a nucleotide triplet with some additional contacts , and making similar base specific interactions with CU or UC dinucleotides . Other sequences can likely make different base specific contacts , and the optimal elements for each domain are not known . Moreover , the flexible linkers separating some of the RRM domains and the requirement for a gap between elements simultaneously bound to domains three and four allow for substantial degeneracy in PTBP1 binding sites . This degeneracy and the lack of understanding of the sequence features that contribute to binding affinity have made it difficult to identify PTBP1 binding sites based on sequence alone , and to assess which sequences surrounding an exon might contribute to PTBP1 regulation . Experiments with model substrates indicate that a single high affinity PTBP1 binding site placed upstream of an exon , or within it , can repress splicing [27] . However , strong repression of an efficiently spliced exon requires an additional binding site either within the exon or downstream from an exon with an upstream high affinity site [17] , [27] , [28] . PTBP1 is also known to enhance the splicing of certain exons [13] , [19] , [20] . The properties of these exons and how they differ from those that are repressed by PTBP1 are unclear , with different studies coming to different conclusions [13] , [19] . An analysis of CLIP data in HeLa cells found that PTBP1 sites near the adjacent constitutive exons could enhance the inclusion of an alternative exon between them [13] . In contrast , examination of exons whose splicing was reduced by double knockdown of Ptbp1 and Ptbp2 found that they frequently had binding sites immediately downstream [19] , whereas splicing repression often involved upstream binding sites: a pattern observed for other splicing regulators . These results are not mutually exclusive . It is possible that the two groups examined different subsets of the many exons regulated by PTBP1 , and that the protein may show additional patterns of protein binding adjacent to its target exons . In this study we sought to understand the sequence features that determine RNA binding by PTBP1 and to examine how they are combined in exons that are targeted by the protein . We first developed a statistical model of PTBP1 binding sites that identifies new features of RNA recognition by the protein . This binding model was then applied to the assessment of exon regulation by PTBP1 across the transcriptome .
To examine the interactions of PTBP1 across many binding sites , we used a set of PTBP1-bound sequences identified by crosslinking immunoprecipitation ( CLIP ) [13] . PTBP1 has four RRMs separated by linker peptides , with each RRM recognizing a pyrimidine triplet . In previous studies we found that a minimal high affinity binding site for the protein extended across 25 to 30 nucleotides , about the average size of the CLIP clusters ( 29 nt ) [27] . Given the triplet recognition and the need for spacers between the direct RRM contacts , it is unlikely that every nucleotide within a CLIP cluster makes a direct base-specific contact with the protein or otherwise contributes to binding affinity . This information about direct binding is hidden in the examination of a CLIP tag , but should affect the triplet frequencies within the entire set of tags . We designed a two-state Hidden Markov Model ( HMM ) based on triplets to assess whether triplets would segregate into two states and whether these two states differed in their PTBP1 binding or non-binding potential . The 48 , 604 CLIP clusters from the human transcriptome were extracted and used to train the HMM ( Figure 1A ) [29] , [30] . This training defined two states showing distinctly different triplet distributions ( Figure 1B ) . Pleasingly , all of the pyrimidine triplets segregated into State 1 . We called this state the PTBP1 binding state , as we confirm below . We found that 20 triplets have higher probabilities to be seen in the PTBP1 binding state . All triplets containing only pyrimidines were included in this 20-triplet set ( Figure 1B ) , with the top-scoring triplet UCU showing the alternating C and U nucleotides seen in many characterized PTBP1 binding sites . Interestingly , multiple triplets containing G residues are also preferred in State1 ( Figure 1B ) . These triplets often contain U residues as the other nucleotides . Some of these triplets , such as UGU , have output ( emission ) probabilities in State 1 that are similar to pyrimidine triplets , presumably also making them predictive of PTBP1 binding . In contrast , triplets containing A residues , even if the other two nucleotides are pyrimidines , were all preferred by the non-PTBP1 binding State 2 . These results indicate that PTBP1 is not strictly pyrimidine specific . At least one of its RRM domains can presumably make specific contacts with G residues . On the other hand , all A containing triplets have modest positive emission probabilities for state 2 and are likely to be either neutral or to inhibit PTBP1 binding . We next tested the HMM scoring , which strongly weights the triplets from state 1 over state 2 , for prediction of PTBP1 binding . We performed cross validation experiments on the Hela CLIP dataset . A background dataset was generated using ten randomly picked sequences from each gene identified as containing a CLIP cluster . Applying the model to this data set gave us a distribution of scores that was compared to scores generated by subsets of the CLIP clusters removed from the training set prior to training . As shown in Figure S1 , sequences from subsets of the CLIP clusters scored significantly higher than background . We also tested our model on an independent iCLIP dataset from human embryonic stem cells ( ESC ) ( Figure S2 ) . Unlike standard CLIP , iCLIP tags define the probable crosslink site as being the 5′ terminus of the tag . We used a Viterbi algorithm to predict the most probable state path predicted by the PTBP1 HMM model for each iCLIP tag . Defining triplets from the State1 ( PTBP1 binding ) and triplets from State 2 ( nonbinding ) , we found that the frequency of predicted binding triplets is highly enriched in the iCLIP cluster regions and peaks precisely at the crosslink site . This indicates that State 1 probability is highly associated with PTBP1 crosslinking in vivo . To more quantitatively assess the relationship between the HMM score and RNA binding , we applied the trained model to a set of 100 , 000 random 69 nucleotide sequences . This length allows for one hexanucleotide binding site for each of the four RRMs with 15 nucleotide gaps , the minimum gap required for simultaneous binding by RRMs 3 and 4 [25] , [26] . The scores are calculated as a log-odds ratio of the probabilities of the sequence having been generated by the HMM over a background model that assigns equal probability to all triplets . The random sequences generated a distribution of scores that was used to normalize the binding scores , with the average score for random sequence set to zero , and the z-score defined as the deviation from the average as shown in Figure S3A [29] . Thus a sequence with a z-score of 2 . 74 is 2 . 74 standard deviations from the average ( empirical p-value = 0 . 005 ) , and is predicted to be a significantly stronger binder than the average sequence ( 500 of the 100 , 000 random sequences have scores equal or greater than this sequence ) . A negative z-score is predicted to bind less well than the average sequence . We isolated thirteen sequences from the mouse transcriptome that exhibited a range of scores from −2 . 62 to +4 . 40 ( Figure 2A ) . These were transcribed in vitro and subjected to electrophoretic mobility shift assay to measure binding to recombinant PTBP1 ( Figure 2B; Figure S3B ) . Sequences yielding negative scores all failed to bind PTBP1 within the protein concentration range tested , with the exception of probe 4 , which bound weakly , below the level that would allow measurement of an affinity constant . Positive scoring sequences all yielded PTBP1 bound complexes that were assayable by gel shift to derive apparent binding affinities . The apparent Kds of these RNAs showed a very strong negative correlation with their binding score from the model ( Pearson correlation coefficient = − 0 . 9 ) , where a higher score predicts a lower Kd and hence a higher affinity ( Figure 2A ) . Thus , the scoring system performed very well in predicting PTBP1 binding affinity . Two sequences ( probes 9 and 11 ) showed variable binding that shifted their Kd's slightly off the fitted curve relating z-score to Kd . These may have secondary structures that reduce binding affinity thus increase their apparent Kd . To look at this , we examined the predicted structure of each probe using the RNA fold program [31] . Probes 9 and 11 did not show an overall free energy of folding substantially lower than other RNAs . However , it is difficult to rule out that they contain a local structure that sequesters some key feature for PTBP1 recognition . In addition to the background model using uniform triplet frequencies , we also tested control sequence sets using different nucleotide frequencies ( Figure S4 ) . Control sets that maintain the mono or dinucleotide frequencies of the PTBP1 CLIP tags while shuffling the triplet frequencies did not perform well . This is not surprising because these sequences are highly skewed in nucleotide content and the shuffling does not change the triplet frequencies dramatically . We also tested a background model based on random sequences selected from genes containing PTBP1 CLIP clusters ( ten sequences from each gene ) . Like the random dataset , this background model generated scores that predicted affinity reasonably well . However , it did generate negative scores for a couple of probes that are shown to bind ( data not shown ) . Thus , the uniform model gave the most accurate scoring of the background models we tested . The data demonstrate that HMM scoring based on triplet frequencies can accurately predict the observed binding affinities across a wide range of Kd values ( from ∼250 nM to 1 nM ) . Probe 6 yields a z-score of 0 . 82 and binds with a Kd of 257 nM , whereas probe 10 scores 2 . 74 in the model and binds with a Kd of 73 nM ( Figure 2B ) . These sequences include G containing triplets that contribute to the binding scores . This method allows any sequence to now be quantitatively assessed for possible PTBP1 binding , which was not previously possible by simply looking for clusters of a limited number of motifs . This HMM based approach should be applicable to the prediction of binding sites and affinity for other multi-domain RNA binding proteins . With our new method of defining PTBP1 binding sites , we next examined PTBP1 target exons for the location of predicted PTBP1 binding . In part , we wanted to reassess two previous studies that came to differing conclusions regarding the placement of PTBP1 sites adjacent to its target exons . One group mapped PTBP1 CLIP clusters adjacent to a limited number of PTBP1 repressed and enhanced exons [13] . This study described PTBP1 repressed exons as enriched for binding sites both upstream and downstream , as has been seen in studies of individual exons . They did not observe PTBP1 CLIP clusters within repressed exons , even though such exons have been described [17] , [32] , [33] . The PTBP1 enhanced exons they examined showed a trend in PTBP1 binding near the flanking constitutive exons . A second study examined exons showing altered splicing on splicing-sensitive microarrays after Ptbp1/Ptbp2 double knockdown [19] . CLIP clusters derived from the first study were mapped to these exons . The authors found CLIP cluster enrichment upstream and within PTBP1/PTBP2 repressed exons . In contrast to the previous study , they found that PTBP1/PTBP2 enhanced exons showed enrichment for CLIP tags in the downstream region . This pattern of binding site placement relative to repressed and enhanced exons has been observed for several other splicing regulatory proteins [14] , [34] . In our study , we defined four groups of exons from a set of exons previously assessed for splicing after Ptbp1 knockdown [20] , [35] . These included 68 PTBP1-repressed exons whose splicing increases after Ptbp1 knockdown , 37 PTBP1-enhanced exons whose splicing decreases after knockdown , 69 control exons that are not affected by Ptbp1 depletion but are known to be alternatively spliced ( PTBP1-non regulated ) , and 1 , 000 constitutive exons . We determined the density of predicted PTBP1 binding states within a 24-nucleotide window sliding along the exon region . We also examined the sequence encompassing the adjacent constitutive exons ( Figure 3A ) . As expected , the non-regulated control and constitutive exon sets did not exhibit high probabilities of PTBP1 binding except in the polypyrimidine tract of the 3′ splice site . On the other hand , the introns upstream of PTBP1 repressed exons show enrichment of potential PTBP1 binding sites starting from 250 nucleotides upstream of the exon . Relative to the control exons , exons repressed by PTBP1 also exhibited substantial enrichment of PTBP1 binding sites within the exon itself and within the first 100 nucleotides of the downstream intron . The repressed exons thus exhibit binding site placement that combines the findings of the two previous studies [13] , [19] . The PTBP1-enhanced exon set also shows enrichment of PTBP1 binding sites within the downstream intron relative to control exons , although the distribution of binding sites across this region was different between the repressed and enhanced exon sets ( Figure 3A ) . Similar to what was seen in the previous study by Llorian , we found little enrichment of PTBP1 sites within enhanced exons [19] . There is a limited enrichment adjacent to the exons flanking enhanced exons . Interestingly however , we find some PTBP1 enhanced exons that have PTBP1 binding sites upstream of the exon . These were not seen in either previous study . Our results are generally consistent with the known placement of PTBP1 binding sites in PTBP1 target exons and imply that rules correlating the position of PTBP1 binding to its effect on a target exon are not as strict as seen for some other splicing regulators . The mechanisms proposed from previous maps of PTBP1 binding do not appear to be generalizable to all PTBP1 targets [13] , [19] , [27] . Binding maps for PTBP1 and other splicing regulators show the averages of multiple exons . Since the data indicated a high level of variability in binding site placement between individual exons , we wanted to visualize target exons relative to each other . To display binding signals for individual exons we created heat maps of the binding scores upstream , within , and downstream of each exon in the PTBP1 target set ( Figure 3B ) . This display makes clear that the location of PTBP1 binding sites within its known target exons is variable . We found that 60% of PTBP1 repressed exons are predicted to have strong binding sites within the upstream intron . Most of these exons also have strong binding sites within either the exon or the downstream intron , patterns that were observed previously [13] , [19] , [27] . However , other patterns of binding site placement are also seen , suggesting PTBP1 dependent exons are following multiple rules . Some repressed exons score highly for PTBP1 binding only within the exon or in both the exon and the downstream intron . About half of PTBP1 enhanced exons have strong PTBP1 binding sites downstream ( Figure 3B ) . These can co-occur with upstream intron-binding sites , but rarely with exon binding sites . Interestingly , there are exons enhanced by PTBP1 with strong upstream binding in the absence of other sites . These data demonstrate the heterogeneity in the position of PTBP1 binding sites for its target exons . This heterogeneity needs to be considered for predicting PTBP1 dependent regulation . PTBP1 repressed exons exhibited significantly higher average binding scores in both the upstream intron and in the exon itself , than either the control group of alternative exons or the PTBP1 enhanced exons ( Figure 3C ) . The average binding scores in the downstream introns were higher for both the PTBP1-repressed and PTBP1-enhanced exons than the control group ( Figure 3C ) , although not at the same statistical significance . The variability of binding site placement within the smaller group of PTBP1-enhanced exons presumably contributes to the weaker statistical correlation of binding scores with positive regulation . We also compared the three exon sets for other features that might contribute to their ability to be regulated by PTBP1 , including exon length , flanking intron length , and 5′ and 3′ splice site strength . Most of these features were not statistically different among the three-exon groups . However , both PTBP1 enhanced and PTBP1 repressed exons were found to carry significantly weaker 3′ splice sites than the control exon set , as measured by the Analyzer Splice Tool ( Figure 3C ) [36] , [37] . These results indicate that PTBP1-repressed exons , and perhaps PTBP1-enhanced exons , exhibit an ensemble of sequence features that define them as PTBP1 regulated and that should allow their identification by sequence alone . Alternative exons are generally regulated by multiple factors that act both positively and negatively on their ability to be spliced . Thus , an exon controlled by a regulator in one context might not be affected by it under other conditions where counteracting factors are present , or required cofactors are absent . This means that the most accurate predictions of splicing regulation will need to consider many different factors . Nevertheless , models based on single factors will be useful for understanding the relative contributions of individual proteins to patterns of splicing regulation . Such models will be easier to interpret regarding the contributions of individual factors to individual exons than more complex models . Moreover in the longer term , models developed for different individual factors can be combined to make more accurate predictions . To assess how well one might model splicing regulation by a single factor , we examined whether the strength and placement of predicted PTBP1 binding sites could be used to predict new PTBP1 dependent exons . We plotted the scores for a variety of sequence features against the percent of exons exhibiting that score that also exhibit PTBP1 dependent exon repression ( Figure S5 ) . These plots produced distinct sigmoidal curves where most exons regulated by PTBP1 were found above or below a particular score . This strongly suggests that a logistic regression model incorporating each of these scores will be predictive of PTBP1 repression . We developed a multinomial logistic regression model and trained it on three classes of regulated exons ( Figure 4A ) [38] . The training set included PTBP1 repressed exons , PTBP1 enhanced exons , and non-regulated exons . Each exon in each class was scored for the four features found to correlate with PTBP1 regulation ( x1 through x4 ) , including the 3′ splice site strength , and the PTBP1 binding scores for each of three regions: the 250 nucleotides upstream of the exon , the exon itself , and the 100 nucleotides downstream of the exon . These intron lengths encompass the regions of binding site enrichment for PTBP1 dependent exons ( Figure 3 ) . The PTBP1-enhanced exons are fewer in number and show more limited enrichment of PTBP1 binding sites than PTBP1-repressed exons making the prediction for these exons less accurate . We first tested models that considered just PTBP1-repressed exons relative to control groups . However , we found that including the enhanced exons as a separate training group improved the prediction of repressed exons , even though enhanced exons themselves are not as easily identified ( data not shown ) . The trained model yielded values for the β coefficients that weight the different features contributing to the regulation . As expected the upstream binding score was weighted most heavily in predicting PTBP1 repression ( Table S1 ) , although binding scores in all three regions contributed to the score for PTBP1 repression . In contrast , we found that only the downstream binding score was significantly associated with PTBP1 enhancement . The upstream score generated a β coefficient close to zero making it essentially neutral in the prediction of enhanced exons . The exon binding score was subject to a negative β coefficient , indicating that exon binding reduces the probability of PTBP1 enhancement . Using these β coefficients , the trained models for repression or enhancement each yield a value of the g-function ( logit ) for an exon ( x ) given by the log of the ratio of the probability of repression or enhancement over the probability that the exon is not regulated . From this , the probability that an exon is repressed by PTBP1 can be determined from the two g-values as shown in Figure 4A . We assessed the multinomial logistic regression model by recursively retraining on exon sets with one exon left out and then scoring the missing exon . This leave-one-out cross validation enabled assessment of the overall performance of the model [38] ( Figure S6 ) . The PTBP1 dependent exon repression logit showed good prediction , with an area under the curve ( AUC ) value of 0 . 72 , substantially greater than random guessing ( AUC = 0 . 5 ) . As expected , the enhanced exon logit was not as accurate as the repression logit ( AUC = 0 . 57 ) , although it was better than random ( Figure S6A ) . Using these data , we assessed the sensitivity and specificity across the range of scores to define a decision threshold for exon repression scores ( Figure S6B ) . Increasing the threshold increases the specificity by eliminating many false positives , but decreases the sensitivity of the model in identifying maximum numbers of repressed exons . We sought to choose a threshold that gave a low false positive rate over one that yielded more regulated exons . We found that above a threshold score of 0 . 65 the false positive rate was 10% or lower ( Figure S6B ) . Applying the model to 4494 alternative cassette exons from UCSC genome browser database , we found 243 exons ( 5 . 4% ) that yielded a PTBP1 repression probability score greater than 0 . 65 and which were not in the training set . The 50 top-scoring cassette exons are listed in Table 1 . These included two exons that were reported previously to be PTBP1 targets . An exon of Gabrg2 yields a probability score of 0 . 92 . Although we could not confirm its repression in N2A cells because of low expression of the transcript , the orthologous exon in rat is a well-characterized PTBP1 repression target [39] . Exon 2 of Ptbp3 ( Rod1 ) , another known PTBP1 target [40] , yielded a repression probability score of 0 . 89 and was confirmed by RT/PCR to show increased inclusion after Ptbp1 knockdown ( Figure 4B ) . We performed additional RT-PCR validation in triplicate on a series of high and low scoring exons from transcripts expressed in N2A cells ( Figure 5 & Figures S7 , S8 and S9 ) . Seven of ten exons scoring above 0 . 65 were de-repressed after Ptbp1 knockdown in N2A cells , yielding a validation rate of 70% . The actual false positive rate is difficult to estimate because exons with high repression scores that are not affected by Ptbp1 depletion in N2A cells might be regulated by PTBP1 in other cells . An indication that this might be occurring is that the average inclusion level ( or percent spliced in value , PSI ) of the putative false positives is significantly higher than the confirmed true positives in N2A cells , indicating that they will be less prone to change upon Ptbp1 depletion and be more difficult to validate ( Figure S8B ) . Thus , the true positive rate may be greater than 70% . Importantly , the high validation rate for exons scoring above 0 . 65 indicates that the binding model and the regulation model based upon it can identify many new PTBP1 targets that were not previously known ( Table1 ) . High scoring exons might also fail to be validated because of regulation by other proteins . Knockdown of Ptbp1 induces expression of its close homolog Ptbp2 , which targets some of the same exons [20] ( Figure S7 ) . To test whether PTBP2 was also targeting the predicted PTBP1 repressed exons , we knocked down Ptbp2 or both Ptbp1 and Ptbp2 expression in N2A cells and re-assayed the exons in triplicate ( Figures S10 , S11 & S8A ) . Although some exons showed greater inclusion in the double knockdown compared to depletion of Ptbp1 alone , this did not validate any additional predicted PTBP1 repressed exons . We did identify some high and low scoring exons showing more complex regulation by the two PTB proteins ( Figure S10 & S11 ) . We also examined a set of low scoring exons ( probability score≤0 . 2 ) by RT-PCR after Ptbp1 and/or Ptbp2 depletion ( Figure 5B and Figure S11 ) . All of these exons ( 8 of 8 ) failed to respond to the loss of PTBP1 and are likely true negatives . Thus , PTBP1 repression scores above 0 . 65 and below 0 . 2 were highly predictive for regulation and its absence , respectively . As expected , intermediate scores were less consistent in their predictive value ( Figure S9 ) . Some exons in the intermediate scoring group were affected by PTB proteins and will be interesting to assess further . The prediction of PTBP1-repressed exons was improved by treating PTBP1-enhanced exons as a separate class , but the probability scores for PTBP1 enhancement did not consistently identify new PTBP1 target exons ( data not shown ) . This is likely in part due to the smaller number of exons in the training set and their heterogeneity , with some possibly being indirect targets . These predictions will likely improve with training on larger numbers of PTBP1 enhanced exons as they are identified . However , it is possible that simply the presence of the PTBP1 binding site is not sufficient for predicting PTBP1 enhancement and that binding sites for other factors will need to be considered . We next tested the model on a genomewide scale , by applying it to a set of 168 , 111 mouse internal exons and ranking them by their probability of PTBP1 repression . This analysis yielded 3824 exons ( 2 . 3% ) with probability scores above 0 . 65 for being repressed by PTBP1 . Among other activities , these exons were enriched in genes that function in calcium ion transport , cytoskeletal organization , intracellular transport , and synaptic transmission , all functions affected by previously known PTB targets ( Table S2 ) . To assess splicing of this large set of predicted PTB targets , we used RNA-seq to generate a large dataset of exons that change after Ptbp1 knockdown . RNA from control and PTBP1-depleted N2A cells was subjected to high density short read sequencing on the Illumina HiSeq platform using a strand specific , paired end protocol [41] . Exons whose inclusion changed between the two samples were identified by alignment to an exon database and quantification of exon inclusion using the SpliceTrap program [42] . After filtering for read coverage and removing the training set , we identified 573 alternative exons whose splicing was assayable in N2A cells . These exons exhibit changes in percent exon inclusion ( delta PSI ) ranging from −29% to 62% upon PTBP1 depletion . The exons were binned by their PTBP1 repression probability scores and plotted for their change in PSI ( Figure 6 ) . The average changes in splicing were significantly correlated with the repression probability . Exons scoring below 0 . 5 distributed around zero change in PSI , but above this score the average exon inclusion is altered by PTBP1 depletion . Most notably , exons with a repression probability score above 0 . 65 exhibited significantly larger changes in splicing than exons with lower scores . Exons with intermediate scores and hence weaker binding sites show smaller changes in splicing than high scoring exons . Setting a threshold of a 5% change in PSI as validation , 22 of 33 exons ( 67% ) that scored above 0 . 65 for PTBP1 regulation were confirmed as PTBP1 repression targets in N2A cells . At least some of the other 11 exons are presumably PTBP1 targets in other cells . To test the model in another cell type , we examined exons reported to change after Ptbp1 knockdown in mouse C2C12 myoblasts , as measured on splicing sensitive microarrays [43] . Very similar to what was observed in N2A cells , we found that exons with high repression probabilities showed significant de-repression upon the Ptbp1 knockdown compared to exons with low repression probabilities ( Figure S12 ) . Of 29 exons assayed on the arrays with a repression probability above 0 . 65 , 19 exons were confirmed as PTBP1 repressed on the array ( q-value<0 . 05 ) , yielding a validation rate of 66% . Thus the model performed very similarly in C2C12 and N2A cells . Among the 11 high scoring exons identified as unchanged after PTBP1 knockdown in N2A cells only 3 were assayed on the array and expressed in C2C12 cells . These again showed high inclusion in C2C12 prior to knockdown and so were difficult to assay for derepression . Thus , it is difficult to use the C2C12 data to draw conclusions about the false positive rate . The logistical model gives us a new tool for studying the regulation of alternative splicing . Using it , we can now scan genomic sequence to score exons for PTBP1 regulation . Applying the model genomewide , the PTBP1 repression probability scores were integrated into the UCSC genome browser . These data , displayed with the RNAseq data from N2A cells are available at our website ( http://www . mimg . ucla . edu/faculty/black/ptbatweb/ ) . A novel PTBP1 repressed exon in the Kcnq2 gene is shown in Figure 6B . The logistic model thus allows the assessment of any exon across the transcriptome for likely PTBP1 regulation .
We have developed two computational models , one that allows accurate prediction of PTBP1 binding sites and another that predicts likelihood of PTBP1 repression of exons across the transcriptome . These models uncovered several new features of RNA recognition by PTBP1 and the properties of its target exons . The PTBP1 binding model was based on triplets following the structures of the PTBP1 RRM domains , whose sequence specific contacts are each primarily to three nucleotides . We find that the set of triplets that increase the probability of binding includes the expected pyrimidine motifs , particularly those with alternating cytosines and uridines . However , many triplets with guanosine residues also increase binding probability . In contrast , adenosine residues have a negative effect on binding . Thus , RNA recognition by PTBP1 is not solely dependent on pyrimidine nucleotides . The recognition of G residues by PTB was unexpected , although some previously characterized PTB binding sites did contain G residues [13] , [44] . With this model , we can now predict PTBP1 binding affinity to any site in the transcriptome . The base-specific contacts that PTBP1 makes with Guanosine are not yet clear . Recent studies of RNA recognition by SRSF2 ( SC35 ) protein have shown that the element GGAG can be recognized by the same RRM as CCAG by flipping the initial two G nucleotides to the syn conformation [45] . It will be very interesting to investigate whether a similar anti to syn switch occurs in RNA bound by PTBP1 , when C residues are replaced with G . Previous characterizations of PTBP1 binding sites have focused on finding enriched short motifs within populations of bound RNAs or regulated exon sequences [13] , [44] , [46] , [47] , [48] . These methods generally identify elements whose short length will allow interaction with only one RRM domain . Searching for new binding sites comprised of clusters of these short elements can identify higher affinity sites but does not consider all elements or rank them . Crosslinking-immunoprecipitation experiments allow large numbers of binding regions to be identified . However , not all the sequence within a CLIP tag will be contacting the protein and it is difficult to relate CLIP signals to binding affinity . The HMM allowed the individual assessment of different short elements within the CLIP clusters , showing that they segregated into two states . The ranking of the triplets for their contributions to one of these states yielded a model where complex clusters of short elements could be assessed for binding and yielded accurate predictions of binding affinity . Many RNA binding proteins are similar to PTBP1 in having multiple domains that may each make different base specific contacts with RNA . The widespread generation of CLIP-seq datasets will allow the modeling of RNA recognition by almost any protein based on a large number of known binding sites . Using the same modeling approach , we also developed a binding model for PTBP2 ( neuronal PTB ) using a published PTBP2 CLIP dataset [49] . PTBP2 is about 70% identical to PTBP1 in sequence , and has only two amino acid changes among the residues making direct contact with RNA [17] . We found that the binding models for two PTB proteins were also nearly identical indicating that the two proteins are likely to differ more in their protein/protein interactions than in their RNA binding sites ( Data not shown ) . Several PTBP1 target exons have been analyzed in detail [17] , [50] . These exons vary in the placement and action of their PTBP1 binding sites . It is common for PTBP1-repressed exons to have a binding site upstream , often encompassing the branch point of the 3′ splice site [39] . Exons can also be repressed by PTBP1 binding within the exon [19] , [32] , [33] . Other exons contain downstream binding sites that are needed in conjunction with an upstream site to achieve splicing repression [51] , [52] , [53] . Although acting as a repressor for most of its targets , PTBP1 also activates the splicing of a group of exons . There have been divergent reports about placement of PTBP1 binding sites needed to mediate PTBP1 enhancement of splicing . The PTBP1 binding model allowed us to examine PTBP1 binding site placement across a large set of known PTBP1 target exons . Nearly all exons had predicted high affinity PTBP1 binding sites nearby . We found that more than half of PTBP1 repressed exons have high affinity binding sites upstream , and a fraction of PTBP1 enhanced exons have high affinity sites downstream . These exons fit with recent results on several other splicing regulators where the placement of the binding site determines the direction of the regulatory effect [12] , [14] , [34] . However , for PTBP1 these rules are not so clear . Some PTBP1 repressed exons have their strongest predicted binding site downstream or within the exon . These results indicate that there are fundamental differences between the mechanisms of PTBP1 mediated splicing regulation , and those governing regulation by certain other splicing factors . To quantify the predictive value of the PTBP1 binding scores for PTBP1 repression , we built a logistic model for PTBP1 regulation . For exons repressed by PTBP1 , binding scores for the upstream , downstream and exon sequences all contribute to the probability of repression . Exons enhanced by PTBP1 were too few to achieve accurate predictions from the model . However , treating these as a separate exon class improves the prediction of PTBP1 repression . We find that for probability scores above 0 . 65 the model is strongly predictive of PTBP1 repression . Applying this criterion across the transcriptome , we identified hundreds of new PTBP1 target exons . Alternative exons are generally regulated by multiple proteins acting in combination , and a particular exon will often be subject to both positive and negative regulation by antagonistic factors . For a model based on one factor , these other proteins will confound predictions . Exons with high PTBP1 binding scores may be counteracted by antagonistic factors in some cell types . Alternatively , synergistic factors may allow an exon with a relatively weak binding site to still recruit PTBP1 . Thus , a model based on one factor will be limited in its predictive power . In this study , our intent was to measure the effect of PTBP1 binding alone before considering the contributions of other factors . The logistic modeling allowed the contributions of different binding site placements to PTBP1 regulation to be measured . Several studies have used Bayesian models to dissect the regulatory properties of exons [7] , [54] . These models can generate accurate predictions by incorporating a wide variety of sequence , expression and conservation data . However , because so many disparate variables are incorporated , it can be difficult to draw mechanistic conclusions from these models regarding any one protein . For example , the presence of high pyrimidine density upstream from the branch point can be predictive of exons showing neuronal specific inclusion [7] , [55] . This is presumably in part due to many neuronal exons being regulated by PTBP1 and PTBP2 . However , a subset of these exons may be regulated by other factors with pyrimidine rich binding sites . In the long term , it will be most accurate to develop predictive binding models for each protein , similar to the PTBP1 model here , and then to incorporate each of these binding models into a larger network model . Such an approach will allow the analysis of the many overlapping regulatory programs controlled by RNA binding proteins .
A Hidden Markov Model ( HMM ) was designed and trained by an expectation–maximization ( EM ) method ( Baum-Welch algorithm ) using published PTBP1 CLIP data [13] , [29] , [30] . In total , 48 , 604 PTBP1-CLIP cluster sequences were used to train model parameters . During the training step , multiple initial values were tested to avoid a local maximum problem . Trained parameters included emission probabilities for nucleotide triplets , initial probabilities and transition probabilities between states [29] , [30] . The trained model was used to score RNA sequences . The raw PTBP1 binding score is defined as a log-odds ratio that compares the score of a sequence from the HMM over the score from a background model . Since CLIP experiments do not have an inherent corresponding negative dataset , we generated computational negative datasets and tested different background models ( Figure S4 ) . We found that a background model that values all triplets equally yielded the most accurate binding scores [29] . Raw scores were further normalized and converted to z-scores . For the 69 mer RNA sequences used in binding assays , scores were normalized by 100 , 000 random sequences with same length ( Figure S3 ) . This yielded very accurate predictions of binding affinity ( Figure 2 ) . When considering binding scores in genomic sequence , exons and upstream or downstream intron regions have different base compositions and will yield different average binding scores . Thus , to score binding sites adjacent to possible regulated exons , it is more informative to score sites relative to equivalent sequence regions . From the annotated mouse genome , we retrieved 168 , 111 internal exons and their flanking introns as separate sequence sets using a python library , Pygr . We scored log odds of these sequences with the trained model . Since the lengths and base compositions of intronic and exonic sequences are different , and binding scores automatically increase with length ( Figure S13 ) [29] , we grouped sequences by their location and sequences in each group were sorted according to length into bins of 1000 sequences each . The average score and standard deviation were determined for each bin . These values were used to transform the raw scores into z-scores for each upstream intron , downstream intron , and exon sequence . We localized the PTBP1 binding sites along each RNA sequence using the Viterbi algorithm [29] , [30] . To test predicted PTBP1 binding scores , we selected thirteen mouse exon/intron RNA sequences ( 69 nucleotides ) exhibiting a range scores . In the selection , other sequence features such as secondary structure were not considered . Target RNAs were transcribed in vitro from dsDNA using T7 RNA polymerase and subjected to an electrophoretic mobility shift assay ( EMSA ) . During the transcription , radioactive α-32P UTP was incorporated into RNA to visualize the probes . The RNA probes were then denatured for 2 min at 85°C and cooled down on ice immediately to reduce secondary structure formation . Binding assays were carried out as previously described with some modifications [27] . Specifically , each gel mobility shift reaction ( 10 µL ) contained the indicated amounts of recombinant human PTBP1 in 6 µL DG buffer ( 20 mM Hepes-KOH ph 7 . 9 , 20% glycerol , 80 mM potassium glutamate , 0 . 2 mM EDTA , 0 . 2 mM PMSF ) , 1 µL 22 mM MgCl2 , 1 µL 0 . 5 mg/ml tRNA , 0 . 5 µL RNase inhibitor ( 20 unit , RNaseOut from invitrogen ) , 0 . 5 µL DEPC treated H2O , and 1 µL 100 nM RNA probe . At first , all reaction components excluding RNase inhibitor , tRNA , and RNA probes were mixed and incubated for 8 min at 30°C . Then RNase inhibitor and tRNA were added and mixed . RNA probe was then added and the reaction was incubated for an additional 15 min . The reactions were put on ice for 5 min and mixed with 1 . 2 µL glycerol loading dye ( 30% glycerol ) . They were separated on 8% native polyacrylamide gels with 25 mM Tris-Gly running buffer in a cold room . Gels were dried and exposed to a phosphor screen . Then images were scanned using Typhoon 9410 and quantified using ImageQuant TL program ( GE Lifesciences ) . The apparent Kd values were estimated by fitting the data to non-linear curves using Prism software . An exon training set was compiled from previous microarray and RT-PCR experiments [20] , [35] . The training set was composed with 68 PTBP1 repressed , 37 PTBP1 enhanced , and 69 non-PTBP1 regulated simple cassette exons . We only considered exons with canonical splice sites ( GU-AG ) . An exon was classified as PTBP1 repressed or enhanced when 1 ) the inclusion level ( PSI ) of its minor isoform was greater than 5% in both the control and knock-down samples and 2 ) the inclusion level of its minor isoform was changed by 30% or more in the Ptbp1 knock down condition compared to the control sample . Next , we collected sequence features for each exon and its flanking exons . The features included PTBP1 binding scores , 5′ and 3′ splice site strengths , exon/intron lengths , and word frequencies . The PTBP1 binding scores were calculated from the PTBP1 binding model described above . The strength of splice sites was calculated by the splice-site analyzer tool [37] . Using a mouse whole internal exon set , we normalized features and fed them into the model . The PTBP1 splicing model is based on a multinomial logistic regression framework using the following steps: 1 ) selection of initial variables with a moderate level of association ( p-value from t-test<0 . 25 ) , 2 ) removal of outlier exons , 3 ) stepwise variable selection [38] . We scored mouse internal exons with the trained PTBP1 splicing model and validated candidate exons with RT-PCR and RNA-seq experiments . Exons from the training set were excluded from the validation . To test alternative splicing events for candidate exons , we assayed exon inclusion levels in cells following Ptbp1 , Ptbp2 , and double Ptbp1 & Ptbp2 knock down . The knockdown experiment was performed as described previously with minor modification [20] . Mouse neuroblastoma ( N2A ) cells were cultured in DMEM with 10% FBS and 2 mM L-glutamine . At 70 to 80% confluency , cells were trypsinized and suspended in the growth medium . DNA–Lipofectamine 2k ( Invitrogen ) complexes were prepared and mixed with cells in a tube according to manufacturer's instructions . Tubes were incubated for 5 h with mixing every half hour . Then cells were centrifuged and cultured in plates for 3 d . Proteins and RNA was extracted from collected cells . Protein samples were subjected to fluorescence immunoblotting to monitor knockdown efficiency of Ptbp1 and Ptbp2 . Total RNA was collected using Trizol ( Invitrogen ) according to the manufacturer's instructions . The RNA was further treated with DNase I to avoid DNA contamination . For RT-PCR ( Reverse Transcription-PCR ) assays , the RNA was reverse transcribed to cDNA with random hexamers using SuperScript enzyme ( Invitrogen ) following the manufacturer's instructions . PCR reactions were performed to assay alternative splicing of particular target exons . First , forward and reverse PCR primers were designed for the flanking exons using PRIMER3 program [56] . To label PCR products , a 5′ fluorescent-labeled universal primer ( 5′-FAM-CGTCGCCGTCCAGCTCGACCAG-3′ ) was added to the PCR reaction and a universal priming site was introduced to the 5′ end of the forward primer ( 5′-CGTCGCCGTCCAGCTCGACCAG-Forward Primer-3′ ) . Each PCR reaction ( 15 µL ) was carried out with 1 . 5 picomole of the forward primer and 6 . 75 picomole of the reverse and universal primers [57] . PCR amplification proceeded with an initial denaturation at 94°C for 4 m followed by 24 cycles of 94°C for 30 s , at a melting temperature of the reverse primer for 45 s , and 72°C for 45 s , with a final extension step at 72°C for 10 m . The samples were mixed with 2× formamide buffer ( Formamide with 1 mM EDTA pH 8 . 0 ) and denatured at 95°C for 5 min . Then samples were chilled on ice and run on 8% denaturing polyacrylamide gels . Gels were directly scanned by Typhoon and quantified by ImageQuant program . RNA-seq libraries were constructed following standard protocols ( Illumina TruSeq RNA Sample Prep Kit ) . To make strand-specific libraries , we added two extra steps to the protocol [41] . After first strand cDNA synthesis , remaining dNTPs were removed by a size selection on beads ( AMPure XP ) . Second-strand cDNA was synthesized with a dNTP mix containing dUTP instead of dTTP . The reaction contained samples eluted in 50 µl resuspension buffer , 2 µl 5× FS buffer , 1 µl 50 mM MgCl2 , 1 µl 100 mM DTT , 2 µl 10 mM dUTP nucleotides mix , 15 µl Second Strand Buffer ( Invitrogen ) , 0 . 5 µl E . coli DNA Ligase ( 10 U/µl;NEB ) , 0 . 5 µl RNase H ( 2 U/µl;Invitrogen ) , 2 µl DNA E . coli Polymerase I ( 10 U/µl;NEB ) . The reaction was incubated for 2 h at 16°C . After sequencing adaptors were ligated , 1 µl USER ( Uracil-Specific Excision Reagent enzyme; NEB ) was added to reactions to degrade the second strand cDNA . The samples were incubated for 15 min at 37°C and the reaction were inactivated at 94°C for 5 min . The samples were put in ice and then subjected to PCR amplification . Average size of inserts was about 225 bp and the libraries were subjected to 100 bp paired-end sequencing ( Illumina HiSeq2000 platform ) . Using SpliceTrap [42] , 60–65% of reads were mapped to exon duos or trios . In total , 180M ( 179 , 511 , 116 ) and 145M ( 145 , 334 , 711 ) paired end reads were used to infer exon inclusion ratios in the control and Ptbp1 knockdown conditions , respectively . The data have been deposited in NCBI's Gene Expression Omnibus [58] and are accessible through GEO Series accession number GSE45119 . | A key step in the regulation of mammalian genes is the splicing of the messenger RNA precursor to produce a mature mRNA that can be translated into a particular protein needed by the cell . Through the process of alternative splicing , mRNAs encoding different proteins can be derived from the same primary gene transcript . The regulation of this process plays essential roles in the development of differentiated tissues and is mediated by special pre-mRNA binding proteins . To understand how these proteins control gene expression , one must characterize what they recognize in RNA and identify these binding sites across the genome in order to predict their targets . Models that allow this prediction are essential to understanding developmental regulatory programs and their perturbation by disease causing mutations . In this study , we use statistical methods to build models of RNA recognition by the important splicing regulator PTBP1 and then apply these models to predict PTBP1 regulation of new gene transcripts . We show that PTBP1 has different specificity for RNA than was previously recognized and that its target exons are more diverse than was known before . There are many similar splicing regulators in mammalian cells , and these analyses provide a general framework for the computational analysis of their RNA binding and target identification . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"sequence",
"analysis",
"biochemistry",
"rna",
"rna",
"processing",
"nucleic",
"acids",
"gene",
"regulation",
"gene",
"expression",
"genetics",
"molecular",
"genetics",
"gene",
"splicing",
"biology",
"computational",
"biology"
] | 2014 | De Novo Prediction of PTBP1 Binding and Splicing Targets Reveals Unexpected Features of Its RNA Recognition and Function |
Transcription in mammalian nuclei is highly compartmentalized in RNA polymerase II-enriched nuclear foci known as transcription factories . Genes in cis and trans can share the same factory , suggesting that genes migrate to preassembled transcription sites . We used fluorescent in situ hybridization to investigate the dynamics of gene association with transcription factories during immediate early ( IE ) gene induction in mouse B lymphocytes . Here , we show that induction involves rapid gene relocation to transcription factories . Importantly , we find that the Myc proto-oncogene on Chromosome 15 is preferentially recruited to the same transcription factory as the highly transcribed Igh gene located on Chromosome 12 . Myc and Igh are the most frequent translocation partners in plasmacytoma and Burkitt lymphoma . Our results show that transcriptional activation of IE genes involves rapid relocation to preassembled transcription factories . Furthermore , the data imply a direct link between the nonrandom interchromosomal organization of transcribed genes at transcription factories and the incidence of specific chromosomal translocations .
Interphase chromosomes are organized in tissue-specific arrangements in nuclei , suggesting that chromosomal position and juxtaposition play a role in gene expression [1–5] . Nonrandom chromosome positioning has also been implicated in the frequency of specific chromosomal translocations . For example , Chromosomes 12 and 15 , which contain the frequent B cell translocation partners immunoglobulin heavy chain ( Igh ) and the proto-oncogene Myc , are preferred neighbors in mouse splenic lymphocytes [3] . Similarly , in human lymphoid cells MYC and IGH are found in the same vicinity in about one-third of nuclei [6] . Any process that brings these genes together would obviously be expected to increase the risk of a translocation between them; however , almost nothing is known about the forces that organize chromosomes in the nucleus . Nascent transcription occurs at RNA polymerase II ( RNAPII ) -rich nuclear foci known as transcription factories [7–11] . These sites are highly enriched in the hyperphosphoryated forms of RNAPII involved in transcription initiation and elongation [7 , 11] . Previous findings suggest that active mammalian genes are transcribed in bursts of activity punctuated by long periods of relative inactivity [9 , 12–14] . This concept is supported by recent live-cell studies showing that gene expression [15 , 16] , and in particular gene transcription [17] , occur in discrete pulses . We and others [9 , 10] have observed a virtually absolute correspondence between transcriptional activity at individual gene alleles and their positioning within transcription factories , whereas identical inactive alleles , often in the same cell , are clearly positioned away from factories . Collectively , these data could be interpreted to imply that the engagement of genes at factories is dynamic; however , they could equally be construed to indicate that a transcription factory nucleates around an individual gene during a transcriptional burst . Arguing against the latter interpretation is the finding that multiple genes in cis and trans can frequently share the same factory , which strongly suggests that genes migrate to preassembled transcription sites for transcription [9] . In this study , we investigated the positioning of immediate early ( IE ) genes relative to transcription factories and other B cell expressed genes during IE induction . We found that before activation the majority of IE alleles are not associated with transcription factories , whereas upon induction , IE genes rapidly relocate to preformed transcription factories . Remarkably , we observed preferential recruitment of the proto-oncogene Myc to the same transcription factory that is occupied by its frequent translocation partner , Igh . Our results suggest that this frequent and preferential juxtaposition may provide the opportunity for a chromosomal translocation , and may in part dictate the incidence with which specific chromosomal translocations occur .
Resting B cells can be stimulated through the B cell receptor signaling pathway to rapidly increase transcription and mRNA expression of the IE genes Fos and Myc [18] . We used RNA fluorescent in situ hybridization ( FISH ) with gene-specific intron probes to investigate the transcriptional activity of several genes during IE gene induction in mouse B lymphocytes ( Figure 1 ) . We found that transcription frequencies vary for several genes in B cells , similar to our previous observations in erythroid cells [9] . Transcription signals for the B cell-specific gene Igh are present at approximately 90% of alleles both before and after induction ( Figure 1A ) . In the vast majority of cells both alleles are actively transcribed ( >80% of cells ) , whereas a smaller percentage of cells have only single signals , consistent with our previous findings [19] . The immunoglobulin light chain genes Igk and Igl also have transcription signals at approximately 90% of alleles , with approximately 80% of cells having two transcription signals ( Figure 1B and 1C ) . These data show constitutive transcriptional activity at nearly all alleles that is unchanged upon IE induction . In contrast , transcription of the IE proto-oncogenes Fos and Myc is significantly lower in unstimulated cells , with 20% and 26% of alleles , respectively , displaying transcription signals ( Figure 1D and 1E ) . Most cells ( ~60% ) have two silent alleles , some cells have one active allele ( ~30% ) , and a minority of cells have two active alleles ( <10% ) . Upon induction , the percentage of loci with transcription signals for Fos and Myc rises dramatically within 5 min to 53% and 75% , respectively . This rise is the result of a dramatic increase in the percentage of cells with two active alleles and , to a lesser extent , an increase in cells with one active allele . The fold increases in the percentage of active alleles are in precise agreement with run-on transcription studies in mouse B cells that demonstrated a 2- to 3-fold induction of Fos and Myc transcription in stimulated cells [18] . These results suggested the possibility that increased IE expression could be accounted for by transcriptional recruitment of additional IE alleles rather than an increase in the “basal” rate of transcription at all IE alleles . We questioned whether the increase in nascent transcription levels seen in run-ons could be accounted for solely by the transcriptional recruitment of additional IE alleles . We used a sensitive reverse transcription PCR ( RT-PCR ) technique capable of quantitating the average absolute number of primary transcripts per cell [20] . In this method serial dilutions of a known amount of a spiked competitor RNA that contains a small internal deletion is compared to the amount of endogenous primary transcript in total RNA preps from a known number of cells . We adapted this method by using PCR primers that flank a 5′ splice donor site . Cleavage of the primary transcript at the 5′ donor site occurs soon after the RNA polymerase has passed the 3′ splice acceptor site , at the end of the intron sequence [21] . Quantitative detection of RT-PCR products from this part of the primary transcript provides an extremely sensitive measure of the number of transcripts being synthesized over that intron . An estimate of the average number of primary transcripts per gene can be calculated by extrapolating to the full size of the transcription unit . We found that the average number of Igh primary transcripts does not change during B cell induction ( Figure 1F; Table 1 ) , consistent with our FISH data , which indicate that Igh transcription is unaffected by stimulation . We detected approximately 4 . 3 and 4 . 5 unspliced primary transcripts on the Igh intron per cell in unstimulated and stimulated cells , respectively . As this intron is approximately half the Igh transcription unit length , we calculated that there are an average of 8 . 5 and 9 . 0 primary transcripts being produced from the two Igh transcription units per cell ( 4 . 3 and 4 . 5 per Igh allele ) . Since RNA FISH shows that 90% of Igh alleles are transcriptionally active , we estimate that there are approximately five transcripts being produced on each active allele . The picture for the IE Fos gene is very different . In unstimulated cells we found that the extrapolated , average absolute number of Fos primary transcripts per cell is 1 . 73 copies . This is less than one primary Fos transcript per allele , indicating that not all Fos alleles are transcriptionally active in unstimulated cells . If we calculate the average number of Fos transcripts per active allele based on our RNA FISH data in which 20% of Fos alleles showed a transcription signal , we arrive at an average number of 4 . 3 Fos primary transcripts per active allele . In stimulated cells Fos primary transcript intron copies per cell increase approximately 2 . 7-fold , consistent with the 2 . 65-fold increase in the percentage of actively transcribed Fos alleles determined by RNA FISH . Comparison of the number of Fos primary transcripts per active allele indicates that the number of Fos transcripts per active allele does not change upon induction ( 4 . 3 versus 4 . 4 copies ) . These data show that our RNA FISH technique is very sensitive and is capable of measuring very small numbers of primary transcripts at a transcription site . In addition , these results show that when we do not see an RNA FISH signal over a particular allele , it is truly “off” and has no primary transcripts associated with it . Collectively , these results show that Fos IE gene induction occurs via transcriptional activation of additional , previously inactive alleles , rather than by simply increasing the “basal” rate of transcription of all alleles . Our previous studies in erythroid cells suggested that gene association with transcription factories is dynamic , with genes moving to preformed factories in order to transcribe [9] . Inducible gene expression in B cells permitted us to examine the dynamics of transcription in relation to transcription factories . We speculated that the activation of previously quiescent IE alleles upon B cell induction may involve repositioning of alleles to transcription factories . However , the Myc gene has a well-characterized attenuation site that is thought to block passage of RNAP II , resulting in a stalled polymerase [22] . This observation leaves open the possibility that silent Myc alleles may be pre-positioned in factories awaiting removal of a transcriptional block . We first examined the positions of actively transcribed genes relative to transcription factories , using RNA immuno-FISH ( Figure 2 ) . We found that 92% of Myc RNA FISH signals are associated with strong RNAP II foci ( Figure 2A ) . Similarly , 90% of transcriptionally active Igh alleles are associated with strongly staining RNAP II foci ( Figure 2B ) , consistent with previous observations of erythroid-expressed genes [9 , 10] . Others have shown that the remaining 10% of RNA FISH signals localize to weakly staining RNAPII foci [10] , indicating that essentially all transcriptionally active alleles associate with transcription factories . This observation agrees with previous studies that showed a good correspondence between pulse-labeled nascent RNA and RNAPII foci [7 , 11] . We conclude that all Myc and Igh transcription occurs at transcription factories . Next , we investigated the position of nontranscribing alleles by DNA immuno-FISH , which detects DNA of both active and inactive alleles and RNAP II proteins . We found that approximately 30% of Myc DNA FISH signals overlapped with RNAP II foci in unstimulated cells , while 70% were not associated with RNAP II foci . These results are consistent with the percentage of transcriptionally active alleles detected by RNA FISH ( Figure 3A and 3C ) , and show that the inactive Myc alleles are not associated with transcription factories , but are instead positioned away from these sites . Three-dimensional DNA FISH measurements between the inactive Myc alleles and the nearest transcription factory show that on average silent Myc alleles are 500 nm from the nearest factory ( Figure 3D ) . After 5 min of stimulation the percentage of Myc loci associated with transcription factories increased to 65% , in agreement with the increased percentage of actively transcribing Myc alleles determined by RNA FISH ( Figure 3B and 3C ) . These results show that Myc induction involves an increase in the percentage of Myc alleles associated with transcription factories . The increase of gene association with factories could be achieved in two ways . Myc transcriptional induction could involve the nucleation of transcription factories on newly activated Myc alleles . Alternatively , transcriptional induction could involve the rapid relocation of silent Myc alleles to preassembled transcription factories . The synchronous induction of IE gene alleles described above allowed us to directly test these alternate scenarios . The Igh locus on mouse Chromosome 12 is positioned 28 Mb telomeric to the Fos locus . Approximately 90% of Igh alleles exhibit RNA FISH signals in B cells ( Figure 1A ) [19] , and are associated with transcription factories , indicating that the Igh locus undergoes nearly constant transcription , similar to the highly expressed Hbb locus in erythroid cells [9 , 23] . We therefore used Igh RNA FISH signals as factory reference points and scored the percentage of Igh transcription signals that have a colocalizing ( overlapping ) Fos signal , before and after induction using double-label RNA FISH . In unstimulated cells approximately 7% of Igh signals had a colocalizing Fos signal ( Figure 4A and 4C ) . After induction , colocalization nearly tripled , with 20% of Igh signals having a colocalizing Fos signal . These results suggest that a significant proportion of the newly activated Fos alleles move to a factory that is already occupied by an Igh allele rather than forming their own factory . We previously showed a low but significant level of interchromosomal associations between the highly transcribed Hbb and Hba genes in erythroid cells [9] . We hypothesized that the preferred neighbor arrangement of Chromosomes 12 and 15 in B cells [3] might allow Myc and Igh to co-associate with the same transcription factory in trans . Using double-label RNA FISH as above , we found that approximately 6% of Igh signals have a colocalizing Myc signal in unstimulated cells ( Table 2 ) . Comparing the percentage of active Myc alleles that colocalized with an Igh signal to those that did not , we found that a remarkable 25% of the transcribing Myc alleles colocalized with Igh in trans prior to induction . Upon induction , as we observed a 2 . 9-fold increase in the percentage of transcribing Myc alleles , we found a 2 . 5-fold increase in the percentage of Igh alleles with a colocalizing Myc signal ( Figure 4B and 4C; Table 2 ) . Again , by comparing colocalizing versus noncolocalizing Myc signals we found that approximately one-fourth of the active Myc alleles ( 22–24%; Table 2 ) were associated with Igh alleles upon induction . Thus , one-fourth of the newly activated Myc alleles , which were previously located away from transcription factories , had moved to a factory occupied by Igh . We confirmed that colocalizing Myc and Igh transcription signals co-associated with a shared transcription factory using triple-label RNA immuno-FISH to detect transcriptionally active Myc and Igh alleles and RNAP II foci ( Figure 4D ) . We found that all colocalizing Myc and Igh signals overlapped with the same transcription factory . In order to put this extraordinarily high frequency of interchromosomal Myc-Igh colocalization into perspective we compared the colocalization frequencies between Igh and five other B cell-expressed genes . One gene , Eif3s6 , is located approximately 20 Mb from Myc on Chromosome 15 , and four other genes , Igk , Igl , Uros , and Actb , are on Chromosomes 6 , 16 , 7 , and 5 , respectively . Since Chromosomes 12 and 15 are preferred neighbors in B cells [3] , we considered the analysis of colocalization between Eif3s6 and Igh to be of particular interest . If the high level of interchromosomal colocalization between Myc and Igh were simply due to the fact that the genes are on neighboring chromosomes , then we might expect Eif3s6 and Igh to colocalize at similar frequencies when transcribed . We found that Eif3s6 and Igh colocalize , but at significantly lower levels than Myc-Igh . Only 11% of Eif3s6 signals colocalized with Igh , compared to approximately 25% for Myc . For the other genes we found that 9% of Uros , 8% of Igk , 6% of Igl , and 2% of Actb transcribing alleles colocalized with Igh ( Figure 5 ) . These considerably lower frequencies of co-association with Igh clearly demonstrate that trans co-association frequencies between different gene pairs can vary greatly . For example Igh-Myc trans colocalization is over 10-fold higher than Igh-Actb , indicating that the Myc and Igh trans colocalization frequency is statistically highly significant . However , the Myc-Igh co-association is also highly preferential , as indicated in the comparison between Myc and Eif3s6 . Myc co-associates with Igh at a greater than 2-fold higher frequency than Eif3s6 . This result demonstrates that not all genes on neighboring chromosomes co-associate at equal frequencies and shows that Myc and Igh preferentially co-associate in trans . Our results suggest that 5 min after induction , many previously inactive Myc alleles are moving to preformed factories that contain transcriptionally active Igh alleles . If this interpretation is correct we would expect to see a net movement of Myc alleles toward Igh alleles upon stimulation of IE gene expression . To directly test this hypothesis we carried out 3D DNA FISH , measuring the separation distances between Myc and Igh alleles in unstimulated and stimulated cells . We found a statistically significant shift in the distribution of measurements upon B cell stimulation , changing from a mean separation distance of 2 . 16 μm to 1 . 83 μm ( p = 0 . 005 ) ( Figure 6A ) . This shows that across the population of cells Myc and Igh alleles are significantly closer together 5 min after induction . In contrast , we found no significant change in the distributions of measurements between Igh and two other genes in trans , Actb and Uros ( Figure 6B and 6C ) . These results show there is no net movement between Actb-Igh and Uros-Igh upon B cell activation , indicating that these genes are not significantly changing their location relative to one another . However , there is net movement of Myc alleles toward Igh upon induction . We did not detect any net movement between Igh and Fos upon induction ( Figure 6D ) , most likely because the range of separation between these physically linked genes is too small to detect subtle changes in relative positioning via light microscopy [9] . We conclude that increased Myc expression during IE gene induction involves the rapid relocation of Myc alleles to preassembled transcription factories , with many alleles migrating to a factory containing the Igh gene . We also assessed separation distances between Myc and Igh alleles in two other tissue types , adult kidney and fetal liver erythroid cells . We detected much greater separation distances between Myc and Igh in these tissues compared to unstimulated B cells ( Figure 6E ) . Myc and Igh were separated by an average of 2 . 78 and 3 . 20 μm in kidney and erythroid cells respectively , compared to an average of 2 . 16 μm in unstimulated B cells . These results are consistent with previously reported observations of tissue-specific positioning of genes [6 , 24] and chromosomes [3 , 25] and show that Myc and Igh are already in the same “nuclear neighborhood” in unstimulated B cells , which most likely facilitates their increased proximity and colocalization upon stimulation . To corroborate the FISH results , we used the capturing chromosome conformation ( 3C ) assay [9 , 26 , 27] , which measures ligation frequency between in vivo formaldehyde cross-linked chromatin fragments . Ligation products were detected with four different primer pairs within the Igh and Myc loci in stimulated B cells ( Figure S1 ) . The primer pair that produced the most robust product ( primer pair d/g ) was used to detect ligation products in unstimulated and stimulated B cells . Over multiple experiments , ligation products were always detected in stimulated B cells , while in unstimulated cells the products were usually , but not always detected . In contrast , Igh/Myc ligation products were never detected in brain or kidney cells , indicating that juxtaposition of Myc and Igh is restricted to tissues in which both genes are expressed . Importantly , Myc and Igh are the two most common translocation partners in Burkitt lymphoma and mouse plasmacytoma . Of these cancers 80% harbor Myc-Igh translocations , while the remaining cases contain Myc-Igk ( 15% ) or Myc-Igl ( 5% ) translocations [28 , 29] . To establish whether there is a relationship between the frequency of these translocations in plasmacytomas and the co-association frequencies of transcriptionally active alleles in normal B cells , we measured the extent to which the transcriptionally active Myc colocalized with Igk and Igl alleles in 10-min stimulated cells by RNA FISH . We found that 11% of transcribing Myc alleles colocalized with Igk and 7% colocalized with Igl , compared to 22% with Igh ( Figure 7 ) . Thus the frequencies of co-association between Myc and the immunoglobulin loci in transcription factories are in line with the appearance of their respective translocation frequencies in mouse plasmacytomas .
Our results show that IE gene induction involves the rapid nuclear relocation of previously inactive genes to preassembled transcription factories . This dynamic transcriptional organization is nonrandom and leads to the preferential juxtaposition of the Myc and Igh genes at transcription factories . Transcriptional colocalization may provide an opportunity and therefore an increased risk of illegitimate recombination resulting in a chromosomal translocation [30 , 31] . We cannot discount the possibility that differences in oncogenic potential result in selective outgrowth of one type of translocation versus another . However , it is striking that the co-association frequencies echo the appearance of specific translocations in plasmacytomas , suggesting that the juxtaposition frequency of specific genes in a transcription factory has a direct effect on their translocation frequency . Upon B cell induction , signaling pathways converge upon the IE genes , a process that causes their relocation to transcription factories . Others have shown recently that upon activation , genes can undergo directed , actin and myosin-dependent relocalization , moving between 1 and 5 μm [32] . We cannot discount the possibility that similar forces may be involved in the relocation of genes to factories . However , we found that inactive Myc alleles are positioned on average 500 nm from the nearest factory , a distance that could conceivably be covered by random chromatin movements [33–35] . A key question concerns the basis of the preferred co-associations of specific genes in a common factory . The relative positions of genes in cis or on preferred neighbor chromosomes would be expected to affect the frequency of co-association in a factory [9] as it does recombination frequency [36] . On the other hand , it is possible that tissue-specific chromosomal positioning is driven by the net effect of thousands of preferential interchromosomal interactions between active ( and inactive [37] ) genes that serve as dynamic anchor points that facilitate chromosome positioning [38] . Preferential co-associations in factories may be the result of 3D spatial clustering of genes with related functions or genes coordinately regulated by common factors [9 , 37 , 39] . Genome-wide examination of the subsets of genes that preferentially co-associate may provide valuable information about these influences . Igh translocations are presumed to occur through aberrant repair during programmed recombination , by recombinase activating gene protein ( RAG ) during V ( D ) J recombination , and activation-induced deaminase ( AID ) during somatic hypermutation and class switching [40] . Cleavage by RAG complexes at cryptic RAG recognition sites at other genes , and altered DNA structures have been implicated in the generation of some human IGH translocations [41 , 42] . However , the current consensus view is that cryptic RAG sites are not present in the major Myc breakpoint region . Igh translocations within the Igh diversity and joining regions occur in the bone marrow pre-B cells , which undergo V ( D ) J recombination [40] . However , most Igh translocations to Myc are found in the Igh class switch region and are believed to occur in germinal center B cells , the site of class switching [40] . There is strong evidence to suggest that genes may be susceptible to double-stranded breaks during transcription . The process of transcription creates considerable torsional stress [43] , which can be relaxed by topoisomerases via introduction of transient double-stranded breaks . In fact , topoisomerase type IIβ-generated double-stranded breaks in the promoter regions of some genes have recently been shown to be required for regulated transcription [44] . Topoisomerase cleavage sites are common features of translocation hot spots [45] . Significantly , topoisomerase type IIβ binding sites have been mapped to the major breakpoint region of the human MYC gene , at the 5′ end of the first intron [46] . Further evidence of a link between transcriptional organization and recombination is suggested by two papers , one of which showed that actively transcribed yeast tRNA genes cluster in the nucleolus [47] , and another which showed that recombination is higher between actively transcribed tRNA genes compared to inactive tRNA genes [48] . Interestingly , double-stranded break repair enzymes Ku70/80 are also associated with transcription factories [49] . In summary , the introduction and repair of double-stranded breaks may be commonplace in transcription factories . Therefore , interchromosomal co-associations between genes in factories may be expected to result in a heightened risk of aberrant repair of double-stranded breaks resulting in chromosomal translocations . It is curious that evolution has allowed the interchromosomal juxtaposition of the Myc and Igh loci in transcription factories to persist , considering the potentially grave risks of such an organization . However , the apparent dangers of illegitimate recombinations may be outweighed by advantages of clustering transcribing genes , which may make efficient use of shared resources , or perhaps provide a degree of transcriptional coordination of subsets of genes .
CD43− resting B cells were isolated from spleens of 6- to 8-wk-old BALB/c mice by magnetic cell sorting using CD43 microbeads ( Miltenyi Biotec , http://www . miltenyibiotec . com ) to deplete other cell types . Induction of B cells was done in PBS supplemented with 10 ng/ml recombinant mouse IL-4 ( Stemcell Technologies , http://www . stemcell . com ) , 20 μg/ml purified rat anti-mouse monoclonal antibodies to CD40 ( clone HM40–3 , Serotec , http://www . serotec . com ) and 10μg/ml goat anti-mouse IgM μ chain , F ( ab′ ) 2 fragment ( Jackson Immunoresearch , http://www . jacksonimmuno . com ) at room temperature for up to 15 min before fixation for FISH . RNA FISH was carried out as described previously [50 , 51] . We visualized Igh transcription with a dinitrophenol-labeled single-stranded DNA probe to the intronic enhancer region [19] , followed by Texas Red detection . We prepared digoxygenin or biotin-labeled single-stranded DNA probes to detect Fos , Myc , Eif3s6 , Uros , Actb , Igk , and Igl , intron sequences as described [19] . Primer sequences used to PCR-subclone the various probes are listed below . DNA FISH was carried out as previously described [52] . The following BAC clones ( BACPAC Resources , http://bacpac . chori . org ) were used: 234 kb RP23-98D8 for Myc; 178 kb RP24-233K8 for Fos; 166 kb RPCI24-258E20 for Igh; 151 kb RP24-132K17 for Uros; and 213 kb RP23-97O1 for Actb . For double-label experiments , we labeled one of the DNA FISH probes directly with AlexaFluor 594 and labeled the other probe with digoxigenin , detected with fluorescein-conjugated antibodies . Immunofluorescence and immuno-FISH was carried out as described [9 , 52] , using a CTD4H8 antibody ( Upstate Biotechnology , http://www . upstate . com ) that was raised against a Ser5-phosphorylated CTD . This antibody is specific to phosphorylated forms of RNAP II [53] . Primers used to amplify RNA FISH probes were the following . Fos intron 1 sense , 5′-GCTTTGTGTAGCCGCCAGGT-3′; antisense 5′-AGAGGAAAGCGGAGGTGAGC-3′ . Fos intron 2 sense , 5′-AAGTAGAGCTGGTGAGCAGCGATT-3′; antisense , 5′-AGAAAAGGACCAACATTCAGTTAAGG-3′ . Myc intron 1 sense , 5′-AGCACAGATCTGGTGGTCTTTC-3′; antisense , 5′-CTCCTTCGAGCAGGGACTTAG-3′ . Myc intron 2 sense , 5′-CTTCTCCACCACTCATTGGCATTA-3′; antisense , 5′-GGGAGGAAGTGGAAGATCACAGTT-3′ . Eif3s6 intron 1 sense , 5′-GTGAGGAAGCTTTGAGAAGGAGGA-3′; antisense , 5′-ATTAATTTTGCTGTTCCCTGCTGA-3′ . Uros intron 6 sense , 5′-TCAGCGCCACAGCAAGGGTT-3′; antisense , 5′-GCCTTCCCTCCTTTGTTCCCAGT-3′ . Actb intron 1 sense , 5′-TCGCTCTCTCGTGGCTAGTA-3′; antisense , 5′-TGGCGAACTATCAAGACACA-3′ . Igh Iμ intron sense , 5′-AGCTGTGGCTGCTGCTCTTA-3′; antisense , 5′-AGCCTCGCTTACTAGGGCTCTC-3′ . Igl J-C intron probe 1 sense , 5′-TGAGTGACTCCTTCCTCCTTTG-3′; antisense , 5′-TGGAGGCAGTGTGTAAAGTGTC-3′ . Igl J-C intron probe 2 sense , 5′-GTTGTCTTGCAAGGGTCTTTTT-3′; antisense , 5′-GTGCGAATAAAAGAAGGGATTG-3′ . Igk J-C intron sense , 5′-AAGACACAGGTTTTCATGTTAGGA-3′; antisense , 5′-AATAGAATTATGAGCAGCCTTTCC-3′ . We examined RNA FISH signals on an Olympus BX41 epifluorescence microscope , and assessed 200 loci for each probe combination , except for Uros-Igh , for which 140 alleles were assessed . Transcription signals scored as colocalizing if the red and green signals overlapped to create a visible yellow signal . To assess the association of Myc DNA FISH signals and RNAP II foci , we captured image stacks of nuclei , using an Olympus BX41 epifluorescence microscope , equipped with a UPlanApo 100× oil objective to reduce chromatic aberration , and fitted with a motorized stage . Images were captured and analyzed using Analysis 3 . 2 image capture software , fitted with a RIDE module ( SIS , http://www . sis . com ) . The stacks were deconvoluted using a nearest-neighbor algorithm with 85% haze removal , and analyzed . We analyzed 81 and 88 alleles in unstimulated and stimulated B cells , respectively . Statistical analysis was carried out using a two-sided Fisher exact test . For Myc alleles that were not co-associated with an RNAPII focus , we measured the separation distance from the edge of the gene signal to the edge of the nearest RNAP II immunofluorescence signal , and analyzed 27 alleles . To measure the distances between Igh and genes in trans by DNA FISH , we collected image stacks using a Zeiss 510 Meta confocal microscope . Separation distances for each Igh allele and the nearest Myc , Fos , Uros , or Actb allele were measured on 3D-reconstructed image stacks using Volocity image analysis software ( http://www . improvision . com/products/volocity ) . In all cases , we made measurements from center to center of the two gene signals . We analyzed at least 83 measurements for gene pairs in unstimulated and stimulated B cells , adult kidney cells , and E14 . 5 fetal liver cells . Changes in the distributions of measurements were assessed by two-sided Student t-test . The assay was carried out essentially as described [20] . RNA competitor fragments were generated by cloning 230 bp Igh and 236 bp Fos fragments that span 5′ exon-intron junctions . The plasmid containing the Igh fragment was digested with HindIII and AflII to release a 17 bp fragment , then religated . The Fos deletion was generated by BsmFI and NheI digestion to remove a 24 bp fragment . RNA was transcribed from linearized plasmids , then checked by gel electrophoresis and quantitated by UV spectrometry . Dilutions of controlled amounts of RNA was spiked into Trizol reagent that contained a known number of cells . RNA was extracted , reverse transcribed , and PCR amplified with nested primers . To measure the numbers of primary transcripts over the length of the intron , the relative intensity of the endogenous RT-PCR product was compared to the spiked competitor RT-PCR product , measured using AIDA quantitation software . For Igh , the lanes with four copies of spiked competitor per cell was used for quantitation . For Fos , quantitation was obtained from the average from the lanes with 1 copy per cell and 0 . 5 copy per cell . The primers used were as follows . Igh-f , 5′-CCTGGGAATGTATGGTTGTGGCTTC-3′; Igh-r , 5′-CCCCCTAAAGCAAT:GACTGAAGACTCA-3′; Igh nested-f , 5′-CCTCGGTGGCTTTGAAGGAACAAT-3′; Igh nested-r , 5′-CCCTAAAGCAATGACTGAAGACTCAGT-3′; Fos-f , 5′-AGCATCGGCAGAAGGGGCAAAGTA-3′; Fos-r , 5′-TGAAGTAGGAAGCTGTCAGGGAAACTG-3′; Fos nested-f , 5′-AGAAGGGGCAAAGTAGAGCAGGTGA-3′; Fos nested-r , 5′-TGTCAAAATCTGACAAGGGAGGGAAAG-3′ . We carried out the 3C assay as described previously [9] . We fixed B cells that had been stimulated for 5 min in 2% formaldehyde for 10 min at room temperature , and digested 1 × 106 nuclei overnight with 600 units of BglII . We ligated digested chromatin ( 2 μg ) with 2 , 000 units of T4 DNA ligase in a final volume of 800 μl . We cloned ligation product detected by primer pair d/g ( see list below ) and verified it by DNA sequencing . We tested the specificity of the 3C primers as previously described [9] . The primers used for 3C analysis were the following . Myc a forward , 5′-TCTACACCCCATACACCTCCA-3′; Myc a nested , 5′-CGAGAATATGCCATGAATTGG-3′ . Myc b forward , 5′-GGGGAGGGAATTTTTGTCTATT-3′; Myc b nested , 5′-GGACAGTGTTCTCTGCCTCTG-3′ . Myc c forward , 5′-TGCCCTCTCAGAGACTGGTAA-3′; Myc c nested , 5′-TTCCCCTTTCCTCTGTCATCT-3′ . Myc d forward , 5′-ATTCTTCCAGGTGGTGATGTC-3′; Myc d nested , 5′-CTTCCCACAGCTCTCTTCCTT-3′ . Igh e forward , 5′-AACCCATCTACCCATGTAGCC-3′; Igh e nested , 5′-CCTCTGACTGCCTCTTTTCCT-3′ . Igh f forward , 5′-ACTGTGATCGGTTTTGGAGTG-3′; Igh f nested , 5′-CTGGGAGGGTTTGGTTCTTAC-3′ . Igh g forward , 5′-CCCAGAACCTGAGAAGGAAGA-3′; Igh g nested , 5′-ACAGAACCGAACCATGACTTG-3′ . Igh h forward , 5′-TTGGGCACTAAACACCACTTC-3′; Igh h nested , 5′-GGTGTGTGCAGGTTTTTGTCT-3′ . Hbb-b1 forward , 5′-CTCAGAGCAGTATCTTTTGTTTGC-3′; Hbb-b1 nested , 5′-AGGATGAGCAATTCTTTTTGC-3′ . Calreticulin Cal1 , 5′-CTCCAGATAAACCAGTATGAT-3′; Cal2 , 5′-AAACCAGATGAGGGCTGAAGG-3′ . Actb Actb 1forward , 5′-CGGTGCTAAGAAGGCTGTTCC-3′; Actb 1nested , 5′-AGCAAGAGAGGTATCCTGACC-3′ . Actb Actb 2forward , 5′-TGTGACAAAGCTAATGAGG-3′; Actb 2nested , 5′-TGAGTAGATGCACAGTAGG-3′ . | Many different types of cancer result from gene translocations . Specifically , two different chromosomes can be joined that fuse growth control genes with powerful regulatory elements , leading to unrestricted control of cell growth . Translocation partner genes must physically encounter each other in the nucleus to undergo a translocation; how they find each other in the crowded nucleus is unknown . We showed previously that gene transcription occurs at a few hundred discrete nuclear sites called transcription factories . In the current study we investigated the effects of activation of the Myc proto-oncogene and examined its location with respect to transcription factories and its common translocation partner , the immunoglobulin heavy chain ( Igh ) gene . We found that switching on the Myc gene leads to its rapid relocation to a transcription factory . Surprisingly , we found that the activated Myc frequently chooses the same transcription factory as the highly transcribing Igh gene . This close juxtaposition of translocation partner genes at a shared transcription factory may provide the opportunity for a chromosomal translocation , and thus may be the first step in the genesis of several types of cancers . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"oncology",
"cell",
"biology",
"mammals",
"immunology",
"mus",
"(mouse)",
"molecular",
"biology",
"genetics",
"and",
"genomics",
"hematology"
] | 2007 | Myc Dynamically and Preferentially Relocates to a Transcription Factory Occupied by Igh |
Implantable electrode arrays are widely used in therapeutic stimulation of the nervous system ( e . g . cochlear , retinal , and cortical implants ) . Currently , most neural prostheses use serial stimulation ( i . e . one electrode at a time ) despite this severely limiting the repertoire of stimuli that can be applied . Methods to reliably predict the outcome of multi-electrode stimulation have not been available . Here , we demonstrate that a linear-nonlinear model accurately predicts neural responses to arbitrary patterns of stimulation using in vitro recordings from single retinal ganglion cells ( RGCs ) stimulated with a subretinal multi-electrode array . In the model , the stimulus is projected onto a low-dimensional subspace and then undergoes a nonlinear transformation to produce an estimate of spiking probability . The low-dimensional subspace is estimated using principal components analysis , which gives the neuron’s electrical receptive field ( ERF ) , i . e . the electrodes to which the neuron is most sensitive . Our model suggests that stimulation proportional to the ERF yields a higher efficacy given a fixed amount of power when compared to equal amplitude stimulation on up to three electrodes . We find that the model captures the responses of all the cells recorded in the study , suggesting that it will generalize to most cell types in the retina . The model is computationally efficient to evaluate and , therefore , appropriate for future real-time applications including stimulation strategies that make use of recorded neural activity to improve the stimulation strategy .
Implantable electrode arrays are widely used in clinical studies , clinical practice and basic neuroscience research and have advanced our understanding of the nervous system . Implantable electronic devices can be used to record neurological signals and stimulate the nervous system to restore lost functions . Sensing electrodes have been used in applications such as brain-machine interfaces [1] and localization of seizure foci in epilepsy [2] . Stimulating electrodes have been used for the restoration of hearing [3] , sight [4 , 5] , bowel control [6] , and balance [7] , and in deep brain stimulation ( DBS ) to treat a range of conditions [8] . Most neuroprostheses operate in an open-loop fashion; they require psychophysics to tune stimulation parameters . However , devices that can combine both sensing and stimulation are desirable for the development of a new generation of neuroprostheses that are controlled by neural feedback . Feedback in neuroprostheses is being explored in applications such as DBS for the enhancement of memory [9] , abatement of seizures [10] , control of Parkinson’s disease [11] , and the control of brain machine interfaces [12] . Models that can accurately characterize a neural system and predict responses to electrical stimulation are beneficial to the development of improved stimulation strategies that exploit neural feedback . Volume conductor models are typically used to describe retinal responses to electrical stimulation , however these are computationally intensive and can be difficult to fit to neural response data [13–15] . Simpler models that can be constrained using neural recordings are necessary for real-time applications . Linear-nonlinear models based on a spike-triggered average ( STA ) have been successfully used to characterize retinal responses to light [16–19] . Models that incorporate higher dimensional components identified through a spike-triggered covariance ( STC ) analysis have been explored to describe higher order excitatory and suppressive features of the visual system [20–25] . Generally , STA and STC models make use of white noise inputs and have the advantage that a wide repertoire of possible inputs patterns can be explored . White noise models have previously been explored to describe the temporal properties of electrical stimulation in the retina [26 , 27] . Spatial interactions between stimulating electrodes has not been previously investigated . An example of a stimulation algorithm that could benefit from an accurate description of the spatial interactions is current steering , which attempts to improve the resolution of a device by combining stimulation across many electrodes to target neurons at a particular point [28] . Two benefits obtained by using neural feedback algorithms are ( 1 ) the accurate prediction of the response to an arbitrary stimulus across the electrode array and ( 2 ) the ability to fit the device to individual patients from the recorded neural responses to a set of stimuli presented in a reasonable amount of time . Here , we combine whole cell patch clamp recordings from individual retinal ganglion cells ( RGCs ) with stimulation using a multi-electrode array to demonstrate a model with the above advantages . We find that a simple linear-nonlinear model accurately captures the effects of multi-electrode interactions and estimates the spatial relationship between stimulus and response . The approach is scalable to a large number of electrodes , which is prohibitive to accomplish with psychophysics . In contrast to conventional volume conductor models of electrical stimulation [13–15] , our model is straightforward to constrain using neural response data and is orders of magnitude more computationally efficient , making it suitable for use in real-time applications .
Methods conformed to the policies of the National Health and Medical Research Council of Australia and were approved by the Animal Experimentation Ethics Committee of the University of Melbourne ( Approval Number: 1112084 ) . Data were acquired from retinae of Long-Evans rats ranging from 1 to 6 months of age . Long-Evans rats were chosen for several reasons . First , rat RGC morphological types have been examined in detail [29 , 30] and have similarities to RGCs found in other species , including the macaque monkey [31] and cat [32 , 33] . Second , the size of the rat retina is larger than the mouse retina and so we were able to cover the entire stimulating electrode array with half of the retina . The animals were initially anesthetized with a mixture of ketamine and xylazine prior to enucleation . After enucleation , the rats were euthanized with an overdose of pentobarbital sodium ( 350 mg intracardiac ) . Dissections were carried out in dim light conditions to avoid bleaching the photoreceptors . After hemisecting the eyes behind the ora serrata , the vitreous body was removed and each retina was cut into two pieces . The retinae were left in a perfusion dish with carbogenated Ames medium ( Sigma ) at room temperature until used . Pieces of retina were mounted on a multi-electrode array ( MEA ) with ganglion cell layer up and were held in place with a perfusion chamber and stainless steel harp fitted with Lycra threads ( Warner Instruments ) ( Fig 1A ) . Once mounted in the chamber , the retina was perfused ( 4–6 mL/min ) with carbogenated Ames medium ( Sigma-Aldrich , St . Louis , MO ) at room temperature . The chamber was mounted on the stage of an upright microscope ( Olympus Fluoview FV1200 ) equipped with a x40 water immersion lens and visualized with infrared optics on a monitor with x4 additional magnification . Electrical stimulation was applied subretinally through a custom-made MEA fabricated on a glass coverslip consisting of 20 platinum stimulating electrodes ( Fig 1 ) . Each electrode had an exposed disc of 400 μm diameter , and a vertical pitch of 1 mm . The stimulating area of the MEA spanned an area of 3 . 5 x 3 . 5 mm2 ( excluding the outer ring which was not used ) . Glass coverslips were cleaned in an oxygen plasma chamber for 20 minutes ( Fig 1B ) . Next , a positive ( UV-removable ) photoresist ( AZ1415H , Microchemicals ) was spin-coated onto the surface at 4000 revolutions per minute for 60 seconds ( Fig 1C ) . A laser-printed chrome on soda glass photolithography mask was used to expose a pattern in the photoresist , then developed chemically ( MIF726 , Microchemicals ) revealing openings for electrode pads and tracks ( Fig 1D and 1E ) . The developed cover slips were loaded into an electron beam deposition chamber ( Thermionics ) and pumped to a vacuum pressure of 1 . 5×10−6 mbar . A 20 nm titanium adhesion layer was deposited at a rate of 0 . 2 Å/sec , followed by a platinum deposition of 130 nm at a rate of 0 . 6 Å/sec . Residual photoresist was removed by soaking in acetone for 30 minutes , rinsing with deionized water , and finally oxygen plasma cleaning for 10 minutes . For electrode isolation , a negative ( UV-curing ) photoresist ( SU8-2002 , Microchemicals ) was spin-coated onto the coverslip and exposed through a different photolithography mask leaving only metal exposed for stimulating electrodes and contact pads ( Fig 1F and 1G ) . The entire device was then cured at 200°C on a temperature-controlled hotplate . Whole cell intracellular recordings were obtained using standard procedures [34] at room temperature . The main reason for recording at room temperature was to ensure that recordings lasted for many hours . To obtain a whole cell recording , a small hole was made in the inner limiting membrane to expose a small number of RGC somas . A pipette was then filled with internal solution containing ( in mM ) 115 K-gluconate , 5 KCl , 5 EGTA , 10 HEPES , 2 Na-ATP , 0 . 25 Na-GTP ( mosM = 273 , pH = 7 . 3 ) , Alexa Hydrazide 488 ( 250 mM ) , and biocytin ( 0 . 5% ) ( Fig 1A ) . Initial pipette resistance in the bath ranged between 5–10 MΩ . Prior to recording , the pipette voltage was nulled , pipette resistance was compensated with the bridge balancing circuit of the amplifier , and capacitance was compensated on the amplifier head stage . Voltage recordings were collected in current clamp mode and amplified ( SEC-05X , NPI electronic ) , digitized with 16-bit precision at 25 kHz ( National Instruments BNC-2090 ) , and stored for offline analysis . Intracellular recordings lasting up to 4 hours were obtained . Stimulation artefacts that were present in the intracellular recording were removed offline by setting the membrane potential to a constant value for the duration of the stimulus . Spikes in the remaining membrane potential waveform could be easily detected by finding peaks that crossed a set value . Spike times were calculated as the time that the action potential reached its peak value . Spike delay times were calculated by taking the difference between the spike time and the preceding stimulus offset time . Intrinsic physiological differences , such as spike width , membrane time constant , and input resistance , among RGC types have been described [35 , 36] , which could lead to differences in response latencies to electrical stimulation . Therefore , we performed a k-means cluster analysis on the spike latency from stimulation offset time . The number of clusters ( k ) to fit was set manually by visual inspection of the clusters . From the cluster analysis , we could detect if there were two or more clusters that might be attributed to direct activation or indirect activation via activation of presynaptic neurons . Unless otherwise stated , responses to electrical stimulation were evaluated by analyzing the short-latency responses . Short-latency responses were spikes that fell within two standard deviations of the mean of the shortest-latency cluster . Long-latency responses fell within the cluster that occurred directly after the short-latency response . Stimulation consisted of biphasic pulses of equal phase duration ( 500 μs ) with an interphase gap ( 50 μs ) and random amplitude . The random amplitudes were sampled from a Gaussian distribution with variance σ2 . Fig 2 illustrates the random amplitude pulses applied to all electrodes . Stimulation waveform signals were generated by a custom-made MATLAB ( MathWorks version 2014a ) interface commanding a multi-channel stimulator ( Tucker Davis Technologies: RZ2 base station and IZ2 multichannel stimulator ) . All stimulus amplitudes were bounded by the limits of the stimulator ( ±300 μA ) . Biphasic pulses were applied to all electrodes at a frequency of 10 Hz and the numbers of short-latency responses were recorded . To avoid overstimulation of a cell , an appropriate value of σ was chosen for each cell . Three stimulus trains of 500 pulses were initially generated with fixed σ = 50 μA and applied to the tissue . Next , a new set of stimulus trains were generated using a σ that varied between 50 μA and 250 μA in steps of 50 μA . The number of spikes detected within 5 ms from the stimulus time was used to compute a response probability . A sigmoidal curve was fit to the data of σ versus response probability to find the value of σ that resulted in half the maximum level of response . For cells where the maximum response probability was close to 1 , σ was chosen to be a value that resulted in a response probability of 0 . 5 . For other cells that saturated at a response probability less than 1 , σ was a lower value . Once an appropriate value for σ was chosen for the cell , a stimulus vector , St→ , of length 20 ( equal to the number of electrodes ) was generated by sampling each element from a Gaussian distribution . If the amplitude of stimulation of an electrode exceeded the stimulator limits ( ±300 μA ) , then the amplitude value was discarded and a new value was generated from the distribution . Each stimulus was applied 3–5 times before a new St→ was generated . The experiment continued for as long as the cell’s responses remained stable ( usually 3–4 hours ) . Once cells started to show signs of deterioration ( e . g . unstable high frequency spontaneous activity ) , the experiment was terminated . After recording , the retinal tissue was removed from the chamber and mounted onto filter paper . The tissue was then fixed for ~45 min in phosphate-buffered 4% paraformaldehyde and stored for up to 2 weeks in 0 . 1 M phosphate-buffered saline ( PBS; pH 7 . 4 ) at 4°C . The tissue was then immersed in 0 . 5% Triton X-100 ( 20 μg/ml streptavidin conjugated Alexa 488; Invitrogen ) in PBS overnight to expose biocytin-filled cells . Tissue was washed thoroughly in PBS , mounted onto Superfrost+ slides , and coverslipped in 60% glycerol . Cells were then reconstructed in 3D using a confocal microscope ( FluoView FV1200 ) . RGC types were initially identified by their focal light response at the beginning of each experiment . ON cells showed an increase in spike rate at the onset of light; OFF cells showed an increase in spike rate at the offset of light; ON-OFF cells showed an increase in spike rate at the onset or offset of light . Additionally , RGC classification was correlated with morphology based on dendritic stratification in the inner plexiform layer ( IPL ) [29 , 30] . The level of stratification was defined as 0–100% from the level of the inner nuclear layer to the level of the ganglion cell layer . The stratification depth ( s ( x ) ) was quantified as a percentage of the inner plexiform layer ( IPL ) thickness , according to s ( x ) =100 ( Ls−xLs−Le ) . ( 1 ) Here , x refers to the depth of a terminal dendrite and Ls and Le represent the IPL-GCL border and the GCL-INL border of the inner plexiform layer , respectively , where depth decreases from the ganglion cell layer towards the photoreceptor layer . Cells that stratified in the inner part of the IPL ( s ( x ) ≤ 40% ) are denoted as OFF-cells . Cells that stratified in the outer part of the IPL ( s ( x ) ≥ 60% ) are referred to as ON-cells . For all cells in this study , the physiological and morphological classifications correlated well . Dendritic field sizes were calculated by tracing out a region connecting the dendritic tips of each cell and fitting an ellipse to the region . The major axis of the ellipse was used to estimate the dendritic field size . Our objective was to find a mathematical description able to accurately capture the spiking probability of RGCs to subretinal stimulation using a MEA . We characterized neural responses using an N-dimensional linear subspace of the stimulus space , combined with a nonlinearity describing the intrinsic nonlinear firing properties of neurons . Using STC analysis , we derived the lower dimension stimulation subspace that led to a short-latency response in the neuron . By projecting the raw and spike-triggered stimuli onto the lower dimension subspace , we estimated the intrinsic nonlinearity . The probability of generating a spike , given stimulus St→ , was estimated as P ( R=spike|St→ ) =NN ( v→1∙St→ , v→2∙St→ , … , v→N∙St→ ) , ( 2 ) where N represents the static nonlinear function operating on arguments in μA and v→i ( i = 1 , 2 , … , N ) represent the N significant principal components . To find v→i ( i = 1 , 2 , … , N ) , the stimulus data were first separated into a matrix containing only stimuli generating a short-latency response , SD , and a matrix containing all stimuli , ST ( Fig 3A ) . We found the low-dimensional linear subspace that best captured the difference between the spike-triggered stimuli and the raw ensemble by performing principal component analysis ( PCA ) on the covariance matrix of the spike-triggered ensemble , Cs=cov ( SD ) , ( 3 ) and comparing it to the variance of the raw ensemble which was approximately σ2 in all stimulus directions due to the Gaussian nature of ST . PCA on Cs produce a set of eigenvectors giving a rotated set of axes in stimulus space and a corresponding set of eigenvalues giving the variance of the spike-triggered ensemble along each of the axes . Eigenvalues that are greater than the variance of the input stimuli reveal the directions where the spike-triggered stimuli have experienced an increase in variance from the raw ensemble . Similarly , eigenvectors that are smaller than the variance of the input stimuli reveal directions where the spike-triggered stimuli have experienced a decrease in variance from the raw ensemble . The eigenvalues that are sufficiently different from the raw ensemble correspond to eigenvectors ( v→i , i=1 , 2 , … , N ) pointing in directions in the stimulus space that carry information about the spiking probability of the neuron . To test if the neural response could be accurately characterized by a one-dimensional model , we examined how many eigenvalues resulting from PCA were significantly different to chance [20] . We compared the eigenvalues obtained by PCA on spike-triggering stimuli to a distribution of eigenvalues for a randomly chosen ensemble of stimuli . This was done by randomly time-shifting the spike train and performing PCA on the corresponding randomized spike-triggered stimuli to recover a new set of eigenvalues . By repeating these 1000 times , we construct a distribution of eigenvalues and set a confidence criterion outside of which we presumed the magnitude of the true eigenvalues to be significant . The confidence criterion used was two standard deviations , or a 95% confidence interval . If the greatest or least eigenvalue fell outside these bounds , we rejected the null hypothesis that the spike-triggered ensemble was not significantly different to the full ensemble . We then projected out the axis corresponding to this eigenvalue from the raw data . We repeated the test until all remaining eigenvalues fell within the bounds of the null distribution , suggesting that the remaining eigenvalues were insignificant in affecting the variance of the neuron . Components having an eigenvalue significantly greater than the variance of the randomly time-shifted ensemble were considered to be components that generate an excitatory response on the cell . Conversely , components that are significantly smaller than the variance of the raw ensemble were considered to be components that suppressed the cell’s response . For the majority of cells , we found that a simplification to one-dimension ( v→1 ) accurately captured the spike-triggering information , thereby reducing the equation to one dimension . Using this simplification , Eq ( 2 ) becomes P ( R=spike|St→ ) =N1 ( v→1⋅St→ ) . ( 4 ) Results in the literature indicate that response thresholds to electrical stimulation for some cell types might differ depending on pulse polarity [37] . To explore difference in response to pulse polarity , we allowed the probability to be described by two different nonlinear functions and we found the electrical receptive fields ( ERFs ) for stimuli having a net effect that was either cathodic-first or anodic-first . Eq ( 4 ) then becomes P ( R=spike|St→ ) =N+ ( w→+⋅St→ ) +N− ( w→−⋅St→ ) +cRs , ( 5 ) where N+ and N− represent static nonlinear functions and w→+ and w→− represent the ERFs for stimuli with positive projections ( v→1∙St→>0 , net anodic-first ) and negative ( v→1∙St→<0 , net cathodic-first ) , respectively . Rs represents the spontaneous firing rate in Hz and c represents a scaling factor of units Hz-1 . To find the nonlinearities and the ERFs , the first principal component ( v→1 ) was used to divide the stimulus space into positive and negative regions by projecting all stimuli of SD and ST onto the first principal component ( Fig 3B ) . Positive and negative regions were defined by the stimuli having either a positive or negative projection onto the first principal component . This produced two spike-triggered stimulus matrices , SD+ and SD− . The means of the matrices are analogous to the spike-triggered average for net anodic-first and net cathodic-first stimuli [16] , and provide an estimate of the ERFs , w→+ and w→− , respectively . Fig 3B shows an example of the stimuli projected onto the first two principal components and the ERFs , w→+ and w→− . After the stimuli were separated into two regions , the nonlinear functions , N+ and N− , were recovered by projecting all stimuli onto the normalized vectors w→+ and w→− and segmenting the projected stimuli into 30 bins ( 15 for the net anodic-first and 15 for the net cathodic-first regions ) such that each bin contained an equal number of spikes . By comparing the number of spike-eliciting stimuli to the total number of stimuli in each bin , an estimate of the spike probability was generated . The nonlinear function from Eq ( 5 ) was then fit to the data , with the following equations for the sigmoidal curves: N+ ( x+ ) =a+1+exp ( −b+ ( x+−c+ ) ) ( 6 ) N− ( x− ) =a−−a−1+exp ( −b− ( x−−c− ) ) , ( 7 ) where x+=w→+∙St→ and x−=w→−∙St→ . Coefficients a+ and a− represent scaling factors that determine the saturation amplitudes , b+ and b− represent the gain of the sigmoidal curves , and c+ and c− represent the thresholds ( 50% of the saturation level ) for the net anodic-first and net cathodic-first stimulation , respectively . Note that the vectors w→+ and w→− might not necessarily be parallel to each other , nor parallel to v→1 . This may result in electrodes that differentially influence the neuron’s response to anodic-first or cathodic-first stimulation . To test the similarity between the positive and negative ERFs , we calculated the correlation coefficient between them . A correlation coefficient close to -1 indicated that the two ERFs are approximately equal in magnitude but opposite in sign , and therefore the cell was equally influenced by the same combination of electrodes . A value closer to 0 indicates that the two ERFs have no correlation , and therefore the cell is not influenced by the same electrodes . Positive correlation coefficients were not expected and did not occur . The spatial extent over which a cell was influenced by electrical stimulation was estimated by computing a weighted mean of the distance between the cell and the electrodes . The distance between the cell and each electrode center was weighted by the electrode’s influence on the cell as given by the ERFs . The weighted mean for both ERFs was given by , D+=∑i=120wi+di∑i=120wi+ ( 8 ) D−=∑i=120wi−di∑i=120wi− ( 9 ) where wi+ and wi− are the weights given by w→+ and w→− respectively , and di is the distance between the cell and electrode i . To test which electrodes significantly affected the cell's response , w→+ and w→− were recalculated 1000 times by projecting the data onto the first eigenvector of the randomly time-shifted distribution of eigenvectors from the significance test . A distribution for w→+ and w→− was constructed from which the true w→+ and w→− could be compared . Electrodes from the true w→+ and w→− were compared to the root mean square ( RMS ) of the distribution and electrodes that were larger than the RMS bounds were considered significant . For cells where more than one significant principal component was obtained from the significance test , we compared the variance explained by the first principal component to that of the next most significant component . This was done by comparing the separation of the first eigenvalue e1 from the mean of the randomized distribution of eigenvalues ( e¯rnd ) with the separation between the next most significant eigenvalue ( e2 ) and the same mean . The strength was defined as G=|e1−e¯rnd||e2−e¯rnd| , ( 10 ) and gives a relative measure of how much larger e1 is compared to the next most significant eigenvalue . e¯rnd was calculated from the first iteration of the hypothesis test . For each cell , 80% of the data were used to fit the model parameters , while the remaining data were used to validate the model . To obtain a quantitative estimate of the performance of the model , the probability of response given a stimulus was calculated from the validation data and compared to the model prediction . The validation stimuli were assigned 1 if they produced a direct response and 0 otherwise . Each stimulus was also assigned a predicted probability using the model ( Eq ( 5 ) ) recovered from the training data . The stimuli were then binned into segments in the range of 0 to 1 depending on their predicted probability and an actual probability for each bin was calculated by the fraction of stimuli assigned a 1 . The mean square error ( EMS ) was then calculated , EMS=1B∑i=1B ( P^i−Pi ) 2 , ( 11 ) where B is the number of bins , P^i is the predicted probability , and Pi is the calculated probability from the data for a particular bin . For all cells , B was equal to 10 . The root mean square error ( ERMS ) of the model , given by ERMS=EMS , ( 12 ) was used as a quantitative measure of the model accuracy . We also compared the error of a one-dimensional model to that of a two-dimensional model . The two-dimensional spike probability was estimated by P ( R=spike|St→ ) =N2 ( v→1⋅St→ , v→2⋅St→ ) , ( 13 ) where v→2 represented the next most significant component , either the second ( excitatory ) or last ( suppressive ) principal component . To find the two-dimensional nonlinearity ( N2 ) , a surface was fit to the spike-triggered data projected onto these two most significant components . The surface fit was obtained using a cubic spline interpolation on MATLAB’s curve fitting toolbox . Once the surface was fit , the validation data was used to calculate the mean model error calculated using Eqs ( 11 ) and ( 12 ) .
Intracellular recordings lasting up to 4 hours were obtained from 25 cells . This population included 7 ON , 13 OFF , 3 ON-OFF , and 2 cells where 3-D morphological reconstructions were not possible . Our comparison of histological and physiological results were consistent with those of Huxlin and Goodchild [29]: ON center cells stratify in the inner IPL ( 40–100% depth ) , while OFF center cells stratify in the outer IPL ( 0–40% depth ) . ON-OFF types stratify in both the inner and outer layers of the IPL . Fig 4 shows an example of an ON-OFF RGC with dendrites stratifying in both inner and outer layers of the IPL . A summary of the stratification depths for the ON , OFF , and ON-OFF cells are given in Table 1 . To fit the model parameters , cells were simultaneously stimulated with biphasic pulses on all electrodes , where the amplitude of the pulses were randomly chosen from a Gaussian distribution of zero mean and standard deviation σ ( here after white noise stimuli ) . To determine an appropriate value of σ for each cell , three short stimulus trains ( approximately 3 min each ) of white noise stimuli with different σ were initially presented to the cell ( σ varied from 50–250 μA in steps of 50 μA ) . The number of times the cell responded within 5 ms was used to obtain a response probability . Each cell responded with a different maximum response probability when stimulated with white noise at the highest value of σ; some cells could respond with a spike probability close to one , while others only responded with a spike probability less than one . However , cells that responded to fewer pulses tended to show an increased level of long-latency activity ( > 5 ms ) , most likely due to intensified network activation . The value of σ used for white noise stimulation for each cell in the rest of the experiment was the value corresponding to half the saturation level . Fig 5 shows examples of two cells with different σ values . Cell 2 responded with a spike probability close to one even at low σ values while cell 1 responded maximally with a spike probability of around 0 . 6 ( Fig 5A ) . The value of σ used for white noise stimulation for cell 1 was 85 μA and for cell 2 was 145 μA . Note that we used this method to calibrate our experiments and the nonlinear curves do not show the maximum probability of firing , as each point is an average over a variety of stimulus amplitudes . Following this calibration , longer trains of white noise stimulation ( approximately 2 minutes each ) with the corresponding value of σ for each cell were used to obtain data for recovering the model parameters . The corresponding Gaussian distributions for cells 1 and 2 are shown in Fig 5B . The experiment for each cell lasted approximately 3–4 hours . Stimulation artefacts were present in the recordings that could be removed by blanking without affecting the ability to detect the cells’ spikes . Fig 6A shows examples of some of the spiking patterns observed during experiments: ( i ) a failed anodic-first stimulus , ( ii ) a successful short-latency anodic-first stimulus , ( iii ) a successful short-latency cathodic-first stimulus , and ( iv ) a successful long-latency cathodic-first stimulus . The top panel in each subplot shows the raw recording and the bottom panels show the same signals with the artefact removed by blanking . Also shown in the bottom panels are the thresholds used to detect spikes ( horizontal lines ) . These figures show that spikes could be easily identified without interference from the stimulus artefact . The spike latencies after a stimulus pulse were analyzed for each cell . Some cells produced a bimodal distribution attributed to the short- and long-latency responses ( N = 13 ) , with four cells showing overlapping distributions for the two latencies . The remaining cells only produced short-latency spikes that were close to the timing of the stimulus pulse ( N = 8 ) . Fig 6B depicts the spike latencies for all cells . The average short-latency cluster mean for all cells was 1 . 75 ms from stimulus offset ( SD 1 ms ) . The longest short-latency cluster for a cell had a mean of 4 . 35 ms ( SD 1 . 37 ms ) . Fig 6C and 6D show the distributions of spike latencies for two sample cells , along with fitted Gaussian distributions obtained from the cluster analyses . Fig 6C shows a cell with two distinct clusters , with a short-latency cluster mean at 1 . 95 ms . Fig 6D shows two overlapped clusters with the short-latency cluster mean at 4 . 35 ms . Our aim was to find a mathematical description that could accurately capture the response probability of neurons to concurrent stimulation using a MEA . To do this we first performed a principal components analysis on the ensemble of stimuli that triggered a short latency spike . For all cells we found that the neural response could be well predicted by projection onto a subspace spanned by the first principal component , v→1 . The variance explained by v→1 was significantly higher than that of next greatest component , v→2 , suggesting that the spiking information was well captured by v→1 . Fig 7A shows the spike-triggered probabilities projected onto v→1 and v→2 from the sample cell in Fig 3B . The histograms show the number of stimuli ( gray ) and responses ( black ) along each axis; the ratio of the bars of the two histograms is used to determine the spike probabilities along each axis . From the histograms , it is clear that the distribution of the spike-triggered stimuli was bimodal in the v→1 axis; however , it remained unimodal along v→2 , similar to Gaussian distribution of the full stimulus ensemble . A statistical hypothesis test was used to determine how many eigenvalues recovered by PCA revealed a significant amount of the spike-eliciting information . The test compares the eigenvalues recovered from the data , to a set of eigenvalues produced by randomly time-shifting the spike train and performing PCA on the new set of stimuli . From the set of time-shifted eigenvalues , a 95% confidence limit was set to determine which eigenvalues from the original spike-triggered data lie outside of the limits . Fig 7B illustrates an example of the hypothesis test . For the sample cell , the test revealed that a large amount of information was contained in the first component , which was excitatory ( v→1 , eigenvalue above the confidence interval , long red arrow in Fig 7B ) . A second suppressive component was also significant , but contained a very small amount of information ( the last component , eigenvalue below the confidence interval , short red arrow in Fig 7B ) . The shaded region shows the 95% confidence intervals from the hypothesis test . The circles represent the eigenvalues obtained from PCA on the spike-eliciting data . After two iterations of the hypothesis test , all eigenvalues were within the 95% confidence intervals . The arrows shown in the figure represent the separation between the mean of the randomly time-shifted distribution of eigenvalues and the raw eigenvalues ( |e1−e¯rnd| and |e2−e¯rnd| in Eq ( 10 ) ) . For this cell , |e1−e¯rnd| was approximately 12 times greater than |e2−e¯rnd| ( G = 12 ) . Note that for this experiment , electrode 12 was not operational and hence only 19 components were produced . The bimodal distribution of response probability along the axis of the first principal component indicates that this neuron responded to two categories of multi-electrode stimulation; one category that produced a positive projection onto this axis , and one with a negative projection onto this axis . For all cells , stimuli with a positive projection onto the axis produced a stimulus at the cell’s location whose net effect was anodic-first , regardless of the fact that some electrodes may have been stimulated with cathodic-first pulses . The opposite was true for the negative projection . We therefore wondered if there may be differences in the one-dimensional stimulus subspace to which the cell responded , between net anodic-first and net cathodic-first stimulation . If so , the PCA analysis would only find the average direction of these two one-dimensional subspaces . To address this we used the PCA initial estimate of the subspace to break the stimulus space up into positive and negative regions determined by whether the stimuli had a positive or negative projection onto v→1 . By separating the data into the two regions , two electrical receptive fields ( ERFs ) , w→+ and w→− , corresponding to net anodic-first and net cathodic-first stimuli , were estimated ( Fig 3B ) . The ratio of spike-eliciting stimuli to total stimuli was then used to determine a spike probability . Fig 7C illustrates the spike probability for the sample cell . The raw data was projected onto w→+ and w→− ( × ) and the nonlinear curve from Eq ( 5 ) ( solid line ) was fit to the data . All cells obtained high r2 ( coefficient of determination ) values for the nonlinear fit; the average r2 for all cells was 0 . 92 ( SD 0 . 04 ) ( see Table 1 for summary ) . This suggests that a double sigmoidal curve is appropriate to describe the nonlinear firing probabilities of RGCs to electrical stimulation . Significant electrodes in w→+ and w→− were determined by comparing the electrodes to a distribution of w→+ and w→− generated in the signifcance test . Fig 7D shows the true w→+ ( solid black line ) compared to the RMS of the distribution ( dashed line ) . Up to three electrodes significantly affected this cell . To visualize the cell's ERF , the electrode amplitudes that generated w→+ and w→− were plotted . Fig 7E depicts ERFs for the sample cell; the green dot shows the approximate location of the recorded cell soma . The filled circles represent the stimulus amplitudes on the electrode that generate w→+ and w→− . Only significant electrodes are colored . For all cells , w→+ produced a stimulus at the cell location that was anodic-first , while w→− produced a stimulus at the cell location that was cathodic-first . In this example , the retina was oriented such that the optic disc was approximately above electrode 9 . w→+ and w→− for this cell had a correlation coefficient of -0 . 91 , indicating that the cell was influenced by the same set of electrodes when the stimulus was net anodic or net cathodic-first . An estimate of the size of the ERFs was determined by calculating a weighted mean of the distance between the cell and electrodes , where the distance was weighted by the ERFs . For this sample cell , D+ and D− were approximately equal to 1 mm , which is also the separation of the electrodes . We investigated the relationship between dendritic receptive field and ERF by comparing the morphological reconstructions to the ERFs obtained from the model . Two sample cells are shown in Fig 8A and 8B . In these images , the morphological reconstruction has been superimposed onto a photograph showing the stimulating array and the patch pipette during recordings . Using the estimate of the dendritic receptive field size obtained from the morphological reconstruction , we plotted the ERFs along with the dendritic receptive fields for 21 cells ( Fig 8C ) . Two cells where morphological reconstruction was possible were omitted due to uncertainty of the location of the cell relative to the array . The dendritic fields were estimated by a circle with a diameter equal to the major axis from the elliptical fit . The electrode colours represent the amplitude of w→+ and the stars represent the approximate location of the optic disc . One cell ( Fig 8C21 ) was only affected by cathodic-first stimulation and hence w→− is shown for this cell . This cell was affected by both anodic- and cathodic-first stimulation in its long-latency responses . Data summarizing the model fit for the population of 25 cells is shown in Fig 9 . The model nonlinear fit recovers an estimate of a cell’s thresholds ( c+ and c− in Eqs ( 6 ) and ( 7 ) ) . Most cells had similar threshold values for both their net anodic-first and net cathodic-first regions ( see Table 1 , Fig 9A ) , and no significant differences were found between or among ON , OFF or ON-OFF cell types ( t-test , p > 0 . 3 ) . The correlation coefficient of the ERFs , w→+ and w→− , for the majority of cells was close to -1 indicating that the cell was influenced by same electrodes for both net anodic-first and net cathodic-first stimulation ( Fig 9B ) . Two OFF cells had a correlation coefficient greater than -0 . 4 suggesting that the cell was differentially influenced by the electrodes depending on whether the stimulus was anodic-first or cathodic-first . The size of the positive ERF ( D+ ) was estimated for 23 cells and compared to the dendritic field size ( Fig 9C ) . Cells were significantly influenced by only one ( open circle ) , two ( closed circle ) or three ( square ) electrodes . The mean size of the ERFs produced by both w→+ and w→− was approximately 1 . 2 mm , and no significant difference was found between the size produced by w→+ and w→− ( t-test , p>0 . 4 ) . The statistical hypothesis test showed that the spike-eliciting information was mostly contained in the first PCA component ( v→1 ) . For 17 cells , some information was also contained in up to two additional components; however this information was relatively small , as the effect of omitting these higher components made little difference to the predicted result . We used a ratio G measuring how much of the variance in the response is accounted for by the first principal component ( v→1 ) compared to the next most significant component ( v→2 ) . G was generally much greater than one , signifying that for most cells a large proportion of the spike-triggered variance is contained in the first principal component . A histogram of G for all cells is given in Fig 9D , which shows that 18 of the 20 cells had a G value >4 . Cells with a single dominant principal component have spatial interactions between electrodes that are linear to a good approximation . 80% of the data was used to recover the model parameters and the remaining data was used to validate the model prediction . Fig 10A compares the validation spike probabilities and the model predicted probabilities for all cells ( grey curves ) . For clarity , we show the curves without error bars . Also shown is the model prediction for one cell with standard error bars ( black solid line ) . The model accurately predicted the responses of the RGCs to electrical stimulation . Small deviations in the prediction could likely be explained by modeling errors due to omission of some significant components . Using a different set of data to validate the model gave slightly different deviations , and slightly different estimates of the error . However , in all cases the model still accurately predicted the responses . The average ERMS for all cells was 6 . 3% error , with a maximum error of 11 . 7% ( see Table 1 ) . There were 17 cells that recovered two or more components that fell outside the 95% confidence interval in the statistical hypothesis test . To test for improvements in the prediction by including higher components , we fit a two-dimensional surface to the probability data ( e . g . Fig 7A ) . The same validation data used from the one-dimensional model was used to compute the error from a two-dimensional model . Note that for some cells the training data was under-sampled in regions that were sampled with the validation data . These points were omitted when calculating the model error . The model error for the two-dimensional model ( ERMS2 ) was compared to the error from the one-dimensional model ( ERMS1 ) . For most cells , little improvement was found in model error , and a few cells resulted in a higher error ( Fig 10B ) . A slightly higher error for a few cells is likely due to over fitting to increased noise , which occurs due to undersampling of the two-dimensional surface fit . For the two cells that had a low value of G , the model error was reduced by approximately half ( 9 . 8% to 4 . 2% ) for one cell and changed very little for the other ( 3 . 3% to 3 . 5% ) ( See the two circled points in Fig 10B ) . We examined if the model could also be applied to the long-latency responses to predict responses that were most likely of presynaptic network activity . Understanding the secondary effects of electrical stimulation is important in a clinical setting to understand differences between the perceived and expected responses . Responses originating from presynaptic origin can have excitatory or suppressive effects on postsynaptic RGCs . Fig 11A illustrates the positive ERF ( w→+ ) for the sample cell ( same cell from Fig 7 ) that had an excitatory long-latency response . The negative ERF was almost the same in magnitude and location as the positive ERF and hence is not shown . In this preparation , the optic disc was placed above electrode 9 . It is this electrode that most strongly influences the long-latency response in this neuron . The accuracy of the model was assessed in the same way as the model error for the short-latency responses . For this cell , the model predicted the long-latency response accurately ( error approximately 7% ) ( Fig 11B ) . It is evident from the corresponding eigenvalues ( Fig 11C ) that there is an excitatory and suppressive component that affects the long-latency responses . Excitatory or suppressive effects applied through the retinal network can be investigated by analysis of long-latency responses . Fig 11D illustrates an example of a cell that was very responsive in its short-latency responses; however , it became suppressed by high amplitude stimulation in its long-latency responses . The corresponding eigenvalues are shown in Fig 11E . Although our results on short-latency responses showed that RGCs were largely indifferent to the pulse polarity , analysis on the long-latency responses could produce responses that favored a particular pulse polarity . Fig 11F shows an example of a cell that was sensitive to both polarities of short-latency stimulation but its long-latency response resulted largely from cathodic-first stimulation . The corresponding eigenvalues are shown in Fig 11G . When little is known about the neural system , a naive stimulation strategy might be to activate multiple nearby electrodes such that the amplitude of stimulation across the electrodes is equal . However , the ERF recovered from the model gives an insight into the stimulus that improves the efficacy of a response in the neuron . To compare a naive stimulation strategy to stimulation using currents on electrodes that are proportional to w→+ , we used the recovered model to compare the response probabilities for both strategies . Fig 12A compares stimulation on one , two , or three of the electrodes closest to the sample cell , to stimulation with currents proportional to w→+ . To make an unbiased comparison , comparisons were made while keeping the total power fixed . Since all of the electrodes were of the same geometry , this was equivalent to keeping a constant norm on the stimulation vector St→ . For this example , a stimulus on three of the closest electrodes , where the current amplitudes on each electrode were equal , resulted in better efficacy than stimulating on only one or two electrodes . However , the efficacy was further improved when stimulating proportionally to w→+ . Fig 12B compares the threshold from a naive strategy ( SN ) , to the threshold of a stimulus proportional to w→+ for all cells . Here , we compare only SN resulting in the lowest threshold , single electrode ( star ) , two electrodes ( triangle ) , or three electrodes ( circle ) . In all cases , stimulation proportional to w→+ results in a higher efficacy for a given power . On average , stimulation proportional to w→+ resulted in a threshold 0 . 8 ( SD 0 . 2 ) times the threshold of the SN resulting in the lowest threshold .
The model we present here was adapted from well established Gaussian white noise models developed to describe light responses in the retina [16 , 18 , 20 , 22] . With minor modifications , we have shown the application of this type of model to describe responses to electrical stimulation . The model is scalable and can be used to also describe retinal responses to electrical stimulation using small , high density electrodes , or to describe long-latency responses . RGC responses to concurrent electrical stimulation across multiple electrodes could be accurately modeled by a nonlinear transformation of a linear spatially filtered stimulus . Simultaneous biphasic pulses were applied to all electrodes , with the stimulation amplitude on each electrode randomly sampled from a Gaussian distribution . The model's linear filter characterizes the neuron's electrical receptive field . The nonlinear function characterizes the neuron's intrinsic nonlinear firing properties . Stimulus-evoked spikes in the recorded neurons were analyzed using principal component analysis to determine the linear filter and reduce the dimensionality of the spike-triggered stimuli . For most cells a single linear filter was sufficient to predict the neural response to a good approximation , indicating that interactions between concurrently stimulated electrodes are predominantly linear . High coefficients of determination for the nonlinear function fits were obtained ( lowest r2 = 0 . 83 ) , demonstrating that the double sigmoid function is an accurate description of the nonlinearity . The model was trained with 80% of the data , with the remainder used for validation . The spike probability from the validation data was compared to the predicted probability , which resulted in an average error of 6 . 4% across the population ( maximum error 11 . 7% ) . For many cells , short-latency responses were within 2–3 ms from the stimulation offset . While the origin of the spikes were not investigated , the latencies are consistent with latencies attributed to direct activation of RGCs in response to 1 ms pulses [38] . Four cells showed overlapping short- and long-latency clusters with spike latencies of up to 6 ms ( Fig 6D ) . It is possible that some of the spikes in the four cells with overlapping clusters might have had a mixture of direct and indirect ( network mediated ) activity . Despite this , the model was able to accurately predict the response . The technique could also be applied to investigate the long-latency responses driven by synaptic activity . The effects seen in the long-latency responses can be excitatory , suppressive ( also observed in [38] ) , or polarity-selective . Importantly , a separate analysis of long-latency responses produced distinct electrical receptive fields compared to the short-latency responses . Long-latency responses in the retina are mediated via the activation of retinal interneurons and might result in high acuity vision [39] . The techniques described can be used to gain a deeper understanding of the retinal network and the effects of electrical stimulation at distant sites . Investigation into the long-latency responses can give insight into the secondary effects of stimulation and how these might influence perception . Models recovered from white noise stimuli have been used to characterize light responses in the retina [16 , 18 , 20] and cortex [40] , electrical responses in the retina [26 , 41] , subthreshold responses in squid axons [42] , and to characterize information transfer from the sensory periphery [43] . The advantage of estimating models with Gaussian white noise is that the neurons can be presented with a wide range of possible inputs and adaptation is reduced compared to more regularly structured stimuli . These properties make these models more suitable for characterizing neural responses to spatiotemporal patterns of electrical stimulation . Additionally , analysis techniques for white noise stimulation have been extensively explored in the retina to describe light responses [16 , 18 , 20 , 21] . We have used a spike-triggered covariance model and demonstrated that it can be accurately applied to describe electrical responses . Jepson et al . [44] demonstrated the versatility of a piecewise linear model in capturing neural response probabilities to electrical stimulation . Their study used a high-density array with small electrode diameters and combined stimulation across two or three electrodes to achieve spatial selectivity . Only fixed ratios of stimulus amplitudes were explored . In contrast , our stimuli consisted of Gaussian white noise that allowed the exploration of a vast range of stimulus inputs across all available electrodes , to find an estimate of the cell’s ERF . This has the potential to be more efficient when simultaneously recording from multiple neurons , as the same stimuli can be used to generate the model parameters for all recorded neurons . Our study also used large diameter electrodes due to their relevance to clinical visual prosthesis stimulation arrays [45] . No strong correlation between the area over which the cell was affected ( D+ or D− ) , and the size of the dendritic field was found . It is possible that much of the relationship is lost when using large diameter electrodes , or that the stimulation was largely axonal . The spatial ERFs were generally as might be expected: i . e . the electrodes closest to the cell significantly affected the response . However , some unexpected results were apparent . For example , cells in Fig 8C17 and 8C18 both had a small dendritic field and they were located in a similar location in relation to the stimulating array . However , cell 17 only had two significant electrodes , whereas cell 18 had three . While the origin of these differences are unknown , complexities in ERF shapes have been previously observed [39 , 46] . As suggested by Sim et al . [46] , the complex shapes could be due to axonal stimulation . Our results demonstrate the utility of our technique in identifying even complex ERFs . Freeman et al . [26] explored single electrode stimulation of the retina using binary white noise to recover a temporal spike-triggered average stimulus . This study demonstrated that electrical white noise models can be used to estimate the temporal relationship between the stimulus and response , while our model describes the spatial relationship across electrodes . We assume that the effect of temporal interactions at 10 Hz is small , an assumption that might account for some of the error in prediction . A model that also incorporates temporal effects is desirable , especially when considering higher stimulation frequencies , but technical challenges remain . A similar model to Chichilnisky [16] could be modified for electrical stimulation and incorporate spatiotemporal ERFs , but the accuracy of the model would decrease as the number of frames increased . The stimulus artefact could also be a problem when stimulating at high frequencies . To obtain longer recordings , one solution is to record extracellular potentials , which is also the only practical solution for patient testing . In this case , the stimulus artefact would be larger than the spike signal , making spike identification more challenging . At low stimulus frequencies , the stimulus artefact could be removed by artefact removal techniques [39 , 47 , 48] , allowing detection of short-latency spikes . A major goal of our work is to develop models that can be applied in real-time , closed-loop applications . The applications of closed-loop systems to modern technology are vast . There are several potential advantages to the development of neuroprostheses that make use of neural feedback . An obvious advantage of neural feedback is much tighter control of evoked neural activity , when that activity can be measured and the stimulus adjusted to match a desired outcome . A second advantage is automation of patient fitting procedures , minimizing the need for time-consuming psychophysics . Furthermore , many stimulation algorithms are limited to stimulation with one electrode at a time , in part because the time required to test myriad possible combinations of simultaneous electrode stimulation is prohibitive using a psychophysical approach . Closed-loop neural stimulation models can also take advantage of control theory and can be designed to adapt to changes in the system being controlled . Open-loop strategies cannot adapt to changes such as changes at the electrode-tissue interface . The model we present is simple and appropriate for real-time computation; however , technical challenges remain . A requirement of in vivo or patient tests of closed-loop control is to obtain high density extracellular neural recordings . Devices that can combine stimulation and recording on the same electrode need to balance high surface area and charge capacity for stimulation , with electrodes of low geometric surface area for single-unit recordings [45] . Recent new materials have led to the development of flexible electrode wires capable of stimulation and recording [49] . However , a high density array capable of stable recordings and stimulation remains to be developed . Our model can be extended by increasing the number of recording electrodes , to describe the response of multiple neurons across the array , thus making it a multi-input multi-output system . Multi-input multi-output control is a widely researched area of control and could be applied to achieve patterns of activation across the array . In a visual prosthesis , a desired pattern of activation could be obtained from RGC activation models in response to patterned light [16 , 50] . Our model can be fit to individual patients based on recorded responses and used to develop control strategies that are patient specific . Devices that can record and stimulate can then be used to try and address some of the more complex problems in the field , namely that of relating stimulation to visual percept [51] . | Implantable multi-electrode arrays ( MEAs ) are used to record neurological signals and stimulate the nervous system to restore lost function ( e . g . cochlear implants ) . MEAs that can combine both sensing and stimulation will revolutionize the development of the next generation of devices . Simple models that can accurately characterize neural responses to electrical stimulation are necessary for the development of future neuroprostheses controlled by neural feedback . We demonstrate a model that accurately predicts neural responses to concurrent stimulation across multiple electrodes . The model is simple to evaluate , making it an appropriate model for use with neural feedback . The methods described are applicable to a wide range of neural prostheses , thus greatly assisting future device development . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"action",
"potentials",
"engineering",
"and",
"technology",
"signal",
"processing",
"membrane",
"potential",
"ocular",
"anatomy",
"retinal",
"ganglion",
"cells",
"electrophysiology",
"neuroscience",
"surgical",
"and",
"invasive",
... | 2016 | A Simple and Accurate Model to Predict Responses to Multi-electrode Stimulation in the Retina |
Many neurodegenerative diseases have a hallmark regional and cellular pathology . Gene expression analysis of healthy tissues may provide clues to the differences that distinguish resistant and sensitive tissues and cell types . Comparative analysis of gene expression in healthy mouse and human brain provides a framework to explore the ability of mice to model diseases of the human brain . It may also aid in understanding brain evolution and the basis for higher order cognitive abilities . Here we compare gene expression profiles of human motor cortex , caudate nucleus , and cerebellum to one another and identify genes that are more highly expressed in one region relative to another . We separately perform identical analysis on corresponding brain regions from mice . Within each species , we find that the different brain regions have distinctly different expression profiles . Contrasting between the two species shows that regionally enriched genes in one species are generally regionally enriched genes in the other species . Thus , even when considering thousands of genes , the expression ratios in two regions from one species are significantly correlated with expression ratios in the other species . Finally , genes whose expression is higher in one area of the brain relative to the other areas , in other words genes with patterned expression , tend to have greater conservation of nucleotide sequence than more widely expressed genes . Together these observations suggest that region-specific genes have been conserved in the mammalian brain at both the sequence and gene expression levels . Given the general similarity between patterns of gene expression in healthy human and mouse brains , we believe it is reasonable to expect a high degree of concordance between microarray phenotypes of human neurodegenerative diseases and their mouse models . Finally , these data on very divergent species provide context for studies in more closely related species that address questions such as the origins of cognitive differences .
Here we compare and contrast gene expression in three different regions of the human brain , motor cortex , caudate , and cerebellum , to identify genes that are differentially expressed between the regions . In other words , we seek to identify genes that show patterned expression . Knowledge of such regionally enriched genes may provide insight into the development and biochemistry of different brain structures . This information may also hold potential biomedical implications . Many neurodegenerative diseases , such as Huntington's disease , have a hallmark regional and cellular pathology affecting one or another of these regions while sparing the others . It is reasonable to assume that unique susceptibilities in disease may relate to distinctive brain gene expression patterns in health . We also perform a parallel analysis on the functionally and anatomically corresponding regions of the mouse brain , anterior cortex , striatum , and cerebellum . This allows us to begin to compare and contrast patterns of gene expression in these tissues across these two species . Our motivation for this cross-species analysis also has a biomedical consideration . In recent years , mice have become the most important model organism for human neurological and neurodegenerative diseases . The brains of humans and mice are clearly different with respect to size , complexity , and cognitive abilities . The belief that mice can accurately model human neurodegenerative or neurological diseases rests on assumptions about deeper biological similarities between mouse and human brains that have not been systematically examined . While it is impossible to directly compare mouse arrays to human arrays , one may compare patterns of gene expression in several corresponding brain regions . Comparing gene expression patterns is one way to obtain objective and global information on how similar the brains of humans and mice are . A practical use for this comparative cross-species gene expression information is as a baseline for the comparison of microarray phenotypes of human diseases and their mouse models . For example , if expression changes in a human disease and its mouse model have a global correlation of r = 0 . 5 , is this the best possible correlation that can be expected or might we reasonably expect more ? Obviously , to answer this question it is useful to know what sort of correlation array data from mice and human brains have initially . Comparative cross-species information of brain gene expression would also seem to have natural implications for studies that address the origin of cognitive differences between humans and other species . It is generally believed that with respect to other species , even our closest relatives the chimpanzees , humans have unique abilities pertaining to language and higher order cognitive functions . Sequencing projects have revealed that the human and chimpanzee genomes are ∼96% identical , that their typical protein amino acid sequences are ∼99% identical , and that both species have essentially the same number of genes [1] . Since an increase in genomic complexity seems inadequate to explain the apparent mental differences between humans and chimpanzees , the idea that these differences may be due to changes in gene expression has attracted new attention [2] . Recent microarray studies have sought to identify general trends in brain gene expression that distinguish humans from other primates [3–10] . The conclusions reached by these studies have been somewhat discordant . One study found more expression differences between human and chimpanzee liver than prefrontal cortex , but by using orangutan liver and cortex as out groups , concluded there had been an accelerated rate of change in brain gene expression during human evolution [3] . A reanalysis of this data supported this interpretation of rapid and recent human evolution [6] . Other investigations have found that elevated levels of gene expression further distinguish the human brain from that of other primates [4–6] . Several studies have cast doubt upon these findings , attributing them to improper array normalization or to hybridization artifacts rooted in measuring nonhuman primate expression with arrays designed for human sequences [7–10] . Since we are comparing human and mouse expression , we cannot make definitive statements regarding differential expression between humans and chimpanzees . However , examination of global similarity of expression patterns in species as divergent as mice and humans can provide useful context for studies that aim to correlate expression changes with cognitive differences between more closely related species . Presumably , if gene expression patterns distinguish human brains from the brains of other primates , then changes due to recent human evolution may be even more apparent when comparing humans to mice .
The first part of our analysis focuses on expression in different regions of nondiseased brain . We determined absolute and relative gene expression in three anatomically distinct regions of human brain: motor cortex ( Brodmann area 4 , BA4 ) , caudate nucleus , and cerebellum . The data consisted of samples of all three tissues from 12 donors . These samples constituted a portion of the control group in a study comparing gene expression in Huntington's disease and non-Huntington's brain [11] . For the present reanalysis , the 36 arrays were normalized using Robust Multiple-array Average ( RMA ) [12] . To assess differential expression between brain regions , three sets of paired t-tests were performed; caudate-to-cerebellum , BA4-to-cerebellum , and BA4-to-caudate using the Bioconductor package LIMMA ( http://www . bioconductor . org/packages/1 . 9/bioc/html/limma . html ) [13 , 14] . To confirm the primary human data , we used a second set of caudate and cerebellum samples from nine different donors [11] . Because of the original study's design , there were no motor cortex samples from these nine donors . The absolute and differential expression analysis of the human samples is provided in Datasets S1 and S2 . A key that associates samples with GEO accession numbers , age , gender , post mortem delay , and other covariables can be found in Table S1 . Comparing the log2 ( fold change ) ( i . e . , log ratio ) for the replicate caudate-to-cerebellum comparisons indicated that these independent data were highly correlated , with Pearson's correlation coefficient r = 0 . 93 ( Figure S1 ) . In the primary caudate-to-cerebellum comparison , 9 , 088 probesets met p < 0 . 001 with respect to differential expression . In the smaller secondary dataset , 8 , 074 probesets met p < 0 . 001 . Of these , 82% ( 6 , 589/8 , 074 ) met p < 0 . 001 in both comparisons , and only four probesets showed discordant directions of change . These results demonstrate that the caudate and cerebellum have quite distinct gene expression profiles . They also show that the relative differences between the regions were robust and reproducible in these post mortem human samples . This is consistent with results from a detailed analysis of the relationship between prehybridization variables and posthybridization assessments of data quality , which found little negative contribution from post mortem interval to data quality in these samples [15] . We next performed identical comparative analyses of anterior cortex , striatum , and cerebellum samples from six five-week old wild type C57BL/6 mice . It is generally accepted that these mouse brain regions are anatomically and functionally homologous to human motor cortex , caudate , and cerebellum respectively . We used young mice since very often identifying the earliest changes in a mouse neurological disease model is of primary experimental interest . The complete mouse RMA and regional comparison data are provided in Datasets S3 and S4 . Counts of probesets meeting p < 0 . 001 for differential expression in one region relative to the others for both the mouse and primary human data are shown in Table 1 . As was found with the human analysis , the three mouse brain regions examined had very distinct gene expression profiles with many statistically significant changes ( Table 1 ) . To provide additional verification of the data , we queried the human and mouse caudate/striatum-to-cortex and caudate/striatum-to-cerebellum comparisons against a published list of 54 striatum-enriched mouse genes ( Table S2 ) [16] . Table 2 shows both the human and mouse array data to be consistent with known mouse striatal genes ( p ≈ 10 ) . Table 3 shows the 30 named genes with the highest regional scores ( see Materials and Methods ) in each species . At their most extreme , the pattern of expression for these genes is “on” in one region and “off” in the other two regions . Table 3 represents only a small subset of regionally enriched genes , and the complete data pertaining to differential expression between brain regions can be found in Datasets S2 and S4 . Several genes appear in both the mouse and human lists . Even considering these short lists of top regional genes , the intersections between human and mouse gene lists are highly statistically significant ( p < 10−7 for each intersection ) . While the primary interest of this study was in regional differences , many factors such as age , gender , tissue heterogeneity , post mortem interval , medication , and cause of death may influence gene expression in the brain and contribute to differences between individuals . To examine the effect of individual variability of the gene expression on the profiles , the between-tissue and within-tissue variances for each probeset were computed from the human RMA signals . This was repeated for the mouse probesets . As post mortem delay was not a concern with the mouse samples , and all of the mice were the same age , mouse individual variability might reasonably be expected to be smaller than human variability . We found that the between-tissue variability was greater for 89% of the human probesets and 85% of the mouse probesets . This suggested that human individual variability in gene expression and factors such as tissue heterogeneity and post mortem interval were not obscuring or significantly contributing to regional differences . It also suggested that compared to expression dictated by regional identity , age and gender appear to have effects of small magnitude or of large magnitude on a small fraction of genes , even in humans . Some evidence for this can also be inferred from the two independent human caudate-to-cerebellum comparisons . In these comparisons , age and gender were not balanced between the groups , yet relative expression levels were highly correlated and the slope of the regression line was 0 . 967 ( Figure S1 ) . To explore the gene expression of each tissue in more depth , we used gene ontology ( GO ) [17] . GO provides means of objectively identifying functional themes in large groups of genes , in this case the genes that were differentially expressed in the pair-wise regional comparisons within each species . A significantly high number of overrepresented GO categories were found for both human and mouse in all three types of regional comparisons . This was true whether considering increased or decreased probesets separately or together . While GO is not intended for rigorous assessment of evolutionary relationships , GO nomenclature is standardized . This allowed us to compare and contrast the regionally enriched functions in the homologous mouse and human brain regions . Of the hundreds of GO categories differentially represented in one region relative to another , many were common between the corresponding human and mouse comparisons ( Tables 4 and S3 ) . Permutation testing showed these intersections to be significantly greater than would be expected by chance ( p < 0 . 0001 ) ( Table S4 ) . Both the functional GO analysis and the intersections among top regional marker genes hinted that relative gene expression levels across brain regions have been conserved between mice and humans . To examine this in depth , it was necessary to contrast expression ratios on a gene-by-gene basis across the human and mouse arrays . This was complicated by the fact that genes are often represented by more than one probeset on each array . To lessen this complication for our initial analyses , genes were identified where only one probeset existed on each array . We also arbitrarily required that human and mouse gene symbols were identical , since it was a clean and simple way to identify genes that have met widely accepted criteria for being orthologous pairs . Using these criteria , 2 , 998 one-to-one orthologous pairs were found on the mouse and human arrays . Taking this set of genes , we first asked whether genes with high variance of expression across the brain regions of one species would cluster the entire set of samples sensibly . This was motivated by our earlier observation that the largest component of a gene's expression variance was due to tissue specificity . Thus , high variance implied patterned expression across the three brain regions . Figure 1 shows that both the mouse and human genes with the largest variance in the one-to-one gene set cluster all of the samples perfectly , first by tissue , then by species . In other words , for these three brain regions , the equivalent human and mouse regions are more alike than different regions within a species or individual . We also note that while we selected the genes based on variability of expression across regions within a species and not conservation between species , there were 43 genes in common on the two lists of 125 most variable one-to-one genes ( p ≈ 0 ) . The heat maps of normalized expression indicated that relatively few genes in corresponding brain regions were on opposite sides of their mean signal within a species . All of these observations suggest a high degree of similarity in the genes with patterned expression in mouse and human brain . Using all genes in the one-to-one set , we next examined relatedness of regional gene expression within and between species by computing normalized Euclidian distances between all possible nonself pairs of tissues ( Table 5 ) . The similarity of corresponding tissues between the species was apparent by their consistently having the minimal between-species distance . The pattern of distances between regions within the human brain was essentially identical to the pattern of distances within the mouse brain , suggesting that no single region of the human brain had diverged from the other two regions any more than regions in the mouse brain had diverged from each other . To expand our analysis beyond the one-to-one subset of genes , ENSEMBL ( http://www . ensembl . org ) information was used to identify a more complete set of mouse-human orthologs . Where more than one probeset represented a gene , we retained only information pertaining to the probeset with the highest mean RMA signal . This collapsed the arrays to a nonredundant set of 8 , 499 genes common to both array types ( Dataset S5 ) . We then correlated log ratios in the appropriate pairs of tissue comparisons over all the genes . The correlation coefficient of the mouse and human caudate-to-cerebellum log ratio was r = 0 . 47 . For the cortex-to-cerebellum comparisons and for the cortex-to-caudate comparisons , r = 0 . 45 . We explored the hypothesis that genes whose sequences had been under stabilizing selective pressure would also be constrained in their pattern of expression . Information about nonsynonymous and synonymous amino acid substitution ratios ( dN/dS ) and percent nucleotide identity for mouse and human orthologs were retrieved from the ENSEMBL database . The set of 8 , 499 orthologous genes was ranked by each of these metrics , and a correlation coefficient between appropriate human and mouse log ratios was computed for each quartile of genes . The quartile-based correlation coefficients are plotted for each class of tissue comparison in Figure 2 . This shows that there is a positive relationship between conservation of sequence and conservation of expression . It seems natural to assign greater confidence in a pairing between two genes that are 95% conserved at the sequence level than between two genes that are 75% conserved . Furthermore , as homology thresholds decrease , the number of potential ortholog pairings increases . Because of these factors , we assume that our rate of incorrectly pairing orthologs may increase as percent nucleotide identity decreases . Pairing errors also likely reduce correlation , thus any bias in the error rate of pairing may introduce a false positive relationship between homology and gene expression . To avoid this potential bias , we examined the relationship between variability of expression across tissues and sequence conservation . Results presented above suggest that in mouse and human brain , the genes with the greatest variability of expression in the three examined brain regions were similar ( Figure 1 ) . We also showed that expression variance was most strongly dependent upon tissue specificity rather than variability between individuals . Variance within a species can be determined in the absence of homology information , so we examined the within-species variance of bins of genes with integral percent identities . Figure 3 shows that there is a clear tendency in both species for genes with higher expression variance across brain regions to have higher identity with their orthologs . Since expression variance is a surrogate for tissue specificity , this indicates that region-specific genes in the brain tend to have greater homology with their orthologs than more widely expressed brain genes . This is consistent with the idea that functional constraints have applied selective pressure on brain gene expression since the mouse and human lineages diverged some 80 million years ago .
Our data indicate that expression patterns across comparable regions of human and mouse brains have generally been conserved since the two lineages diverged . This is consistent with classical comparative neuroanatomy , which has long indicated general conservation of gross mammalian brain structure and conservation of cell types within comparable regions [18 , 19] . Conservation of patterned gene expression in the mammalian brain is consistent with standard assumptions of biological uniformity justifying the use of model organisms . Further underscoring conservation of mammalian brain gene expression , we find that in the three brain regions examined , equivalent regions in mouse and human brain are more alike than different regions within a species . This is apparent whether considering the 125 genes with the most variable expression within a species ( Figure 1 ) or whether considering Euclidian distances based on expression of thousands of orthologous gene pairs ( Table 5 ) . Our finding is consistent with other studies contrasting brain gene expression in dogs and humans [20] , chimpanzees and humans [7] , and mice and humans [10] . We do not mean to suggest , and our findings should not be interpreted to mean that gene expression in human and mouse brains is identical . Here we are mainly concerned with the genes that show an extremely patterned expression across three particular brain regions . Because our study examines expression of the tissue , we cannot discern evolutionary changes within specific cell types . However , within the three regions examined here , the overall trend is for regional-marker genes to have been conserved . For example , considering the human motor cortex-to-cerebellum comparison , in the 100 human genes with greatest evidence for differential expression between these regions , there are only nine discordant changes in the 100 mouse orthologs . The overall correlation between human and mouse log ratios in this set of 100 genes is r = 0 . 86 . In the top 250 human caudate-to-cerebellum changes , r = 0 . 81 , and there are 39 discordant changes . In the top 500 changes , r = 0 . 75 , and there are 93 discordant changes . Essentially identical correlations and trends appear in cross-species correlation of the other two regional comparisons ( unpublished data ) . It is very likely that our data somewhat underestimate the true correlation , since factors such as post mortem delay , tissue dissection , and gender ratios were not strictly controlled . Other technical sources of variability include possibly measuring different splice variants in mice and humans , comparing young mice to old humans , and differences in cell-type composition arising from comparing whole mouse tissues to small portions of the human tissues . Finally due to evolution of genomic sequence , Affymetrix ( http://www . affymetrix . com ) must almost always use probes of different sequences to assay human and mouse gene expression . Probe sequence has a profound influence upon the signal detected in a microarray experiment [21] . Overall we observe correlation of relative expression levels in mice and humans on the order of r = 0 . 45 . This leaves the proportion of unexplained variance due to technical factors and evolutionary changes as roughly 80% ( 1 − 0 . 452 = 0 . 8 ) . If an estimate of the variance due to technical factors can be made , in theory it is possible to determine the proportion of unexplained variance due to the evolution of gene expression . From the correlation observed in our independent human regional comparisons , one can arrive at an estimate of 14% of the variance being due to technical noise for a within-species regional comparison ( 1 − 0 . 932 = 0 . 14 ) . Perhaps twice this or 28% may serve as an estimate of cross-species technical noise . Thus at the high end , our data suggest 52% ( 80%−28% ) of the variance could be due to evolutionary changes . It may be more accurate to suppose that cross-species correlations are subject to the same technical noise effects as within-species correlations on different generations of Affymetrix microarrays . In that case , typically r = 0 . 7 [21] . Therefore , the estimate of variance due to noise is 1 − 0 . 72 = 0 . 51 , which leaves 0 . 8 − 0 . 51 = 0 . 29 or 29% as our estimate of the unexplained variance due to evolution of expression in mice and humans . Since we are examining log ratios , evolutionary contributions from both tissues in each species are combined in this number . We presume the true variance due to evolution within each single tissue would be less than 29% . It might be reasonable to expect that gene expression variability would be significantly larger between individual humans than between inbred mice housed in uniform conditions . There is little evidence for this in our profiles . We find the fraction of genes that vary between individuals more than between tissues is roughly the same in the two species . These findings could be unique to the regions examined , or they may be a consequence of the between-region variability being so much larger than individual variability for both humans and inbred mice . A more interesting alternative is that this implies that the constraints on brain gene expression are quite strict and that many commonly presumed sources of individual variability are just not that influential . Gender may be one of the largest contributors to individual gene expression variability . In analyses to identify gender-dependent gene expression differences in human brains , differential expression was limited to a rather small set of genes when the X and Y chromosomes were excluded ( L . Jones , unpublished data ) . Examining orthologous mouse and human genes , we find that conservation of amino acid and nucleotide sequence is correlated with conservation of regional expression . Since this relationship could have been an artifact of our ability to identify homologous genes , we re-examined this relationship by beginning with genes that showed evidence for regional expression within one or the other species . This showed that the genes with higher variability of expression between brain regions within a species also tended to have greater sequence homology with their orthologs than genes that are expressed in multiple brain regions . This is somewhat surprising if one imagines that evolutionary constraints act additively on genes widely expressed in different tissues . It may be that regional gene expression in the brain is particularly highly constrained since the proper behavior of the organism depends upon each brain region functioning smoothly with the others . Wider surveys of tissue gene expression tend to support constraints on brain gene expression , finding that the brains of humans and chimpanzees show fewer differentially expressed genes than kidney , heart , liver , and testes [3 , 22] . Particular interest has been devoted to finding differences between humans and chimpanzees . Some studies have concluded that there is a bias for genes to be more highly expressed in human cortex relative to chimpanzee [4 , 6] . While our data cannot directly address chimpanzee and human gene expression , and this claim was made for a rather small number of genes , we see little evidence that the human cortex has uniquely undergone extensive and rapid evolutionary change . Based on the Euclidian distances shown in Table 5 , we find it is the cerebellum that is the outlier tissue both within and between the two species . It is quite possible that complexity of higher order brain functions relate to splicing or protein modifications that escape microarray analysis , but our data suggest some boundaries on the idea that gene expression differences explain differences in cognitive abilities between species . Few genes appear to have evolved new patterns of regional expression . The minority of genes that do show discordant regional expression between adult mice and humans may indeed be key genes regulating brain functions . Alternatively , since general expression patterns in the adult brain have largely been conserved , perhaps it is gene expression during development that ultimately wields the most influence upon higher brain functions by specifying the complexity of neuroanatomy . Humans have at least two orders of magnitude greater numbers of neurons and neuronal connections than mice [18 , 19] . Our data suggest the active genes in those neurons and connections are quite similar in adult mice and humans , species with extremely different cognitive abilities . This similarity should become greater as more closely related species , such as chimpanzees and human , are considered . The most important genes relating to cognitive differences may be genes that specify how the machinery is assembled . Transgenic mice have become the most common model organism for human neurodegenerative diseases [23] . Scrutiny of models has previously involved comparing histopathological and neurochemical phenotypes , or extrapolating from mouse neurobehavioral tests to human disease signs and symptoms . We suggest that the transcriptional signature of the human disease can be used to objectively and globally assess both genetic and phenotypic models; the assumption being that a model that recapitulates the human disorder should have a similar expression profile . Ideally , such assessment involves reference to a range of expression profiles so that the biological specificity of the disease phenotype can be addressed and to provide outlier groups to place relatedness in context [24] . We believe that contrasting healthy mouse- and human-brain gene expression profiles provides a reasonable context with which to assess likeness between mouse models and human neurodegenerative diseases . The high correlation between regional gene expression in healthy brain suggests that mouse models of human neurodegenerative diseases may quite accurately recapitulate the human microarray phenotype and should be held to a high standard . Here we have focused on the general similarity rather than specific differences between two species . Using several different methods , we find that regional gene expression in the mouse anterior cortex , striatum , and cerebellum is very similar , respectively , to gene expression in human motor cortex , caudate , and cerebellum . Classical comparative neuroanatomy has identified a general conservation of mammalian brain structure , with differences between species arising from elaboration of ancestral forms . Our data indicate that this general conservation continues down to the gene expression level , and that expression patterns in our brains may be less far removed from ancestral forms than apparent differences in mental abilities might suggest .
Post mortem human tissue was gathered with ethical approval and permissions , dissected , and processed as specified [11] . The samples were hybridized to Affymetrix HG-U133A arrays containing 22 , 283 probesets . The primary dataset consisted of caudate , cerebellum , and motor cortex samples from eight men and four women , whose ages ranged from 36 to 77 with an average age of 58 years . Confirmation of the primary data was performed with an independent second group that consisted of caudate and cerebellum samples from seven men and two women whose ages ranged from 22 to 72 with an average of 49 years . Clustering included all human and mouse samples . Postnatal day-35 C57BL/6 mice , five females and one male , were killed by cervical dislocation . The brain was immediately dissected into ice-cold phosphate-buffered saline . Tissue microdissections were performed at 4 °C on one hemisphere at a time with the brain on a bed of dry ice . The cortex was divided into an anterior and posterior portion with the line of division at the point where the striatum and hippocampus meet . Tissue was collected into 5-ml polypropylene Falcon tubes , submerged in liquid nitrogen , and stored at −80 °C . Total RNA was isolated by adding 1 ml of Qiazol reagent ( Qiagen , http://www . qiagen . com ) to each frozen sample and homogenizing the tissue with a polytron for 40 s at medium speed . Residual salts and proteins were removed with an RNeasy Lipid Kit per the manufacturer's instructions ( Qiagen ) . RNA concentration was determined with spectrometer . The Affymetrix single-cycle probe synthesis kit was used to generate cRNA probe per the manufacturer's instructions . For the cortical and cerebellar samples , 5 μg of total RNA was used as starting material . For striatal samples , 2 μg of total RNA was used . Biotinylated-cRNA was checked on a bioanalyzer prior to and after the fragmentation reaction . Samples representing tissue from a single mouse were hybridized to MOE_430A_2 chips containing 22 , 690 probesets , n = 6 for each tissue . The raw image data is available at http://www . hdbase . org . Primary analysis of microarray data was performed using Bioconductor , a freely available software package designed for the analysis of genomic data ( http://www . bioconductor . org ) . We first preprocessed and normalized the CEL files with RMA . The primary and secondary groups of human samples were normalized and analyzed separately . Then we fit a linear model ( gene expression ≈ donor + tissue type ) for each of the three paired comparisons of tissue using the Bioconductor library package LIMMA to calculate log ratios , moderated paired t-statistics , and corresponding p-values . We did not further adjust p-values for multiple testing . Here we primarily used p-values for ordering genes . Additional adjustments , such as a Bonferroni or Benjamini-Hochberg correction , would not affect how we ordered genes since such adjustments are typically monotonic operations . To select sets of genes whose expression was highly enriched in one of the three regions under consideration , we chose as arbitrary criteria that probesets met p < 0 . 001 and log ratio ≥ 1 in both relevant pair-wise comparisons . To rank probesets , the log ratios of the two relevant comparisons were summed in the appropriate fashion to provide a positive regional score . For example , the largest values of log2 ( BA4/caudate ) + log2 ( BA4/cerebellum ) would be candidate BA4 genes . Finally , probesets whose summed regional score was >2 in more than one region were culled from the list . The variance for a probeset , across n samples , was calculated by where xi is the RMA signal for probeset i on array n . After selecting variable genes , to minimize systematic differences of scale between the mouse and human arrays , prior to clustering we separately normalized the mouse and human RMA data to give each probeset zero mean and unit variance . Hierarchical clustering and heat maps using the 125 ( an arbitrary number ) most variable probesets were generated using Ward's linkage method , which uses an analysis of variance approach to evaluate the distances between clusters . In short , this method attempts to minimize the sum of squares of any two ( hypothetical ) clusters that can be formed at each step [25] . Heat maps were generated with the Comprehensive R Archive Network ( CRAN ) package GREGMISC . Euclidean distances between samples were calculated using RMA signals by where there are g probesets and x and y are any two mouse or human samples . Euclidian distances between regions were calculated using the mean RMA probeset signals for each tissue . To extract ortholog identities , the ENSEMBL database ( http://www . ensembl . org/Multi/martview ) was queried using mouse ENSEMBL identities provided in the Affymetrix annotation . Human ENSEMBL numbers , dN ( number of nonsynonymous substitutions/number of nonsynonymous sites ) , dS ( number of synonymous substitutions/number of synonymous sites ) , dN/dS , and percent identity were retrieved and associated with mouse probesets . dN and dS values were generated using the codeml program included in the PAML package [26 , 27] . Codeml performs pair-wise Maximum Likelihood calculations of dN and dS for each set of orthologs . We used the F3 × 4 codon evolution model . This takes into account bias derived from the different probabilities of transition versus transversion mutations and bias due to different nucleotide frequencies at the three codon positions . Incorrect ortholog assignments manifest as anomalously high dS values . We therefore applied a cut off of twice the median dS as the criterion for retaining the dN/dS ratio . Since a large but unknown fraction of genes are coregulated , assumptions of independence are not met . We therefore report extreme statistical significance ( p < 10−20 ) as p ≈ 0 , as we do not wish to imply that we believe all assumptions are correct . While additional computation might improve our estimate of p , results when assuming independence are so extreme that our conclusions per statistical significance would not change . p-Values for the intersections of lists of regional marker genes in Table 3 were calculated assuming a hypergeometric distribution drawing two lists of 30 from a pool of 8 , 500 genes . p-Values for intersection of most variable mouse and human genes in Figure 1 were calculated assuming a hypergeometric distribution drawing two lists of 125 genes from a pool of 2 , 998 genes . p-Values for correlation coefficients were calculated with a likelihood ratio test assuming observations are independent realizations from a joint bivariate normal distribution . Categories overrepresented in lists of probes were differentially expressed between different tissue regions ( e . g . , caudate versus cortex ) within species . For the human HG-U133A arrays , 70 . 6% of the probesets had an assigned GO category . For the mouse MOE430A_2 arrays , 66 . 2% of the probesets had an assigned GO category . For each GO category , the total number of probes in that category and the number of probes appearing on a list of differentially expressed probes ( p < 0 . 05 ) were calculated . A p-value for overrepresentation of each category was calculated using Fisher's exact test if either the number of probes on the list or the number not on the list was less than ten , otherwise a Pearson chi-square was used . The number of categories achieving a given p-value for overrepresentation was calculated , and its significance assessed by permutation ( to account for the overlap in categories ) . The permutation procedure was as follows: generate a list of differentially expressed probes of equal length to the actual list by sampling probes at random ( without replacement ) ; calculate the number of probes on the list for each GO category , and hence a p-value for overrepresentation; count the number of categories with a p-value for overrepresentation less than the specified criterion , and compare to that in the actual data; repeat the process 5 , 000 times . Overlaps in overrepresented categories between species for a given regional comparison were examined . These analyses were restricted to the 3 , 119 GO categories defined for both human and mouse . The number of categories significantly overrepresented ( p < 0 . 05 ) for both mouse and human in the actual data was calculated for each comparison and direction of expression . Significance was again assessed by permutation ( to reflect the fact that several probes are in more than one category ) . A random list of differentially expressed probes of equal length to that observed in human was generated and used to calculate p-values for overrepresentation for the human GO categories , as before . The n most significant categories were selected ( n being the number of significantly overrepresented categories in the actual human data ) , and the overlap between these and the significantly overrepresented mouse categories calculated . The process was repeated 10 , 000 times . For all three regional comparisons and all expression directions , the number of overlapping categories in the actual data was higher than that obtained in any of the simulated replicates .
The GEO database ( http://www . ncbi . nlm . nih . gov/geo ) accession number is GSE3790 . Affymetrix Web site ( http://www . affymetrix . com ) annotations for human HG-U133A and mouse MOE430_2 are from ( http://www . affymetrix . com/support/technical/byproduct . affx ? product=hgu133 ) and ( http://www . affymetrix . com/support/technical/byproduct . affx ? product=moe430–20 ) . | Animal models of human neurodegenerative and psychiatric disorders , particularly mouse models , have assumed a central role in biomedical research aimed at discovering the causes of disease and generating novel , mechanism-based treatments . But to what degree can a mouse brain serve as a model for a human brain ? Here we begin to address this question by looking at patterns of gene expression across three corresponding regions of mouse and human brains . We find that within each species , the different regions ( motor cortex , striatum , and cerebellum ) have very distinct gene expression profiles . It is likely that these differences reflect distinctions in regional neurochemistry and function . We then show that genes that are enriched in one of the three areas relative to the other two in mice have the same pattern of expression in humans . Looking at the relationship between conservation of expression and amino acid sequence , we find that genes showing patterned expression generally have been more conserved than more uniformly expressed genes . This suggests that in the brain , constraints on the evolution of DNA sequence and gene expression can also be particularly high for genes with regional or tissue-specific expression . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods",
"Supporting",
"Information"
] | [
"evolutionary",
"biology",
"neuroscience",
"homo",
"(human)",
"mus",
"(mouse)"
] | 2007 | Conservation of Regional Gene Expression in Mouse and Human Brain |
An important goal in researching the biology of olfaction is to link the perception of smells to the chemistry of odorants . In other words , why do some odorants smell like fruits and others like flowers ? While the so-called stimulus-percept issue was resolved in the field of color vision some time ago , the relationship between the chemistry and psycho-biology of odors remains unclear up to the present day . Although a series of investigations have demonstrated that this relationship exists , the descriptive and explicative aspects of the proposed models that are currently in use require greater sophistication . One reason for this is that the algorithms of current models do not consistently consider the possibility that multiple chemical rules can describe a single quality despite the fact that this is the case in reality , whereby two very different molecules can evoke a similar odor . Moreover , the available datasets are often large and heterogeneous , thus rendering the generation of multiple rules without any use of a computational approach overly complex . We considered these two issues in the present paper . First , we built a new database containing 1689 odorants characterized by physicochemical properties and olfactory qualities . Second , we developed a computational method based on a subgroup discovery algorithm that discriminated perceptual qualities of smells on the basis of physicochemical properties . Third , we ran a series of experiments on 74 distinct olfactory qualities and showed that the generation and validation of rules linking chemistry to odor perception was possible . Taken together , our findings provide significant new insights into the relationship between stimulus and percept in olfaction . In addition , by automatically extracting new knowledge linking chemistry of odorants and psychology of smells , our results provide a new computational framework of analysis enabling scientists in the field to test original hypotheses using descriptive or predictive modeling .
Around the turn of the century , with its acknowledgement as an object of science by the Nobel society [1] the hidden sense associated with the perception of odorant chemicals , hitherto considered superfluous to cognition , became a focus of study in its own right . Odors are emitted by food , which is a source of pleasure [2]; they also influence our relations with others [3] . The olfactory percept encoded in odorant chemicals contributes to our emotional balance and wellbeing: olfactory impairment jeopardizes this equilibrium [4 , 5] . Neuroscientific studies have revealed that odor perception is the consequence of a complex phenomenon rooted in the chemical properties of a volatile molecule ( described by multiple physicochemical descriptors ) further detected by our olfactory receptors in the nasal cavity [6] . A neural signal is then transmitted to central olfactory brain structures [7] . At this stage , a complete neural representation , called “odor” is generated and then , it can be described semantically by various types of perceptual qualities ( e . g . , musky , fruity , floral , woody etc . ) . While it is generally agreed that the physicochemical characteristics of odorants affect the olfactory percept , no simple and/or universal rule governing this Structure Odor Relationship ( SOR ) has yet been identified . Why does one odorant smell of rose and another smell of lemon ? Given the fact that the totality of the odorant message was encoded within the chemical structure , chemists have tried for a long time to identify relationships between chemical properties and odors . Topological descriptors , eventually associated with electronic properties or molecular flexibility , have been tentatively connected to odorant descriptors . For instance , molecules carrying a sulfur atom and/or having low molecular weight or low structural complexity are often rated as unpleasant [8–10] . In addition to the hedonic valence of odors , others have looked for predictive models describing odor perception and quality ( see [11–14] ) . Indeed , this was the aim of a crowd-sourced challenge recently proposed by IBM Research and Sage called DREAM Olfaction Prediction Challenge . The challenge resulted in several models that were able to predict pleasantness and intensity as well as 8 out of 19 semantic descriptors ( namely “garlic” , “fish” , “sweet” , “fruit” , “burnt” , “spices” , “flower” and “sour” ) with an average correlation of predictions across all models above 0 . 5 [15] . Although these investigations brought evidence that chemical features of odorants can be linked to odor perception , the stimulus-percept problem raised a number of issues . For instance , the stimulus-percept relationship is generally viewed as bijective in that one physicochemical rule describes or predicts one quality . However , some cases suggest the existence of more than a single rule to relate chemistry and perception . Indeed , chemicals belonging to different families can trigger a “camphor” or a musky smell [16] . On the other hand , a single chiral center can render a compound odorless or shift its perceived odor completely , as is the case for ( + ) and ( - ) -carvone [17] . These examples strengthen the notion that the connections between the chemical space and the perceptual space are subtler than previously thought with multiple physicochemical rules describing a given quality . At best , the bijective SOR rules may be only be applicable to a very small fraction of the chemical space , with the remaining part of the perceptual space being best described using a multiple rules approach . The complexity of available databases , they include both thousands of chemical properties and a large heterogeneity in perceptual descriptions , [18–21] means that the manual generation of multiple rules is not feasible . In other words , to better understand the stimulus-percept issue in olfaction , there is a clear need to extract knowledge automatically and in an intelligible manner . Such an approach is positioned upstream of predictive modeling since it will enable modeling that extracts descriptive rules from the data that link subgroups belonging to both chemical and perceptual spaces . The main aim of our study was to develop such a computational framework to discover new descriptive structure-odor relationships . To achieve this , we first set up a large database containing more than 1600 odorant molecules described by both physicochemical properties and olfactory qualities . We then developed an original methodology based on the discovery of physicochemical descriptions distinguishing between a group of objects given a target or class label , namely odor qualities . This approach has been widely studied in Artificial Intelligence ( AI ) , data mining and machine learning . Specifically , supervised descriptive rules were formalized through subgroup discovery , emerging pattern/contrast-sets mining [22] . In all cases , we face a set of objects associated with descriptions and these objects are related to one or several class labels . This new pattern mining method , a variant of redescription mining [23] , allows the discovery of pairs consisting of a description ( of physicochemical properties ) and a label ( or sub-set of labels , olfactory qualities ) . The strength of the rule ( SOR in our application ) is evaluated through a new quality-control measure detailed in the Methods section .
We designed and set up a database describing odorant molecules by both their perceptual and physicochemical properties . Here , data from different sources were extracted and grouped: ( i ) for odorant identification and olfactory qualities , we referred respectively to the PubChem website ( https://pubchem . ncbi . nlm . nih . gov/ ) and the textbook by Arctander [24]; ( ii ) for physicochemical properties , we referred to the Dragon software package ( http://www . talete . mi . it/index . htm ) . Olfactory qualities were thus gathered from the book “Perfume and Flavor Chemicals” , published in 1969 by Steffen Arctander . In this book , Arctander gives a complete description , including olfactory and trigeminal qualities as well as flavors , of 3102 odorants ( detailed physicochemical properties of 1689 odorants among these 3102 odorants were retrieved , see below ) . These odorants were further identified by chemical name , molecular weight and corresponding olfactory qualities . Here , the 74 olfactory qualities selected by Chastrette and colleagues [25] were used as a reference list . These qualities were selected in a study of the whole of Arctander’s book by excluding those that did not provide qualitative olfactory information and those that were the least frequent . Note that before selecting this source , we ran a comparison with other existing Atlases and websites used for research , teaching and applicative purposes: specifically , the Dravnieks Atlas [26] , the Boelens Atlas ( see [27] ) , and the Flavornet website ( http://www . flavornet . org ) . These sources ( atlases , book and website ) were compared along a series of parameters ( the comparison took into account all odorants for which we collected CID numbers ) . The first parameter of interest was the number of molecules studied in the source , and was respectively 1689 , 138 , 263 , and 660 for the Arctander , the Dravnieks , the Boelens and the Flavornet ( here , only molecules for which we found a PubChem Compound Identification or CID are taken into account ) . The second parameter was the number of evaluators ( and their expertise level ) who smelled the compounds and provided the olfactory qualities: one trained evaluator for the Arctander , a large panel of evaluators for the Dravnieks ( although there seems to be a large heterogeneity in the expert profile of these panelists , and little information as to the extent of training that panelists were given ) , six trained evaluators for the Boelens , and no information is given regarding the panelists for the Flavornet website . Third , when considering the way olfactory qualities were collected in the source , both the Arctander and the Flavornet used a binary format ( presence/absence of quality ) , and both the Dravnieks and the Boelens used a scale of intensity or agreement . Fourth , we compared the number of olfactory qualities used in each atlas/book/website and observed the following distribution ( the average number of qualities per molecule is in brackets ) : 74 ( 2 . 88 ) for the Arctander , 146 ( 29 . 99 ) for the Dravnieks , 30 ( 12 . 86 ) for the Boelens , and 197 ( 2 . 72 ) for the Flavornet . Note also that the minimum ( and the maximum ) number of qualities for one molecule was: Arctander ( min: 1; max: 10 ) , Dravnieks ( min: 5; max: 52 ) , Boelens ( min: 0; max: 22 ) , Flavornet ( min: 1; max: 5 ) . Thus , this analysis showed that whereas some sources are characterized by a large number of molecules ( e . g . Arctander and Flavornet ) , others contain only a limited number of odorants ( e . g . Boelens and Dravnieks ) . Moreover , there is great heterogeneity between these different sources with regards to the number and the degree of expertise of the evaluators . Some sources involve a large number of evaluators but with heterogeneous profiles ( e . g . Dravnieks ) and others involve a limited number of experts ( e . g . Boelens and Arctander ) . Finally , whereas some sources have , on average , between 10 and 30 qualities per odorant ( e . g . Boelens and Dravnieks ) , the average number is around three for others ( e . g . Arctander and Flavornet ) . In view of these parameters , and because the descriptive approach used in this study requires a large database , we used the Arctander book because it contained the highest number of odorant molecules ( 1689 ) and a reasonable number of qualities per odorant ( 2 . 88 on average ) . Odorant physicochemical properties were then obtained using Dragon , a software application that enables the calculation of 4885 molecular descriptors ( Talete ) . Descriptors included in our dataset ranged from the simplest atom types , functional groups and fragment counts , to topological and geometrical descriptors . As Dragon requires 3D structure files , these were collected from the PubChem website ( https://pubchem . ncbi . nlm . nih . gov ) by using the compound identifier number of each odorant ( CID ) . Individual odorant CIDs were obtained by using the CAS Registry Number and/or the chemical name of the odorant as an entry in the PubChem website . In total , 1689 CIDs were found for the 3102 odorants . In the following section , we study the set M of odorant molecules that are described by n physicochemical properties denoted F . Each property fi ∈ F is a function that associates a real value with a molecule: fi: M → image ( fi ) with image ( fi ) an interval of R . The olfactory qualities are denoted by O and class is a mapping that associates a subset of O to a molecule: class: M → 2O . Here , we developed an original subgroup discovery approach to mine descriptive rules that specifically characterize subsets of olfactory qualities ( O ) . The specificity of this approach is intended to be able to extract rules with several olfactory qualities as targets , and also to treat unbalanced classes robustly , i . e . , the fact that some olfactory qualities are very rare ( e . g . “musty” ) compared to others ( e . g . “fruity” ) . Subgroup discovery is a generic data mining method aimed at discovering regions in the data that stand out with respect to a given target . We instantiated this framework in order to identify the conditions on some odorant physicochemical properties that are strongly associated with olfactory qualities . A structure odor rule ( SORule ) , denoted D → Q , is defined by a physico- chemical description D and a set of olfactory qualities Q ⊆ O . The description is a set of n intervals D = ⟨[x1 , y1] , [x2 , y2] , … , [xn , yn]⟩ , each being a restriction on the value image of its corresponding physicochemical property: [xi , yi] ⊆ image ( fi ) . The molecules whose values on physicochemical descriptors belong to the intervals of the description D are members of the coverage of D: coverage ( D ) ={m∈M∀i=1…n , xi≤fi ( m ) ≤yi} We count the number of molecules in the coverage with support ( D ) = |coverage ( D ) | . The quality of a rule is evaluated with respect to the olfactory qualities of the molecules in its coverage . First , the precision measure gives the proportion of the molecules of the coverage of D that also have ( part of ) the olfactory qualities Q: P ( D→Q ) =|{m∈coverage ( D ) class ( m ) ⊆Q}|support ( D ) This is the percentage of times the rule is triggered for molecules whose qualities are in Q . On the other hand , it is also important to know if the rule covers all the molecules of quality Q . This is what the recall measure evaluates: R ( D→Q ) =|{m∈coverage ( D ) class ( m ) ⊆Q}||{m∈Mclass ( m ) ⊆Q}| These two measures behave in opposite ways: when one increases , the other decreases . One way to globally evaluate a rule is to use the F1 measure , the harmonic mean between the precision and recall measures: F1 ( D→Q ) =2P ( D→Q ) R ( D→Q ) P ( D→Q ) +R ( D→Q ) As mentioned above , the olfactory qualities are more or less frequent in the data . To take that into account , the Fβ measure gives more importance to the precision measure for rare olfactory qualities , while favoring the recall measure for frequent qualities: Fβ ( D→Q ) = ( 1+β ( support ( Q ) ) P ( D→Q ) R ( D→Q ) β ( support ( Q ) ) P ( D→Q ) +R ( D→Q ) with support ( Q ) = |{m ∈ M |class ( m ) ⊆ Q}| and β ( x ) = ( 0 . 5× ( 1+tanh ( xβ-xlβ ) ) ) 2 Here , the terms xBeta and lBeta are determinant in choosing the appropriate sigmoid model , and are values that can be set by the experimenter . Given that , our approach aims to discover rules D → Q whose support support ( D ) is greater than a threshold minSupp and with |Q| is lower or equal to a value maxQual . Those parameters make it possible to identify rules that are supported by sufficient odorant molecules , and also that are specific to a small set of olfactory qualities . The maxQual parameter enforces that the right-hand side of the rule contains a limited number of olfactory qualities to be interpretable by the analyst . Similarly , a maxProp parameter allows to limit the number of ( physicochemical ) conditions in the left-hand side of the rules . To illustrate the previous definitions , let us consider the toy olfactory dataset given in Table 1 . This dataset contains 6 molecules identified by their IDs M = {1 , 2 , 3 , 4 , 5 , 6} . Each molecule is described by its molecular weight MW , its number of atoms nAt and its number of carbon atoms nC , that is , F = {MW , nAt , nC} . Besides , the molecules are also associated with their olfactory qualities among O = {fruity , vanillin , woody} . Let us consider the description D=⟨[128 , 151] , [23 , 29] , [[9 , 12]⟩ Its coverage is coverage ( D ) = {2 , 3 , 5 , 6} . If we consider the odorant quality Q = {vanillin} , as there is 2 molecules of coverage ( D ) with this quality , the precision of the rule is equal to: P ( D→Q ) =24 As there are 3 molecules in the whole dataset with that quality , the recall of the rule is: R ( D→Q ) =23 Its F1 measure is thus equal to: F1 ( D→Q ) =227 Detailed information regarding the principle of the algorithm are provided as S1 Text .
Our olfactory dataset includes 1689 molecules described by 74 olfactory qualities . The dataset is multi-labeled , each molecule being associated with one or several olfactory qualities . On average , each molecule refers to 2 . 88 olfactory qualities among the 74 possible labels . Moreover , the frequency of olfactory qualities across odorants is unbalanced: on average a quality is used in 65 . 79 molecules ( standard deviation: 105 . 28 ) , the maximum is reached for the “fruity” quality ( used in 570 molecules ) , the minimum for musty ( used in only 2 molecules ) . Fig 1 illustrates the entire building process of the database . Fig 2 presents a world cloud of the 74 olfactory qualities . With regard to the physicochemical properties , our original database contained more than 4000 physicochemical features . For the purpose of a rational approach where features can be interpreted on a chemical basis , we selected attributes that were relevant , but more importantly easily interpretable . This approach is strongly inspired by the so-called 3D-olfactophore , where such easily interpretable features computed on odorants sharing the same olfactory percept are gathered in the 3 dimensions of space . Such features are typically Hydrogen bond donor/acceptor , Aromatic cycle , Charged atom , etc . This methodology is typically useful for molecular scientists to learn about structure-property relationships and design new molecules which fulfill the properties of these olfactophores [28] . Here the features we used were a series of physico-chemical properties . Thus , we selected constitutional , topological and chemical descriptors that represent molecular features which can be easily interpreted and extrapolated for further predictive models . They include the following categories: constitutional indices ( n = 29; ex . “Molecular weight” ) , ring descriptors ( n = 7; ex . “Number of rings” ) , functional group counts ( n = 40; ex . “Number of esters” ) , molecular properties ( n = 6; ex . “Topological polar surface area” ) . To select these descriptors , we screened the whole set of descriptors proposed by Dragon . We carefully selected descriptors able to provide information interpretable by any molecular scientist . The cost of selecting interpretable descriptors is a reduction in the description of the dataset . To evaluate the loss of information on the variance of a given molecular dataset , descriptors were computed on a set of 2620 odorants provided by Saito and colleagues [29] . Finally , 347 descriptors remained after filtering the following: correlated ( above 0 . 85 ) , constant for the whole dataset ( no variation across parameters ) , not available for the whole dataset . After the dimensionality reduction , our selected 82 descriptors accounted for 37 . 2% of the original variance . When choosing randomly 82 descriptors within this set of 347 , the variance always falls below 25% , suggesting that our descriptors performed quite well at describing a molecular set with a certain degree of variability . Finally , when projecting the entire set of molecules on to the two first components of a PCA , the dataset remains well split and molecules were still distinguishable . First , the physicochemical rules were generated for each of the 74 qualities based on the 82 descriptors . This was done using the following parameters: maxoutput ( 100 ) , beamwidth ( 30 ) , MaxQual ( 1 ) , MaxProperties ( 8 ) , max Supp ( 700 ) , XBeta ( 110 ) , IBeta ( 20 ) , and four different minSupp ( 5 , 10 , 20 and 30 ) ( see Methods section and S1 Text for a detailed definition of these parameters ) . Second , an algorithm search for the best rules or combination of rules ( with a maximum of 12 rules ) for each of the 74 qualities and the four different minSupp ( from 5 to 30 ) . At this stage , the rules or combination of rules were ranked as a function of their Precision . Here , to evaluate the best rule or combination of rules that can describe each quality , we calculated for each rule ( or combination of rules ) the distance ( Euclidian ) from the “ideal” situation defined as the data-point with an error of “0” ( error was calculated as one minus precision ) and the best recall ( value of 1 in the y-axis , meaning that all molecules that belong to the quality are described by these physicochemical rules ) . The point ( s ) with the smallest distance was ( were ) selected as the best rule or combination of rules for a given quality . From this selection , we built a list of rules and/or combination of rules for each quality ( see S1 Table ) . We showed that around 90% of the olfactory qualities were described by 1 to 6 rules and 66% ( 49 qualities among 74 ) were described by 3 , 4 or 5 rules ( see Fig 3a ) . Moreover , for the same quality , different rules or combinations of rules were selected because their distance to the “ideal” situation ( recall: 1; error: 0 ) was the same ( see an example in Fig 3b ) . Fig 3c shows an example of the chemical structure of the molecules described by the same quality ( jasmine here ) and rules/combinations of rules . To compare olfactory qualities according to their description by physicochemical rules , we plotted all physicochemical rules ( and/or combination of rules ) of each quality in a 2D space comprising error ( x-axis ) and recall ( y-axis ) ( Fig 4 ) . As can be seen , whereas some qualities were close to the “ideal” situation others were very far . First , 38 qualities ( 51 . 35% , named “Group 1” ) exhibited an error rate lower than 0 . 5 and a recall greater than ( or equal to ) 0 . 5 ( sulfuraceous , vanillin , phenolic , musk , sandalwood , almond , orange-blossom , jasmine , hay , tarry , smoky , lilac , piney , camphor , grape , anisic , buttery , gassy , fatty , waxy , acid , minty , aromatic , mossy , violet , citrus , peppery , caramelic , medicinal , tobacco , pear , lily , sour , orange , animal , honey , hyacinth , rose ) . Second , 17 qualities ( 22 . 97% , named “Group 2” ) exhibited an error rate lower than 0 . 5 but a recall lower than 0 . 5 ( amber , geranium , metallic , fruity , pineapple , ethereal , plum , woody , balsamic , creamy , green , berry , oily , spicy , floral , winey , herbaceous ) . Third , 18 qualities ( 24 . 32% , named “Group 3” ) showed an error rate greater than ( or equal to ) 0 . 5 and a recall greater than ( or equal to ) 0 . 5 ( leathery , aldehydic , mushroom , coco , mimosa , tea , nut , root , peachy , earthy , powdery , orris , apple , leafy , apricot , musty , brandy , narcissus ) . Fourth , one quality ( 1 . 35% , named “Group 4” ) showed an error rate greater than ( or equal to ) 0 . 5 and a recall lower than 0 . 5 ( banana ) . To further examine whether the generated physicochemical rules were specific to a given perceptual quality , in other words whether they provided a good and relevant model , we used Bootstrap confidence intervals to evaluate whether the generated F-measure of the rules/models was significative . Here , knowing that a given set of rules covers X molecules , we sampled 100 , 000 sets of X molecules ( with replacement ) and calculated the F-measure of each sample according to the studied quality . Next , the confidence intervals ( CI: 99% ) of these sets were computed . Afterwards , the F-measure of the set of discovered rules was compared to this CI . Results showed that for all 74 qualities , the F-measure was significant in that its value was outside ( and greater ) the CI at 99% . Finally , to examine how the model built with 82 physicochemical descriptors performed compared to a model built with all 4000 descriptors , we calculated the F-measure for each quality ( computed on the basis of all sets of rules ) in both types of models . Results showed that , on average , the F-measure was significantly greater ( p<0 . 0001 ) in the model with 82 physicochemical descriptors ( mean = 0 . 592 , SEM = 0 . 012 ) compared to the model with all 4000 descriptors ( mean = 0 . 487 , SEM = 0 . 011 ) , reflecting that the use of a small but explicative and intelligible set of descriptors enhances performance . To sum up , we provide here a computational framework that enables the automatic extraction , from a complex and heterogeneous dataset , descriptive rules linking subgroups in a chemical space onto subgroups in a perceptual space . As can be seen in Fig 3a , only 3 qualities could be best described by a single physicochemical rule whereas more than two thirds of the qualities needed between 3 and 5 rules to be described . When dealing with the confidence of the rules , a gradient was observed whereby some rules were associated with a good rate of recall and minimum rate of error , whereas other rules exhibited a lower confidence in describing olfactory qualities . Note that all the generated rules are available to the reader in S1 Table . The computational approach that we developed is available at the following address: https://projet . liris . cnrs . fr/olfamine/ Here , we analyzed some of the best-known qualities in the field of olfactory evaluation , namely "fruity" , "floral" , "woody" , "camphor" , "earthy" , "spicy" , "fatty" . The analysis of the rules and combinations of rules ( see S1 Table ) , shows that the number of rules is quite high for these qualities ranging from six ( floral ) , seven ( camphor , earthy ) , eight ( spicy , woody ) , nine ( fatty ) to twelve ( fruity ) . From a physicochemical point of view , translated into interpretable rules , the floral quality is characterized by either aromatic and strongly hydrophobic molecules or non-aromatic and moderately hydrophobic odorants . For camphor , molecules are rather small in size , moderately hydrophobic , and eventually cyclic . The earthy quality is characterized by moderately hydrophobic molecules with unsaturations . The spicy quality is characterized by rather rigid molecules , eventually aromatic . Woody quality includes hydrophobic molecules , rather not cyclic nor aromatic . For the fatty , the molecules have a larger carbon-chain skeleton which is highly hydropobic with aldehyde or acid functions . Finally , for the fruity quality , molecules are described as having moderate hydrophobicity and being medium to large in size . To push the interpretation further , we examined qualities associated with generated physicochemical rules with the highest level of confidence . Here , we attempted ( i ) to understand the rules based on a priori knowledge and ( ii ) to examine whether the rules could raise new scientific assumptions . We analyzed a total of eleven qualities corresponding to the first quartile of the distribution of all rules . Based on the Euclidian distance to the “ideal” situation; 473 rules were generated by our analysis ( see Fig 4 ) . These qualities were: sulfuraceous , vanillin , phenolic , musk , sandalwood , almond , orange-blossom , jasmine , hay , tarry , smoky . The “sulfuraceous” quality was described as follows: R1: [0 . 0<nCsp2<0 . 0] [0 . 0<nHAcc<0 . 0] [11 . 611<Se<22 . 069] [144 . 039<SAtot<222 . 269] [0 . 0<TPSA ( Tot ) <50 . 6]; R2: [1 . 0<nS<2 . 0] [1 . 0<nC<6 . 0] [0 . 0<N%<0 . 0] [25 . 0<C%<33 . 3] [38 . 8<TPSA ( Tot ) <64 . 18]; R3: [1 . 0<nS<2 . 0] [-0 . 264<Hy<0 . 323] [102 . 715<SAtot<222 . 269] [0 . 0<O%<6 . 3] . These descriptions suggest , somewhat intuitively , that sulfuraceous odorants encompass molecules with one or two sulfur atoms and are moderately heavy , with a maximum of six carbon atoms . Four rules defined the “phenolic” quality: R1: [216 . 155<SAtot<218 . 661] [0 . 0<nCrs<0 . 0] [0 . 0<nOHp<0 . 0] [30 . 4<C%<45 . 0] [0 . 0<Ui<2 . 322]; R2: [1 . 117<Mi<1 . 118] [-0 . 768<Hy<-0 . 158] [0 . 0<nR = Ct<0 . 0] [43 . 5<H%<50 . 0] [0 . 0<nOxiranes<0 . 0] [0 . 0<nR = Cp<0 . 0]; R3: [2 . 807<Uc<2 . 807] [3 . 0<nCp<5 . 0] [0 . 4<ARR<0 . 545] [2 . 0<Ui<2 . 0] [-0 . 888<Hy<-0 . 277] [37 . 8<C%<40 . 0] [0 . 0<nOHt<0 . 0] [0 . 0<nOHp<0 . 0]; R4: [0 . 6<ARR<0 . 75] [1 . 0<nArOH<2 . 0] [2 . 807<Uc<3 . 17] [170 . 356<SAtot<222 . 475] [0 . 893<MLOGP<2 . 778] [0 . 0<nArCO<0 . 0] . Thus , odorants having a “phenolic” quality are of moderate size , with few unsaturations and low hydrophilicity ( and high lipophilicity ) . It can be regarded as a cyclic molecule . A good consistency is observed between the 4 rules . For “vanillin” , the following rules were observed: R1: [0 . 5<ARR<0 . 545] [3 . 0<nCb-<4 . 0] [3 . 0<nHAcc<3 . 0] [1 . 0<nArOR<2 . 0] [0 . 0<nR = Cp<0 . 0] [0 . 0<nArCO<0 . 0] [38 . 1<C%<46 . 2]; R2: [3 . 0<nCb-<3 . 0] [3 . 0<nO<3 . 0] [0 . 0<nArCOOR<0 . 0] [-0 . 727<Hy<0 . 66] [42 . 1<H%<50 . 0] [0 . 0<nArCO<0 . 0] [38 . 1<C%<42 . 3]; R3: [2 . 0<nCsp3<2 . 0] [1 . 0<nArOR<2 . 0] [0 . 699<MLOGP<1 . 75] [0 . 0<nArCOOR<0 . 0] [0 . 0<nArCO<0 . 0] [2 . 0<nCb-<4 . 0] . These descriptions suggest that odorants belonging to this group are mostly cyclic molecule ( like the prototypical molecule vanillin ) , with 3 Hydrogen bond acceptors branched on saturated carbons atoms on an aromatic cycle . When considering the “musk” quality , the following rules emerged: R1: [3 . 72<MLOGP<4 . 045] [2 . 0<nCrs<15 . 0] [1 . 0<nCIC<1 . 0] [333 . 936<SAtot<436 . 545]; R2: [4 . 0<nCb-<6 . 0] [33 . 0<nBT<47 . 0] [0 . 0<nCbH<2 . 0]; R3: [0 . 0<RBN<0 . 0] [11 . 0<nCs<16 . 0]; R4: [238 . 46<MW<270 . 41] [57 . 1<H%<63 . 8] [402 . 5<SAtot<440 . 301] [0 . 0<nR07<0 . 0] [-0 . 931<Hy<-0 . 763] [0 . 0<ARR<0 . 316] [0 . 0<RBN<12 . 0] [0 . 0<nCt<3 . 0] . Musky molecules are heavy and hydrophobic compounds . This is reflected by a rather large logP , surface area or molecular weight . From a general point of view , these descriptors reflect well the features of musky odorants . For the “sandalwood” quality , two rules were observed: R1: [3 . 0<nCrt<5 . 0] [1 . 0<nHDon<1 . 0] [0 . 0<nR04<0 . 0] [1 . 0<nCrq<2 . 0]; R2: [3 . 0<nCrt<5 . 0] [1 . 0<nHDon<1 . 0] [-0 . 429<Hy<-0 . 325] [2 . 0<nR05<3 . 0] . Sandalwood odorants are quite diverse and minor modifications within their structure can abolish the sandalwood note . The rules which are mined here correspond to models which are very simple and hardly capture the subtlety of this odorant family [28] . The description presented here corresponds to the prototypic beta-santalol structure which has a campholenic skeleton . The “almond” quality was described by four rules: R1: [0 . 0<nCp<0 . 0] [152 . 443<SAtot<165 . 41] [1 . 0<nO<2 . 0] [2 . 0<Ui<2 . 585]; R2: [0 . 706<ARR<0 . 8] [0 . 0<nArCO<0 . 0] [1 . 0<nO<1 . 0] [3 . 0<Uc<3 . 807] [0 . 143<MLOGP<3 . 571] [-0 . 917<Hy<-0 . 71] [0 . 0<nCb-<2 . 0]; R3: [1 . 0<nH<5 . 0] [0 . 0<nOxiranes<0 . 0] [1 . 0<nHAcc<3 . 0] [1 . 0<nN<2 . 0] [23 . 79<TPSA ( Tot ) <90 . 27] [0 . 0<O%<14 . 3] [0 . 0<ARR<0 . 75]; R4: [1 . 0<nArCHO<1 . 0] [11 . 0<nBT<20 . 0] [45 . 0<C%<47 . 1] [-0 . 864<Hy<-0 . 668] [1 . 0<nHAcc<2 . 0] . These descriptions suggest that odorants evoking an almond-like quality are compounds bearing at least one oxygen and/or other hydrogen bond-accepting atom but also bearing an aromatic cycle . This means that the structure bears several unsaturations . These chemicals are thus relatively small and can be compared to the prototypical structure of benzaldehyde . Four physicochemical rules described the “orange-blossom” quality: R1: [10 . 0<nCsp2<10 . 0] [9 . 23<TPSA ( Tot ) <58 . 89]; R2: [1 . 0<nArNH2<1 . 0] [213 . 361<SAtot<326 . 286] [0 . 0<nR = Cs<0 . 0] [0 . 0<nCt<0 . 0] [37 . 9<C%<51 . 5]; R3: [0 . 773<ARR<0 . 857] [39 . 4<H%<45 . 5] [9 . 23<TPSA ( Tot ) <52 . 32] [3 . 0<nCb-<5 . 0]; R4: [47 . 243<Se<53 . 454] [4 . 0<nCbH<9 . 0] [3 . 287<MLOGP<5 . 007] [3 . 0<nHAcc<4 . 0] [0 . 231<ARR<0 . 462] . These descriptions characterize very diverse structures ranging from very small to medium or large compounds . As a general rule , one can note the presence of unsaturations , consistent with a terpenic structure , associated with a quite hydrophobic feature . The “jasmine” quality was described by six rules: R1: [12 . 0<nC<13 . 0] [43 . 37<TPSA ( Tot ) <44 . 76]; R2: [336 . 137<SAtot<337 . 327] [0 . 0<nR = Cs<0 . 0]; R3: [7 . 0<nCsp2<8 . 0] [1 . 0<nCb-<1 . 0] [2 . 0<nCp<3 . 0] [50 . 0<H%<53 . 3] [4 . 0<nCsp3<5 . 0] [1 . 0<nCs<3 . 0] [1 . 0<nRCOOR<1 . 0]; R4: [1 . 0<nCb-<1 . 0] [2 . 034<MLOGP<2 . 386] [2 . 0<nHet<3 . 0] [7 . 0<nCsp2<8 . 0] [1 . 0<nCp<2 . 0] [-0 . 807<Hy<-0 . 727] [0 . 0<nArCOOR<0 . 0] [0 . 0<nArOR<0 . 0]; R5: [5 . 0<RBN<6 . 0] [1 . 0<nRCO<1 . 0] [291 . 434<SAtot<350 . 346] [10 . 0<nC<13 . 0] [0 . 0<nArCO<0 . 0]; R6: [1 . 0<nR = Ct<1 . 0] [4 . 0<nCs<8 . 0] [2 . 0<nCconj<4 . 0] [0 . 0<nCt<0 . 0] [-0 . 912<Hy<-0 . 873] . This rule characterizes ( i ) molecules composed mainly of carbons and oxygen atoms , ( ii ) molecules with an aromatic core and embranchments conferring a large flexibility , and ( iii ) compounds with an optimal chain length around five carbon atoms . These rules are in line with the prototypical molecule jasmonate . For the “hay” quality , six rules were generated: R1: [1 . 0<nArCO<1 . 0] [-0 . 164<Hy<0 . 647] [1 . 239<MLOGP<2 . 001]; R2: [13 . 3<O%<13 . 6] [2 . 322<Ui<2 . 322] [1 . 191<MLOGP<1 . 75] [0 . 0<nR = Cp<0 . 0] [179 . 198<SAtot<300 . 766]; R3: [0 . 556<ARR<0 . 6] [177 . 465<SAtot<205 . 275] [26 . 3<TPSA ( Tot ) <50 . 44] [-0 . 603<MLOGP<2 . 001] [11 . 1<O%<20 . 0] [0 . 0<nCs<0 . 0] [0 . 0<nPyridines<0 . 0] [1 . 0<RBN<2 . 0]; R4: [2 . 0<nCb-<2 . 0] [179 . 249<SAtot<209 . 869] [2 . 322<Ui<2 . 585] [26 . 3<TPSA ( Tot ) <37 . 3]; R5: [1 . 111<Mi<1 . 116] [2 . 0<nHAcc<2 . 0] [2 . 0<nCb-<4 . 0] [0 . 0<nR05<0 . 0] [0 . 0<nCconj<1 . 0] [1 . 49<MLOGP<3 . 719]; R6: [0 . 0<RBN<0 . 0] [2 . 0<nO<2 . 0] [0 . 0<nR = Ct<1 . 0] [-0 . 807<Hy<-0 . 668] [26 . 3<TPSA ( Tot ) <30 . 21] [130 . 383<SAtot<214 . 985] [-0 . 145<MLOGP<2 . 265] . These rules characterize relatively hydrophobic molecules composed of aromatic cycles , being either heterocyclic or linked to a heteroatom outside of the cycle . These atoms confer to the molecule the possibility to accept Hydrogen bonds . “Tarry” quality was also described by six rules: R1: [6 . 0<nCsp2<6 . 0] [-0 . 213<Hy<0 . 031] [1 . 348<MLOGP<1 . 859] [0 . 0<RBN<1 . 0]; R2: [3 . 0<nCb-<4 . 0] [2 . 807<Uc<2 . 807] [1 . 0<nHet<2 . 0] [18 . 46<TPSA ( Tot ) <40 . 46] [138 . 18<MW<178 . 3] [38 . 7<C%<40 . 0] [0 . 0<nCs<0 . 0]; R3: [0 . 733<ARR<0 . 773] [-0 . 905<Hy<-0 . 158] [1 . 0<nHet<1 . 0] [174 . 318<SAtot<277 . 868]; R4: [0 . 0<nDB<0 . 0] [174 . 318<SAtot<175 . 125] [2 . 807<Uc<3 . 585] [15 . 862<Se<17 . 534]; R5: [5 . 9<N%<7 . 7] [0 . 6<ARR<1 . 0] [0 . 565<MLOGP<1 . 834] [-0 . 828<Hy<0 . 031] [12 . 753<Se<16 . 636]; R6: [-0 . 213<Hy<-0 . 158] [16 . 0<nBT<21 . 0] [0 . 0<nRCOOH<0 . 0] [0 . 0<nROH<0 . 0] [1 . 58<MLOGP<2 . 193] [15 . 79<TPSA ( Tot ) <29 . 46] . With regard to this quality , it is not easy to establish specific characteristics of the molecules of this group , but overall these molecules are flexible , presenting heteroatoms while having low hydrophilicity due to the presence of double bonds . Finally , the “smoky” quality is described as follows: R1: [1 . 0<nArOH<1 . 0] [1 . 859<MLOGP<2 . 193] [1 . 117<Mi<1 . 121] [0 . 0<nCconj<0 . 0]; R2: [7 . 0<nC<7 . 0] [2 . 807<Uc<2 . 807] [41 . 2<C%<43 . 8] [-0 . 158<Hy<-0 . 107]; R3: [0 . 0<RBN<0 . 0] [4 . 0<nCar<5 . 0] [99 . 023<SAtot<129 . 741] . In this case , a robust rule is hard to establish because the physicochemical descriptors refer either to aromatic compounds with a hydroxyl group or flexible molecules with rotatable bonds . To evaluate the validity of the generated physicochemical rules , we applied them to novel sets of odorants . For a given quality , we checked whether novel odorants that fulfill physicochemical criteria according to our descriptive model indeed evoked significantly more of the studied quality than novel odorants than do not fulfill these physicochemical rules . To this end , we isolated from 4 different databases , 4 sets of odorants not present in the Arctander database and therefore not used to build the descriptive rules . These databases were from the Dravnieks study [26] ( n = 45; i . e . 45 odorants not present in our original dataset could be used ) , the Boelens & Harding study [30] ( n = 56 ) , one set from the Keller et al . study [15] ( n = 118 ) , and one set from the Licon et al . study [31] ( n = 19 ) . Within each of these four novel sets , olfactory quality was coded using a continuous variable ( Dravnieks: from 0 to 100; Boelens & Haring: from 0 to 9; Keller et al . : from 0 to 100; Licon et al . : from 0 to 100 ) . Note that , for the Keller et al . study , perceptual data were provided for 2 levels of odorant concentrations ( « High » and « Low » ) . Our descriptive model was tested in qualities that were common between the Arctander database and these four different databases . Moreover , for statistical purposes and for a given quality , only when the rules were filled for at least five odorants , comparisons were performed between odorants that filled the criteria for the rules and those that did not filled the rules . The qualities that satisfy these criteria were: 1/ for the Dravnieks study: Woody ( n = 5 ) , Camphor ( n = 5 ) , Earthy ( n = 5 ) , 2/ for the Boelens & Haring study: Woody ( n = 10 ) , Fruity ( n = 9 ) , Green ( n = 8 ) and Balsamic ( n = 5 ) , 3/ for the Keller et al . study: Fruity ( n = 15 ) , and Sulfuraceous ( n = 16; which was compared to a semantically proximal perceptual quality present in the Keller database , namely « Decayed » ) , and 4/ for the Licon et al . study: Camphor ( n = 5 ) . Results are presented in Fig 5 . Within each set , an analysis of variance ( ANOVA ) comparing perceptual values for a given quality for odorants that fulfill the physicochemical rules ( Rule ( 1 ) , black bars ) vs . those that did not fulfill the rules ( Rule ( 0 ) , grey bars ) was performed . For the Dravnieks dataset , the statistical analysis revealed that odorants that fulfill the rules for woody , earthy and camphor , were respectively perceived as significantly more woody ( F ( 1 , 43 ) = 14 . 19 , p<0 . 001 , η2 = 0 . 248; Fig 5a . i ) , earthy ( F ( 1 , 43 ) = 6 . 128 , p = 0 . 017 , η2 = 0 . 125; Fig 5a . ii ) and camphoreous ( F ( 1 , 43 ) = 28 . 63 , p<0 . 001 , η2 = 0 . 400; Fig 5a . iii ) . In the same line , a significant increase in camphor quality was observed for odorants that fulfill the rules for this quality in the Licon et al . dataset ( F ( 1 , 17 ) = 6 . 804 , p = 0 . 018 , η2 = 0 . 286; Fig 5b ) . Validation was also observed within the Boelens & Haring dataset , but the results were more mixed . Whereas a significant increase was observed for woody ( F ( 1 , 54 ) = 88 . 47 , p<0 . 001 , η2 = 0 . 621; Fig 5c . i ) and balsamic ( F ( 1 , 54 ) = 15 . 86 , p<0 . 001 , η2 = 0 . 227; Fig 5c . iv ) in odorants that fulfill the physicochemical rules for these respective qualities , this was not the case for the green quality ( F ( 1 , 54 ) = 0 . 227 , p = 0 . 636 , η2 = 0 . 004; Fig 5c . ii ) . On a descriptive level , Fig 5c . iii shows that odorants that fulfill the physicochemical criteria for the quality fruity seem to be perceived as more fruity , but this was not significant ( F ( 1 , 54 ) = 1 . 989 , p = 0 . 164 , η2 = 0 . 036 ) . However , when considering the Keller et al . dataset , validation was reached for fruity: odorants that fulfill criteria for the fruity quality were perceived as more fruity ( for both low ( F ( 1 , 116 ) = 9 . 219 , p = 0 . 003 , η2 = 0 . 074; Fig 5d . i ) and with high levels of concentrations ( F ( 1 , 116 ) = 11 . 76 , p<0 . 001 , η2 = 0 . 092; Fig 5d . ii ) ) , than odorants that did not fulfill the rules . The statistical analysis of this dataset shows also that odorants that fulfill the physicochemical criteria for the quality sulfuraceous were perceived as more decayed at both low ( F ( 1 , 116 ) = 10 . 49 , p = 0 . 002 , η2 = 0 . 083; Fig 5d . iii ) and high levels of concentrations ( F ( 1 , 116 ) = 24 . 42 , p<0 . 001 , η2 = 0 . 174; Fig 5d . iv ) . To sum up , the present validation involved four sets of stimuli for a total of 238 odorants . It allowed us to test the descriptive model on seven perceptual qualities and for six of them ( woody , earthy , camphor , balsamic , sulfuraceous , fruity ) , the rules generated by our model have been consistent with the ratings provided in these independent datasets .
The interaction between the odorant molecule and the olfactory receptor ( s ) induces a percept called “odor” . Chemists have previously attempted to characterize this phenomenon by working to obtain descriptive and/or predictive rules connecting physicochemical properties to odors [32] . Such is the case with olfactophores or the exploitation of more specific molecular features for predicting intensity or pleasantness [8 , 11 , 33 , 34] . Recently , a large database of compounds as well as a large number of human panelists were used in order to predict percepts , intensity and pleasantness [15] . In our study , we also considered that , to a certain extent , the odor quality of a molecule is encoded in its chemical structure . Our aim was to provide a descriptive model of the relationship between molecules and their perceived odors . To achieve this aim , we set up a new computational framework that considers the scientific assumption that , rather than relying on single physicochemical descriptions , the relationship between the chemical space of odorants and the perceptual space of odors should be examined through multiple descriptions . We developed a new method based on a subgroup discovery algorithm to mine descriptive ( physicochemical ) rules characterizing specific subsets of class labels ( olfactory qualities ) . Thanks to this data-mining approach , we were able to provide new descriptive structure-odor rules with a gradient of confidence ( taking into account both the recall and the precision ) that varied from one quality to another . Validation of these descriptive models was achieved for a series of olfactory qualities associated with rules with medium levels of confidence ( woody , earthy , balsamic , fruity ) to higher levels of confidence ( sulfuraceous and to a less degree camphor ) . Our findings contribute to a better understanding of the olfactory system by elucidating the relationships between the chemistry and the psychobiology of smells . Indeed , the function of the olfactory system is to detect and discriminate volatile environmental molecules in order to make sense of them . This implies the construction of dedicated percepts that can influence behavior . In order to understand this system , relating the worlds of chemistry and perception is a requirement . Our findings provide descriptive elements of responses and highlight the physicochemical rules that describe olfactory perceptual qualities . Beyond these aspects , our algorithm would benefit from a more systemic approach through the inclusion of neurobiological representational states , ranging from olfactory receptors and olfactory bulb to primary and secondary olfactory areas . This will allow us to better understand how the interaction between the chemical features of odorants and olfactory receptors is mediated and processed in the brain to build olfactory percepts . One question that may be raised from the current finding is how our descriptive approach is different from other machine learning methods and how it may help chemists and neuroscientists interested in olfaction solve scientific issues ? In contrast to classical predictive machine learning tasks where the goal is to turn the data into an as-accurate-as-possible prediction machine , exploratory data analysis such as ours aims to automatically discover new insights about the domain in which the data was measured ( e . g . , olfaction ) . To this end , the notion of interpretability is fundamental as it is the premise of descriptive rules . Indeed , these rules are composed of conjunctions of conditions on attributes that conclude on some olfactory qualities . In contrast to black-box models , these rules , assessed by intuitive and mathematically well-funded measures are easy to assimilate for a domain expert . This , in turn , makes development of new hypotheses possible . In sum , our data-mining method should be regarded as an approach that can extract knowledge from a dataset characterized by its complexity , size and heterogeneity . Our approach is therefore situated at the upstream of any hypothetical-deductive approaches . The generation of descriptive rules allows researchers to start such a hypothetical-deductive approach , and to formulate new scientific assumptions , to establish an experimental methodology and finally to develop and test the validity of predictive models . Our algorithm has made it possible to extract significant knowledge about a series of olfactory qualities . First , qualities with a chemical terminology ( sulfuraceous , vanillin , phenolic ) have a great reliability in the rules generated . These rules contained expected attributes such as the presence of sulfur atoms to describe “sulfuraceous odors” , suggesting that our algorithm was efficient in extracting relevant and meaningful knowledge . Our results went beyond the sole description of these expected physico-chemical attributes . The generated rules contained also unexpected features such as “phenolic odors” , where the presence of moderate size molecules , with few unsaturations and low hydrophilicity were put forward . Structure-odor relationships for some qualities such as musky [35 , 36] , sandalwood [37–39] and to a lesser degree almond and jasmine [40] have already been explored in the past . Our descriptive model could bring new information for most of these qualities , thus enabling the testing of innovative hypothesis in the field . Importantly , we revealed the existence of descriptive rules for qualities that have not , to the best of our knowledge , been investigated before . These qualities include orange-blossom , hay , tarry and smoky . The generated rules will help scientists to better understand the chemical composition of the stimuli that evoke these odors and bring new insights about the way these molecules can interact with the olfactory system at the receptor level . Last but not least , our approach showed also that it was difficult to generate reliable rules for some qualities , particularly the most represented in the database ( e . g . , fruity , floral and woody ) . Although the recall associated with these rules was not high , they were characterized by a low rate of error , and validation was achieved for some of them including the well-known fruity and woody qualities . Finally , it is noticeable that a series of interesting qualities were described by rules with a good level of confidence but may be not precise enough to warrant detailed interpretation at this stage . These qualities are those that belong to the second quartile ( Fig 4 ) and include , for instance , camphor for which validation with novel odorants was performed using two different external datasets . A methodological issue that may be raised from our study relates to the choice of Arctander’s book in our methodology . Before answering this question , one must detail why such linguistic sources are used in olfactory research . In general terms , whereas emotional reactions are very prominent in olfaction [41] , lexical and linguistic processes are relatively limited: spontaneous odor identification performances are around 50% ( see [42] ) . Such an absence has led scientists and those in the industry to develop different sources ( atlases , books , websites ) listing the olfactory qualities of a series of odorant molecules ( Arctander book [24] , the Dravnieks Atlas [26] , the Boelens Atlas [27] , and the Flavornet website ( http://www . flavornet . org ) ) . A comparison of these sources led us to consider the Arctander’s book since it contained the highest number of odorant molecules and a reasonable number of qualities per odorant . The book , in being developed by a single scientist , gave the advantage of allowing us to integrate more homogeneous data with less variable response profiles than those collected in other atlases . However , this same feature also opens up the possibility that certain odorants that evoke a given quality could be missed . One should therefore not discard the possibility that certain molecules that evoke , for instance , the quality “fruity” were not considered by our model in the validation phase because they were just below the perceptual threshold set by Arctander for that particular quality . Given the variability of olfactory perception between individuals , it is conceivable that the same quality of “fruity” could have been the perceptual threshold of another rater . As a consequence of these factors , we face a double challenge: on one hand , there is a clear need to implement some flexibility in olfactory databases , whereby a given molecule can be described by one or several qualities with an associated level of confidence instead of a binary response . On the other hand , in order to account for interindividual variability in olfactory perception , olfactory databases need to consist of data from a large number of individuals . Future work will need to overcome these factors , for example , by asking raters to provide a level of confidence alongside each response , or by using a fuzzy logic algorithm in order to provide the model with responses ranging in quality from not at all plausible to extremely plausible . In this way , our model will benefit from a better characterization of olfactory percepts , as the rules generated would be more suited to the complexity of human perception . On a more general front , one interesting perspective in this research field would be to implement a new Atlas that integrates response diversity accompanied by all the strengths present in each individual atlas ( see Methods section; large number of molecules , large panel of evaluators in the qualitative description of each odor ) . Such an atlas could serve as a basis for a large number of: ( i ) fundamental research studies ( to better understand the perceptual olfactory space and its relation to the chemical space and the neuronal space ) , ( ii ) applied research studies ( to better understand the olfactory properties of new compounds developed by the perfume and flavor industry ) , ( iii ) education and teaching actions ( to standardize olfactory learning procedures in perfume schools or culinary arts schools ) . To sum up , current psychological and biological models of olfaction consider that olfactory perception is not totally universal . Although the sense of smell includes invariant aspects , a wide range of olfactory responses are characterized by their diversity from one person to another . In other words , while some molecules can induce very similar behavioral responses and perceptions among individuals , other molecules induce diverse perceptions , not only between individuals but also within the same person according to physiological and cognitive factors . It is undoubtedly in the invariant part of olfaction that we can establish the best predictive models linking chemistry to perception . In this case , a model including bijective rules can even be considered . Nevertheless , the more one moves towards the area of perceptual space of odors that is characterized by its heterogeneity between individuals , the higher the predictability threshold ( i . e . bad prediction ) becomes . This variability characterizes what could be called "the glass ceiling of olfactory diversity" . New methods are thus needed to break or circumvent this glass ceiling . Such methodology should integrate the notion of multiple rules for linking the chemical space to these diverse perceptions . Our approach is providing some new elements to this challenging issue . In conclusion , the present findings provide two important contributions to the fields of computation and neurosciences . First , although direct SOR seems illusory for some olfactory qualities if additional protagonists of the sense of smell are not taken into account , our approach suggests that descriptive rules exist for some qualities . Second , the present approach showed that several sub-rules should be taken into account when describing structure-odor relationships . From these findings , by correlating the multiple molecular properties of odors to their perceptual qualities and evoked-neural activities , experts in neuroscience and chemistry may generate new and innovative hypotheses in the field . In terms of application , this work can add to our knowledge of the complex phenomenon of smells and tastes . Indeed , by implementing such a descriptive structure/odor model within a dedicated data-analytics platform we could improve our understanding of the effects of molecular structure on the perception of those objects with highly-valued odorant properties such as foods , desserts , perfumes and flavors . This , in turn , would enable the optimization of product formulation with respect to the needs and expectations of consumers . | An important issue in olfaction sciences deals with the question of how a chemical information can be translated into percepts . This is known as the stimulus-percept problem . Here , we set out to better understand this issue by combining knowledge about the chemistry and cognition of smells with computational olfaction . We also assumed that not only one , but several physicochemical models may describe a given olfactory quality . To achieve this aim , a first challenge was to set up a database with ~1700 molecules characterized by chemical features and described by olfactory qualities ( e . g . fruity , woody ) . A second challenge consisted in developing a computational model enabling the discrimination of olfactory qualities based on these chemical features . By meeting these 2 challenges , we provided for several olfactory qualities new chemical models describing why an odorant molecule smells fruity or woody ( among others ) . For most qualities , multiple ( rather than a single ) chemical models were generated . These findings provide new elements of knowledge about the relationship between odorant chemistry and perception . They also make it possible to envisage concrete applications in the aroma and fragrance field where chemical characterization of smells is an important step in the design of new products . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"smell",
"chemical",
"compounds",
"statistics",
"social",
"sciences",
"neuroscience",
"data",
"mining",
"perception",
"physicochemical",
"properties",
"cognitive",
"psychology",
"scientists",
"forecasting",
"odorants",
"mathematics",
"materials",
"science",
"information",
"... | 2019 | Chemical features mining provides new descriptive structure-odor relationships |
Cells typically vary in their response to extracellular ligands . Receptor transport processes modulate ligand-receptor induced signal transduction and impact the variability in cellular responses . Here , we quantitatively characterized cellular variability in erythropoietin receptor ( EpoR ) trafficking at the single-cell level based on live-cell imaging and mathematical modeling . Using ensembles of single-cell mathematical models reduced parameter uncertainties and showed that rapid EpoR turnover , transport of internalized EpoR back to the plasma membrane , and degradation of Epo-EpoR complexes were essential for receptor trafficking . EpoR trafficking dynamics in adherent H838 lung cancer cells closely resembled the dynamics previously characterized by mathematical modeling in suspension cells , indicating that dynamic properties of the EpoR system are widely conserved . Receptor transport processes differed by one order of magnitude between individual cells . However , the concentration of activated Epo-EpoR complexes was less variable due to the correlated kinetics of opposing transport processes acting as a buffering system .
In cells external signals from ligands are transmitted by receptors to intracellular signaling cascades . Receptor signaling is regulated by receptor transport processes between the plasma membrane and other cellular compartments that are subsumed under the term receptor trafficking [1] . In absence of ligand , receptors are transported to the plasma membrane and are taken up again by the cell . After ligand binding , activated receptors at the plasma membrane can be internalized . To shut down signal transduction , endosomal acidification induces ligand dissociation from the receptor . Subsequently , the receptor is either degraded or transported back to the plasma membrane . These transport processes therefore strongly influence the ability of cells to integrate signals from external ligands and thereby the translation into cellular responses . In a variety of receptor systems , receptor trafficking was quantitatively studied by a combination of experiments and ODE models based on population average data [2–4] . For example , endocytosis , degradation and receptor recycling were quantitatively studied in the epidermal growth factor receptor ( EGFR ) [5–10] , the erythropoietin ( Epo ) receptor [11 , 12] , the insulin receptor [13 , 14] , chemotactic peptide receptors on neutrophils [15–17] , the transferrin receptor ( TfR ) [18 , 19] , the low density lipoprotein receptor ( LDLR ) [20 , 21] , interferon-α and tumor necrosis factor receptors [22 , 23] . These studies established a canonical receptor trafficking model that accounts for exchange of free receptors between the plasma membrane compartment and an intracellular receptor pool , internalization of ligand-bound receptors , degradation , and receptor recycling [2–4 , 24] . Quantifying receptor trafficking processes helped to characterize physiologically relevant differences between receptor systems . In particular , kinetic parameters for ligand binding , the internalization of free or ligand-bound receptors and for synthesis and degradation of receptors showed large differences between receptor systems , and could be used to categorize receptors according to functional roles in cells [2 , 4 , 24] . Growth factor receptors such as the EGFR are characterized by a high membrane abundance and a strongly accelerated internalization of ligand-bound compared to free receptors at the plasma membrane , a phenomenon denoted as ligand-induced receptor downregulation [5 , 15 , 25] . Due to an accelerated internalization upon ligand binding , short reaction times of receptor signaling to changes in ligand concentrations are facilitated [24] . From a systems perspective , this increases the accuracy of signal transduction within involved signaling pathways [4 , 24] . On the contrary , transport receptors as the TfR or the LDLR typically do not exhibit an accelerated internalization upon ligand binding but show a high rate of receptor internalization compared to the rate of ligand unbinding [24 , 26–28] . Cytokine receptors , as the EpoR or the interleukin 3 receptor , are characterized by a low membrane abundance and an efficient clearance of ligand from the medium and rapid recovery of receptor levels at the plasma membrane [4 , 12] . The last four decades contributed to a broad understanding of dynamic properties of receptor systems but most studies described receptor trafficking based on measurements of cell population averages . Because trafficking processes depend on a multitude of biochemical processes including for example vesicle formation and cytoskeleton-dependent transport [1 , 29] , heterogeneous expression of involved proteins can give rise to cell-to-cell variability [30] . In this context , an open question is whether cellular heterogeneity in different receptor trafficking processes can dissolve borders between categories of receptor systems , potentially leading to subpopulations of cells showing features as endocytic downregulation , fast replenishment or an efficient receptor recycling . As a result , cell-to-cell variability in receptor trafficking might cause a diverging behavior of cells in response to an external stimulus . For this reason , it is an important question whether receptor systems exhibit robustness to cellular variability in trafficking processes . A prime example for the importance of receptor transport processes in regulating systems properties is the receptor for the hormone erythropoietin ( Epo ) [11] . Ligand-induced signal transduction through this cytokine receptor , the EpoR , comprises primarily activation of JAK2/STAT5 , PI3K/AKT and MAPK pathways , and is absolutely essential for differentiation , proliferation and cell survival of erythroid progenitor cells to ensure renewal of mature erythrocytes [31 , 32] . Transport processes regulating EpoR induced signal transduction are ( i ) receptor internalization and inactivation followed by subsequent degradation , and ( ii ) receptor recycling encompassing ligand-induced receptor endocytosis and subsequent transport back to the plasma membrane [12 , 33] . It was reported that the activation of kinases and phosphatases [34] , ubiquitination of the receptor [35] , and cargo protein and cytoskeleton dependent processes such as assembly of actin oligomers [36] modulate transport of the EpoR . A characteristic property of the EpoR system is that only a small fraction of the total receptor amount is present at the cell surface [37 , 38] . By dynamic pathway modeling in combination with binding studies utilizing radioactively labeled Epo we recently showed that extremely rapid receptor turn-over ensures responsiveness of the system for a very broad ligand-concentration range as it is for example observed during continuous erythrocyte renewal and accelerated production in response to severe blood loss [11 , 12 , 39] . Further , data-based mathematical models revealed that ( 1 ) Epo-induced activation of the JAK2-STAT5 signaling cascade occurs in cycles continuously monitoring the activation status of the receptor [11 , 12 , 39] and ( 2 ) the two induced negative regulators bind to the receptor and divide the labor to control signaling for a wide range of Epo concentrations [31 , 32] . The so far established mathematical models were calibrated based on cell population data obtained for suspension cells . The kinetics at the level of single cells is smoothed and underlying biochemical signaling networks might be misinterpreted due to averaging population heterogeneities [40–42] . Furthermore , since the EpoR is also expressed on some tumor cells such as non-small cell lung carcinoma cell lines [43] , it of much interest to investigate to which degree principles learned in suspension cells can be transferred to adherent cancer cells . Here , we developed an approach based on live cell imaging , image segmentation of subcellular compartments , and cell ensemble models to investigate the extent of variability in receptor trafficking and interrelations between the dynamics of transport processes . Single-cell measurements of EpoR concentrations in different cellular compartments were used to estimate kinetic parameters of receptor trafficking processes for individual cells . By model discrimination we determined which receptor transport processes essentially contributed to receptor trafficking of EpoR . Calibrating cell ensemble models with a combination of single-cell datasets improved the identifiability of single-cell kinetic parameters , which was a prerequisite for analyzing correlations between kinetic parameters of receptor transport processes . Despite the large variability in the EpoR trafficking reactions we observed that the correlation between the kinetics of different transport processes had a buffering effect on the concentration of Epo-EpoR complexes at the plasma membrane and in the endosomal compartment . This correlation of the kinetics of different processes involved in the same cellular signaling system might represent a general motif of biological systems to confine cell-to-cell variability .
The EpoR is transported to the plasma membrane , can bind Epo , and is subjected to endocytosis , degradation and transport back to the plasma membrane [12 , 33] . To quantitatively study these processes at the single-cell level , we developed an approach employing an EpoR-GFP fusion protein ( EpoR-GFP ) and Epo labeled with the organic dye Cy5 . 5 ( Epo-Cy5 . 5 ) . The EpoR-GFP fusion protein was stably expressed in the NSCLC cell line H838 and a fluorescent membrane marker , mCherry fused to a myristoylation-palmitoylation ( MyrPalm ) domain ( MyrPalm-mCherry ) accumulating at the plasma membrane was co-expressed ( Fig 1A ) [44] . After recording the first image stack , cells were exposed to Epo-Cy5 . 5 at a concentration of 4 . 2nM corresponding to a biological activity of 10U/ml Epo [45] . Subsequently , Epo internalization was studied for at least five hours by recording three-dimensional stacks of confocal microscope images . Analyzing Epo-Cy5 . 5 in combination with EpoR-GFP and the membrane marker MyrPalm-mCherry enabled simultaneous recording of complementary information on Epo-uptake , EpoR-internalization and EpoR-degradation essential for studying protein turnover by kinetic modeling . While the GFP signal indicated the amount of EpoR-GFP and was affected by EpoR-GFP degradation , the Cy5 . 5 signal represented the sum of intact and degraded proteins since the dye molecule Cy5 . 5 is not targeted by protein degradation mechanisms . Intensities for membrane and cytosolic compartments were extracted from microscopic data to obtain time-resolved measurements , which were proportional to local concentrations of EpoR-GFP and Epo-Cy5 . 5 , and were used for model fitting . For this purpose , we developed a segmentation software to semi-automatically define three-dimensional regions of interest ( ROIs ) for the plasma membrane using the MyrPalm-mCherry signal , and for EpoR-GFP/Epo-Cy5 . 5 containing vesicles the EpoR-GFP and Epo-Cy5 . 5 signals ( Fig 1A to 1C , S1 Fig and S1 Movie; for details , see S1 Text ) . In the ROI for the plasma membrane , Epo-Cy5 . 5 intensities were associated with the amount of Epo-EpoR complexes , while EpoR-GFP intensities were associated with the total amount of EpoR ( Epo-ligated plus free receptors ) . Further , based on an intensity threshold for Cy5 . 5 , we distinguished between EpoR in voxels containing only EpoR-GFP or Epo-Cy5 . 5/EpoR-GFP ( S1 Text ) . We extracted the Cy5 . 5 fluorescence signal in the cytosolic compartment to obtain a quantitative measure of the amount of internalized Epo-Cy5 . 5 . Epo that was bound to internalized EpoR can be either secreted from the cell or degraded . Because Cy5 . 5 that was coupled to Epo is not proteolytically degraded , the intracellular Cy5 . 5 signal was assumed to reflect the amount of intact and degraded Epo . To obtain quantities for model fitting that were proportional to EpoR and Epo concentrations , intensities were normalized by cellular volumes , which were defined by the volumes enclosed by outer borders of membrane ROIs . The described procedure was applied to analyze for example 16 single Epo-treated H838 cells . As shown in Fig 1C for a representative single cell , we observed the strongest signal changes within the first hour after addition of Epo-Cy5 . 5 , indicating fast binding and internalization . The membrane EpoR-GFP fraction and the signal from EpoR-GFP vesicles showed in the exemplary cell only a slight increase , implying that Epo did not have a large influence on the total amount of the EpoR . On the contrary , the intensity from Epo-Cy5 . 5-containing vesicles continuously increased . While the Cy5 . 5 intensity at the plasma membrane reached a steady state after about ten minutes , intracellular Cy5 . 5 intensity showed a prolonged increase suggesting a slow decay of internalized Epo-Cy5 . 5 . We asked whether initial conditions such as EpoR concentrations in cellular compartments were predictive for EpoR trafficking in the presence of Epo , and evaluated associations between characteristic measures of single-cell trajectories before and after adding Epo-Cy5 . 5 . In particular , we examined which experimental quantities were predictive for the amount of membrane bound Epo-Cy5 . 5 , which can be assumed to reflect the amount of active EpoR [11 , 12] . For all Epo-treated cells , characteristic parameters were extracted from segmented imaging data , resulting in total EpoR concentrations or EpoR numbers in arbitrary units . Absolute numbers of EpoR-GFP or Epo-Cy5 . 5 in cellular compartments were estimated by summing up fluorescence intensities in segmented compartment ROIs , while cellular concentrations were estimated by dividing fluorescence intensity sums in cellular compartment ROIs by cell volumes . For scale-free comparisons , single-cell measures were divided by the means of all cells to obtain fold changes relative to single-cell averages ( Fig 1E–1G , S2 Fig ) . Among all cells , the membrane EpoR-GFP fraction contained on average 7 . 6% ( SD: 2 . 1% ) of the total cellular amount of EpoR-GFP . Interestingly , EpoR concentrations in the membrane ROI ( [EpoR-GFPmem] ) were significantly correlated with EpoR concentrations in intracellular vesicles ( [EpoR-GFPves]; Fig 1E; p = 0 . 0066 for Pearson correlation coefficients ) . This implies that the kinetics of EpoR transport from the cytosol to the plasma membrane was correlated with kinetics of EpoR transport from the plasma membrane back to the cytosol . The observation of correlated trafficking parameters will be further addressed below . Furthermore , while there was no significant correlation between the total cellular concentrations of EpoR-GFP ( [EpoR-GFPtot] ) and the concentration of Epo-Cy5 . 5 in the plasma membrane ROI ( [Epo-Cy5 . 5mem] ) at the end of the experiment after 5 hours ( Fig 1F; p = 0 . 25 ) , absolute amounts of cellular EpoR-GFP ( NEpoR-GFP , tot ) were significantly correlated with the amounts of membrane Epo-Cy5 . 5 ( NEpo-Cy5 . 5 , mem ) at 5h ( Fig 1G; p = 0 . 0083 ) . While total amounts of EpoR-GFP and internalized Epo-Cy5 . 5 were significantly correlated , there were no significant correlations between EpoR-GFP and Epo-Cy5 . 5 concentrations in different cell compartments ( S2 Fig ) , which indicates that the EpoR transport kinetics strongly varied between cells . Taken together , we established an experimental setup to quantitatively study the dynamics of the EpoR and the internalization of Epo by live-cell microscopy . To mechanistically study cell-to-cell variability in EpoR transport processes , we developed different mathematical models ( EpoR model ) based on ordinary differential equations ( ODE ) and estimated the model parameters by model fitting to single-cell measurements . The EpoR model variants , consisting of a basic model and variable extensions , described the two observed species , free EpoR and EpoR bound to Epo , in different cellular compartments or at the plasma membrane . The basic EpoR model describes reversible binding of Epo to the EpoR at the plasma membrane ( EpoRm ) and formation of active EpoR ( EpoRm* ) ( black arrows in Fig 2A ) . EpoR permanently cycle between the plasma membrane ( EpoRm ) and the intracellular compartment ( EpoRi ) . The intracellular pool of the EpoR is subject to degradation and refilled by synthesis . Active EpoR at the membrane EpoRm* are internalized to the endocytic recycling compartment ( EpoRRE* ) . In the model reaction describing EpoR binding to free Epo , Epo is not consumed because the amount of Epo in the medium largely exceeds the total amount of EpoR , as described in the methods section , and can therefore be assumed to remain constant . The basic model was extended by variable parts A to D , which described different possible ways for EpoR transport back to the plasma membrane or degradation . By appending variable combinations of parts A to D to the basic model , 16 possible model variants were formulated to systematically test the contribution of different processes to EpoR trafficking in our cellular system . Since receptor recycling and degradation of ligand-bound receptors were described for several receptor systems as the EGFR , IL3R or TfR [2 , 12 , 46–49] , we explored their role in EpoR trafficking in our cellular system , and whether their contribution was essential or could be neglected , which was not examined in previous modeling studies on EpoR trafficking . In model variants , internalized Epo is either released back into the extracellular space ( parts A and C ) or degraded ( Epodeg , i ) and accumulates inside the cell ( parts B and D , Fig 2A ) [11 , 12] . After internalization of Epo-EpoR complexes , receptors recycle back to the plasma membrane ( A and B ) or are degraded ( C and D , Fig 2A ) [12 , 23] . From the endocytic recycling compartment , receptors are recycled via path A directly to the membrane EpoRm or via B to the intracellular pool ( EpoRi ) . All model variants were fitted to data from our single-cell experiments . To enrich our experimental dataset by kinetic data on EpoR synthesis and degradation , we performed two auxiliary experiments . First , Epo-GFP expressing H838 cells were bleached by applying a short laser pulse . Thereafter replenishment due to EpoR-GFP synthesis was followed in ten treated H838 cells by recording the increase of the GFP signal . Furthermore , EpoR-GFP degradation was studied in seven single H838 cells treated with cycloheximide ( CHX ) at a concentration of 5μg/ml to inhibit protein translation and by recording the subsequent decrease of the GFP signal . Inhibition of translation by CHX was similarly used in previous systems biological studies to quantitatively study protein degradation [50–52] . The rationale for doing these additional experiments on EpoR synthesis and degradation was that the trafficking dynamics in unperturbed experiments are likely to be a complex superposition of EpoR endocytosis , recycling , synthesis and degradation effects . Therefore , we assumed that a combination with EpoR synthesis and degradation experiments were required to make kinetic parameters for EpoR turnover identifiable . In general , combining experiments on receptor trafficking with experiments on receptor turnover is reasonable because time scales of these processes might be different . For all model variants , cell ensemble models were constructed [40] . In the cell ensemble models , each single cell of a heterogeneous cell population was described by the same set of ODEs , and cell-to-cell variability was introduced by allowing receptor trafficking parameters and initial EpoR concentrations to be different between cells ( as further described below ) . Cell ensemble models comprised single-cell models for Epo internalizing cells , and simplified models for photobleached and CHX treated cells , in which reactions for Epo uptake were excluded . One single-cell model describing an Epo treated cell contained 6 ODEs and between 7 and 11 parameters ( S1–S3 Tables; for details , see S2 Text ) . Models of photobleached cells contained a reduced set of reactions describing only synthesis , degradation , transport of the EpoR between the plasma membrane and the intracellular pool , and an additional reaction describing removal of detectable EpoR species by photobleaching . Trajectories of CHX treated cells , in which synthesis was inhibited , were described by ODE models describing EpoR degradation and transport between the plasma membrane and the intracellular pool of EpoR ( S4 and S5 Tables ) . Models of photobleached cells consisted of 3 ODEs with 5 kinetic parameters while models of CHX treated cells contained 3 ODEs with 3 kinetic parameters . The parameters for Epo binding and unbinding , kon , Epo and koff , Epo were defined as being equal for each single-cell model , whereas all other kinetic parameters were allowed to vary between cells . This assumption was made , because kon and koff are biophysical constants , whereas receptor trafficking parameters describe lumped reactions that are controlled by concentrations of various intracellular regulatory proteins . Hence , in line with previous studies , we assumed in our model that cell-to-cell variability arises from heterogeneous expression of cellular proteins [40 , 53] . An ensemble model describing the complete available dataset of 16 Epo treated , 10 photobleached , and 7 CHX treated cells comprised between 156 and 220 kinetic parameters . Experimental single-cell datasets for GFP and mCherry fluorescence were linked via scaling factors to model variables in absolute concentration units . Taking together kinetic parameters , scaling factors , and initial concentrations [EpoRm] ( t0 ) and [EpoRi] ( t0 ) resulted in a total number of 230 to 294 parameters for different model variants , which were estimated by model fits of a total of 3996 data points . To estimate the scaling factor between normalized GFP fluorescence intensities in cellular compartment ROIs and absolute receptor amounts , average total cellular EpoR-GFP levels were determined by quantitative immunoblotting ( S3 Fig ) . Immunoblotting and image stack segmentations showed that each cell contained on average 142 . 000 receptors and had a mean volume of about 5 . 47pl , which resulted in an average cellular concentration of [EpoR]tot = 43 . 1nM . Fitting cell ensemble models to sets of single cells treated under different conditions , i . e . , by adding Epo-Cy5 . 5 , CHX or bleaching , can in principle lead to systematic differences between sets of estimated kinetic parameters . However , this is unlikely because the same cell line was used in all conditions . Therefore , kinetic parameters of cells treated under different conditions should follow the same probability distribution [40] . Because kinetic parameters of single cells implicitly depend on concentrations of regulatory proteins that are typically log-normally distributed in cell populations [54 , 55] , we assume log-normal distributions of single-cell parameters for EpoR trafficking processes , EpoR synthesis and degradation . To minimize differences between parameter distributions for the three experimental data sets generated by adding Epo-Cy5 . 5 , CHX or bleaching , we added constraint terms to the likelihood function used for parameter estimations , which penalized for differences in parameter means and variances between experimental sets ( for details , see S2 Text ) . Restricting parameter estimations by these constraint terms was advantageous with regard to model discrimination and parameter identifiability , as described below . We found that the model variant “ACD” , with parts for direct EpoR recycling to the plasma membrane ( part A ) and EpoR degradation with either exocytosis ( part C ) or intracellular accumulation of consumed Epo ( part D ) , could significantly better explain the set of experimental data than the other variants ( Fig 2B ) . This was indicated by the smallest values for the corrected Akaike information criterion ( AICcorr ) , which finds the most parsimonious model by weighing the number of parameters with goodness of fit and experimental noise , thereby preventing overfitting . Next , we compared the model selection results for different sets of experimental data . Thereby , we assessed to which degree cell ensemble models including constraint improved the model discrimination . Already the comparison between cell ensemble models calibrated solely with data from Epo-treated cells showed that the variant “ACD” performed significantly better than the other variants . Including data for bleached and CHX treated cells further increased the AICcorr difference to other variants and allowed more distinct model discrimination . In contrast , fitting model variants to data from only a single cell , instead of fitting cell ensemble models to data from several cells simultaneously , was not sufficient to determine an optimal model variant ( Fig 2C ) , a situation comparable to conventional ODE models calibrated only with population average data , which ignore cell-to-cell variability . The optimal model variant ACD is visualized in Fig 2D . The complete set of single-cell data for Epo internalizing , bleached or CHX treated cells is shown together with the best-fit ACD model trajectories in Fig 3 . In addition , scatter plots of experimental data plotted against corresponding model simulations are shown in S4 Fig , and residuals as well as residual distributions are shown in S5 Fig . Overall , it can be concluded that , our set of single cell data could be well explained by the model . The kinetic parameters associated with the reactions ( grey text in Fig 2D ) are further analyzed below . We hypothesized that cell ensemble models improved parameter estimations by combining complementary experimental datasets . To test this , we analyzed parameter identifiability for different combinations of datasets in cell ensemble models in comparison to individual single-cell models . Fig 4A visualizes relative confidence interval sizes , confidence intervals divided by parameter values , obtained from profile likelihood estimation ( PLE ) for parameters of four exemplary cells and different experimental datasets in a color-coded manner , Fig 4B for an exemplary parameter as error bars . Essentially , confidence interval sizes decreased significantly when using cell ensemble models instead of models fitted to data from one cell at a time , and for fitting cell ensemble models to data from all three experimental conditions instead of only Epo internalizing cells . For all parameters estimated in cell ensemble models , upper confidence intervals were defined by PLE . Only for few parameters , lower confidence intervals included zero indicating that those parameters were not identifiable and that involved reactions might be eliminated in these cells . Similarly , standard deviations from the best 0 . 5% of 1000 fits , ordered according to their squared sum of residuals , for all model parameters showed that combining datasets for Epo-internalizing H838 cells , bleached H838 cells and CHX treated H838 cells significantly improved the accuracy of single-cell parameter estimations ( S6–S8 Figs ) . In absence of constraint terms ( S8 Fig ) , single-cell estimates of EpoR transport parameters were of similar magnitude as in presence of constraint terms ( S6 Fig and S7 Fig ) which indicates that including constraint terms improved the identifiability of single-cell parameters but did not affect the variabilities of single-cell parameters . The globally defined parameters for Epo binding and unbinding were not identifiable , which were , however , not in the focus of this study . All scaling factors were identifiable with small confidence intervals ( S6 Table and S6 Fig ) . In summary , we found that the EpoR model variant ACD was optimal , which is consistent with EpoR trafficking reactions described in the model by Becker et al . that was developed based on cell population average data [12] . In comparison to the model by Becker et al . , our model additionally accounts for the intracellular pool of free EpoR , synthesis and degradation of the EpoR . We observed that our EpoR model could not be further reduced but that all components were required to explain the experimental data . Using cell ensemble models allowed clear discrimination between model variants and improved parameter identifiability . Improving the identifiability of single-cell parameters was necessary to analyze correlations between kinetic parameters within a population of cells , which will be further described below . After determining an optimal model variant , we asked how sub-compartment receptor pools remained largely unchanged in the presence of Epo and why intracellular ligand accumulation was slow . We investigated how EpoR trafficking reactions effectively contributed to these experimental observations . To this end , we extracted the concentrations of EpoR species from the model and analyzed fluxes ( concentration changes per minute ) through each of the reactions for each cell and at different time points . Model predictions of single-cell concentrations of EpoRm , EpoRi , EpoR*m and EpoR*RE , and reaction fluxes for all EpoR reactions are shown in Fig 5A and 5B . We superposed means and standard deviations for the best 0 . 5% of 1000 fits for single cells and average fluxes ( Fig 5B ) . After adding Epo , the largest fraction of the EpoR at the plasma membrane is quickly bound to Epo . The transport from the intracellular pool of EpoR ( EpoRi ) to the plasma membrane compensates for the internalization of Epo-bound EpoR ( EpoRm* ) resulting in EpoR concentrations , which are , in agreement with characteristics observed in single-cell trajectories ( Fig 1 ) , almost at steady state . Fluxes for EpoR recycling ( FEpoR* , REtoM ) reach similar magnitudes as fluxes of unoccupied EpoR from the intracellular pool to the plasma membrane ( FItoM ) . Reaction fluxes in different cells varied approximately by a factor of ten implying that EpoR transport dynamics and the consumption of Epo strongly diverge between cells , an observation , which is further analyzed below . Average fluxes at the end of the experiment ( t = 300’ ) , when fluxes were close to steady states , are illustrated in Fig 5C . Analysis of fluxes showed that a large fraction of internalized EpoR was recycled to the plasma membrane ( FEpoR* , REtoM ) , while a smaller receptor fraction was degraded , mostly with exocytosis of Epo . Notably , about one percent of the total amount of free EpoR cycles per minute between the plasma membrane and the intracellular compartment ( FItoM , FMtoI ) . To conclude , similar to previous studies [11 , 12] we observed an important contribution of receptor recycling and the fast transport of the receptor between the plasma membrane and the cytosol , and showed that the reaction fluxes varied approximately up to an order of magnitude between different cells . Next , we addressed how the observed strong variability in reaction fluxes affects signal transduction . Specifically , we asked how the concentration of Epo-EpoR complexes at the plasma membrane indicative for the fraction of activated receptors [11 , 12] , and the concentration of internalized Epo-EpoR complexes were dependent on EpoR transport processes . First , we compared our single-cell parameter estimates with the corresponding kinetic parameters from the mathematical model by Becker et al . [12] . Interestingly , although Becker et al . had used a different cellular system , the murine suspension cell line BaF3 stably expressing the EpoR instead of the human adherent NSCLC cell line H838 stably expressing the EpoR-GFP , all parameters from their population average data model were inside ranges of the single-cell parameters in our model ( Fig 6A ) , and were significantly correlated with single-cell parameter means ( ρ = 0 . 92 , p = 0 . 01 ) . As observed in the study by Becker et al . , the kinetic parameters for internalization of Epo-bound EpoR ( kEpoR* , MtoRE ) were in the range of the parameters for internalization of free EpoR at the plasma membrane ( kEpoR , MtoI ) , indicating that ligand binding did not substantially accelerate internalization . To further study cell-to-cell variability , we calculated the coefficients of variation ( CV ) , which equal standard deviations divided by means , for single-cell parameters and the concentration of Epo-EpoR complexes at the cell membrane after 5 hours of Epo-stimulation , [EpoR*m] ( 5h ) , and of internalized Epo-EpoR complexes [EpoR*RE] ( 5h ) , when reactions were close to a steady state . Of note , we analyzed the variability of kinetic parameters between cells , which should not be confused with analyzing parameter variances in one single-cell model to assess whether single cell parameters can be uniquely estimated . Here , identifiability of single-cell parameters and small parameter confidence intervals were prerequisites for analyzing the variabilities of parameters in a heterogeneous population of cells . For kinetic parameters , we observed large CVs of above one besides slightly smaller CVs of about 0 . 7 for the parameters for EpoR synthesis ( ksyn ) and for degradation of Epo-bound EpoR with exocytosis of consumed Epo ( kEpoR* , deg , REtoEx ) ( Fig 6B ) . However , for initial concentration estimates of EpoR , and of Epo-EpoR complexes after 5 hours , CVs had substantially smaller values between 0 . 2 and 0 . 5 . To analyze this divergence in variabilities , we determined concentration control coefficients for [EpoR*m] ( 5h ) and [EpoR*RE] ( 5h ) . Concentration control coefficients r were calculated as normalized derivatives of parameters k as r = k/[EpoR*m] ( 5h ) ∂[EpoR*m] ( 5h ) /∂k or r = k/[EpoR*RE] ( 5h ) ∂[EpoR*RE] ( 5h ) /∂k , and were expected to have values above one in case of strong sensitivity towards changes of a parameter and below one in case of weak sensitivity [56 , 57] . All control coefficients were smaller than one , indicating robustness of the system towards parameter changes ( Fig 6B ) . Importantly , the strong divergence between large CVs for kinetic parameters and a small CV for the concentration of Epo-EpoR complexes after 5 hours , [EpoR*m] ( 5h ) and [EpoR*RE] ( 5h ) , could be explained by positive correlations between kinetic parameters ( Fig 7 and S9 Fig ) . In particular , the parameters kEpoR , MtoI , kEpoR , ItoM , kEpoR* , MtoRE , and kEpoR*REtoM , which described EpoR transport reactions , were positively correlated with high significance ( Fig 7A and 7B , S9 Fig ) . The positive correlation of the parameters kEpoR , MtoI and kEpoR , ItoM was in line with the experimental observation that EpoR concentrations at the plasma membrane were correlated with EpoR concentrations in intracellular vesicles ( Fig 1E ) . Further , the kinetics of processes involved in increasing and decreasing Epo-EpoR complexes at the cell membrane [EpoR*m] or internalized Epo-EpoR complexes [EpoR*RE] were positively correlated , and therefore , variabilities canceled out . Intuitively , this positive correlation between opposing processes is biochemically reasonable because different transport processes depend on the same molecular key components , such as motor proteins or constituents of the cytoskeleton [29] , which will be discussed further below . Simulating the case , in which positive correlations between kinetic parameters were removed , could further illustrate to which degree positive correlation between EpoR trafficking processes reduced noise . To this end , we derived a multivariate log-normal parameter distribution from estimates of single-cell parameters . First , we sampled vectors of single-cell parameters from the derived multivariate distribution using the complete covariance matrix , and simulated values for Epo-EpoR complexes , [EpoR*m] and [EpoR*RE] , after 5 hours for each parameter vector . Then , we set covariances for parameters describing the transport of EpoR* ( kEpoR* , MtoRE , kEpoR* , REtoM , kEpoR* , deg , REtoEx , kEpoR* , deg , REtoI ) to zero , and again sampled parameter vectors from the modified multivariate distribution to simulate values for Epo-EpoR complexes [EpoR*m] and [EpoR*RE] after 5 hours . As expected , reducing parameter covariances resulted in a clear increase of the CV for [EpoR*m] ( 5h ) and [EpoR*RE] ( 5h ) , whereas sampling from the complete covariance matrix resulted in a CV similar to the value obtained from parameter estimates after model fitting ( Fig 6B ) . We concluded that positive correlations between single-cell parameters for intracellular EpoR transport processes reduced variability of the concentration of Epo-EpoR complexes at the plasma membrane , which implies that inter-relations between trafficking processes effectively dampened variability in the output of the system . Next , we explored which cell-to-cell differences were essential to describe the data . We tested , whether in any of the reactions , global parameter values could be used to describe the same reactions in different cells and allow further model simplification . Single-cell parameters in the optimal model variant ACD , which were estimated individually for each cell , were sequentially defined as global parameters that were equal for all cells . Only ksyn was allowed to be variable in every case to account for different EpoR concentrations in individual cells . After fitting restricted model versions to the experimental dataset , differences in AICcorr to the unrestricted model , in which all parameters could vary between cells , were calculated ( Fig 8A and 8B ) . Subsequent fixing of additional parameters causing the smallest increase in AICcorr showed that fixing the parameters for EpoR degradation ( kEpoR* , deg , REtoEx , kEpoR , deg , kEpoR* , deg , REtoI ) resulted only in subtle AICcorr increases ( Fig 8A , left panel; Fig 8B , lower trajectory ) suggesting that variability of these parameters was least important . On the contrary , sequential fixing of additional parameters causing the largest increase in AICcorr showed that variability of the parameters for EpoR transport to the plasma membrane ( kEpoR , ItoM ) and for EpoR internalization ( kEpoR , MtoI ) was most consequential ( Fig 8A , right panel; Fig 8B , upper trajectory ) . Apart from the distinct impact of parameter variabilities , AICcorr suggested that all variabilities were essential to fully describe the data indicating that the model could not be further simplified by assuming equal kinetic parameters for the same receptor transport processes in different cells . To conclude , using cell ensemble models instead of separate single-cell models , and including datasets for bleached and CHX treated cells in ensemble models improved the identifiability of single-cell parameters . The dynamics of EpoR transport processes was similar in adherent H838 cells as previously described for BaF3 suspension cells . Interestingly , a positive correlation between parameter describing opposing receptor trafficking processes provides an explanation for the observed moderate cell-to-cell variability of Epo-EpoR complex concentrations at the plasma membrane despite the large variability in the kinetics of EpoR transport processes in individual cells .
An interesting finding of this study was that single-cell parameter estimates indicated large cell-to-cell variability in EpoR transport processes , whereas the concentration of Epo-EpoR complexes at the plasma membrane representing the activated EpoR was much less variable . Model analysis showed that the positive correlations between kinetic parameters describing opposing EpoR transport processes effectively canceled out parameter variabilities and were responsible for the dampening of cellular heterogeneity in Epo-EpoR complexes at the cell membrane and in the intracellular compartment . Receptor trafficking parameters can be assumed to result from molecular properties of receptors and on the process of vesicle trafficking . Therefore , from the perspective of cellular physiology , two explanations can be considered to explain kinetic parameter correlations . First , properties of receptor molecules , their posttranslational modifications and effects due to receptor signaling might take influence on different trafficking processes in the same manner . Second , vesicle trafficking processes that are responsible for receptor transport to the plasma membrane , internalization of ligand-bound or free receptors might be co-regulated . This appears likely because vesicle trafficking reactions share key components involved in vesicle trafficking as microtubules , myosin or actin filaments that define common paths for vesicles . In general , vesicle trafficking typically requires a small number of different motor proteins , while adaptors bound to transport protein complexes , as Rab proteins that differentially regulate transport of different cargos , are more diverse [29 , 58] . Transport proteins are in some aspects co-regulated [29 , 59] , which might support synchronization of different trafficking processes . It was shown that the velocity of transport mediated by dyneins , myosins and kinesins is regulated by the concentration of ATP [60–63] , and that vesicle trafficking is slowed down after loss of ATP [64 , 65] . Therefore , the metabolic status of the cell might determine the kinetics of different vesicle transport processes and contribute to synchronized dynamics of transport processes . Moreover , overall correlations were observed for all cellular proteins , especially for proteins involved in the same biological pathways [66 , 67] . For this reason , also the kinetics of more specific trafficking mechanisms might be correlated , which are dependent on classes of regulatory proteins as kinesins or Rab GTPases [68 , 69] . Previous studies used trafficking parameters observed at the cell population level to categorize different receptors [2 , 4 , 24 , 70] . Accounting for variability in receptor trafficking changes this picture because features of different functional categories of receptors might coexist in cell populations . Therefore , to fully characterize the functional properties of receptor systems in cell populations , variances and covariances of single-cell kinetic parameters have to be taken into account . In ODE models describing comprehensively characterized cellular signal transduction networks , ODEs can reflect biochemical reactions in detail rather than summarizing several biochemical processes in single reactions . In this case , the same kinetic parameters can be assumed for different cells , and cell-to-cell variability can be inferred by different initial concentrations of involved signal transduction proteins [40 , 71 , 72] . As a consequence , correlations between signaling species become important for quantitative predictions . In a previous modeling study of programmed cell death , it was shown that for describing experimental data from a heterogeneous population of cells undergoing apoptosis , correlations between initial protein concentrations had to be taken into account to obtain realistic model predictions [53] . Furthermore , comparable to our study , correlations between initial concentrations of opposing signaling species , which were either anti- or pro-apoptotic , buffered variability of cell death times . The motif of limiting variability by correlated kinetics of opposing reactions can be seen in the context of other mechanisms , which limit variability in biological systems such as negative feedback , for example due to ligand-dependent receptor internalization or inhibition of upstream kinases by downstream kinases , or incoherent feed-forward loops [70 , 73–79] . Dampening of cell-to-cell variability by co-regulation of different trafficking processes would , however , not be regarded as a direct regulatory mechanism that results from the structure of a specific signal transduction network as it is the case for negative feedback loops . In-depth analysis of how different receptor transport processes are mechanistically inter-regulated and depend on the cellular population context that was shown to be relevant for explaining cell-to-cell variability in endosomal trafficking [30] , will be an important topic of future work . In several cellular systems , single-cell dynamics significantly deviate from the behavior observed at the level of cell populations [40 , 42 , 80] . On the contrary , we observed for the EpoR that the model by Becker et al . [12] , which was based on cell population average data , corresponded to the model variant that explained single-cell data best . In addition to the model by Becker et al . , our model accounts for the intracellular pool of free EpoR , for EpoR synthesis and degradation . Although , that study had used a different EpoR-expressing suspension cell line , in this study , we obtained similar kinetic parameters for EpoR trafficking in adherent EpoR-GFP expressing H838 cells . This finding suggests that dynamic properties of the EpoR system are conserved between different types of cells . We observed that internalization of Epo-EpoR complexes was not substantially accelerated compared to free EpoR , in contrast to other receptor systems , which is consistent with the finding that EpoR is internalized in a ligand-independent manner [81] , and was similarly observed in BaF3 cells at the cell population level [12] . Several other receptors as the epidermal growth factor receptor ( EGFR ) , the insulin receptor , the growth hormone receptor or the leukemia inhibitory factor receptor show substantially accelerated internalization of activated receptors [14 , 47 , 82–84] , which facilitates a high temporal resolution in receptor signaling [1 , 3 , 24] . Contrarily , EpoR signaling rather depends on fast transport of EpoR between membrane and cytosolic compartments and rapid ligand depletion [11 , 12] . Confocal microscopy combined with 3D image segmentation was the method of choice for the time-resolved quantification of fluorescently labeled proteins in cellular compartments but offered lower throughput compared to other experimental methods for studying cellular heterogeneity as fluorescence-activated cell sorting ( FACS ) . Nevertheless , significant correlations between single-cell parameters could be identified with the given set of single-cell data . An essential aspect of our study was the refinement of the cell ensemble modeling approach . These advances comprised the implementation of constraint terms [40] that minimized the deviations of kinetic parameter distributions in sets of single cells treated under different experimental conditions and that were added to the log-likelihood function for parameter estimations . The approach of merging single-cell trajectories from qualitatively different experiments is widely applicable and can be transferred to various other models of cellular signaling pathways . Taken together , we could show by combining quantitative live-cell imaging of erythropoietin receptor trafficking with mathematical modeling that receptor transport processes largely differed between individual cells . Receptor concentrations in cellular compartments were nevertheless robust to variability in trafficking processes due to the correlated kinetics of opposing transport processes .
Stable cell lines were generated from the human NSCLC cell line H838 ( ATCC CRL-5844 ) that was purchased from American Type Culture Collection ( ATCC , Manassas , VA , USA ) . From wild-type H838 cells , EpoR-GFP expressing cell lines were selected with 2 . 0 μg/ml puromycin ( Sigma-Aldrich , Taufkirchen , Germany ) , and MyrPalm-mCherry expressing cell lines were selected with 0 . 8 mg/ml G418 ( Sigma-Aldrich ) . Cell lines were maintained in DMEM ( Invitrogen , Darmstadt , Germany ) containing 10% fetal calf serum ( Biochrom AG , Berlin , Germany ) , 100 μg/ml penicillin and streptomycin ( Invitrogen ) . Medium for stably transfected cell lines additionally contained 0 . 2 mg/ml G418 or 0 . 2 μg/ml puromycin . For microscopy , cells were maintained in 8-well Lab-Tek chambers ( Thermo Scientific , Asheville , NC , USA ) with a density of 40 . 000/well . Before experiments , cells were washed and maintained in DMEM without growth factors for 3 hours to prevent basal phosphorylation of EpoR . We used the murine EpoR , which was well characterized in previous studies and is functionally equivalent to the human EpoR [12] . To express the fluorescently labeled EpoR , we utilized the retroviral expression vector pMOWS-puro encoding the murine EpoR C-terminally fused to GFP that was previously described in [85] and results in a protein of approximately 90kDa . For retroviral transduction of H838 cells , Phoenix ampho cells were transfected by the calcium phosphate precipitation method . Transducing supernatants were generated 24 hours after transfection by passing through a 0 . 45 μm filter ( Millipore , Billerica , MA , USA ) . H838 cells were treated with 1ml of supernatant supplemented with supplemented with 8 μg/ml polybrene ( Sigma-Aldrich ) on a 6-well plate at a density of 2·105 cells per well and spin-infected for 3h at 340g . Stably transduced H838 cells expressing EpoR-GFP were selected in the presence of 1 . 5μg/ml puromycin ( Sigma-Aldrich ) 24 hours after infection . The myristoylation-palmitoylation ( MyrPalm ) fusion construct with mCherry was a kind gift of Joel Beaudouin . It was constructed as described in [44] . To generate H838 cells stably expressing MyrPalm-mCherry and EpoR-GFP , EpoR-GFP expressing H838 cells were transfected with X-tremeGENE 9 ( Roche Pharma , Basel , Switzerland ) and selected with 2mg/ml G418 . Cells were treated with Epo-Cy5 . 5 ( Roche Diagnostics , Penzberg , Germany ) , which is a fully bioactive EpoR ligand [45] . Cy5 . 5 fluorescence was shown to be not pH dependent in the physiologic pH range , compared to fluorescein , because of its missing 3’-hydroxyl substituent [86] . Immunoblot samples were lysed with lysis buffer ( 20 mM Tris/HCl , pH 7 . 5 , 150 mM NaCl , 1 mM phenylmethylsulfonyl fluoride ( Sigma-Aldrich ) , protease inhibitor cocktail , 1% Triton X-100 ( Serva , Mannheim , Germany ) , and 10% glycerol ) . Cell lysates were analyzed using SDS PAGE gels ( Invitrogen ) . Proteins were transferred to PVDF membrane ( Merck Millipore ) using wet blotting . Detection was performed using the Pico Chemiluminescent Substrate from Thermo Scientific and a CCD camera ( Intas , Göttingen , Germany ) . EpoR-GFP concentrations in EpoR-GFP-expressing H838 cells were quantified utilizing recombinant eGFP ( BioVision , Mountain View , CA , USA ) . Cell lysates were combined with different amounts of GFP ranging from 0 . 2 to 10 ng and then loaded onto gels ( S3 Fig ) . To detect EpoR-GFP and GFP in immunoblots , we used an antibody recognizing GFP ( clones 7 . 1 and 13 . 1 ) from Roche ( Basel , Switzerland ) . Horseradish peroxidase-conjugated secondary antibodies ( Southern Biotech , Birmingham , AL , USA ) were used for detection . Live-cell experiments were performed in a 37°C , 5% CO2 incubation chamber on a CSU-22 Yokogawa spinning disk confocal ( Yokogawa Electric Corporation , Tokyo , Japan ) on a Nikon Ti inverted microscope equipped with 60x Plan Apo NA 1 . 4 objective lens ( Nikon , Tokio , Japan ) , a Hamamatsu C9100-02 EMCCD camera ( Hamamatsu Photonics , Hamamatsu , Japan ) and a PerkinElmer Photokinesis bleaching/photoactivation unit ( PerkinElmer , Waltham , MA , USA ) , using Volocity software ( PerkinElmer ) . GFP ( EpoR-GFP ) fluorescence was excited at 488 nm and collected with a 527/55 emission filter ( Chroma Technology Corp , Bellows Falls , VT , USA ) and an exposure time of 200 ms . For bleaching , we used the FRAP module of Volocity software . Cherry ( MyrPalm-mCherry ) fluorescence was excited at 561 nm and collected with a 615/70 emission filter ( Chroma Technology Corp ) and an exposure time of 300 ms . Cy5 . 5 ( Epo-Cy5 . 5 ) fluorescence was excited at 640 nm and collected with a 705/90 emission filter ( Chroma Technology Corp ) and an exposure time of 200 ms . Laser intensity was kept at a low level , at which no effect of bleaching was observed . A binning of 2x2 pixels was used . At each time point , z-stacks with 26 slides at 0 . 7μm step size were recorded . In live-cell imaging experiments , cells were treated with Epo-Cy5 . 5 at a concentration of 4 . 2 nM in a total volume of 400 μl . To facilitate an even distribution of the ligand , cells were kept in 200 μl medium while recording the first image stack , before adding 200 μl Epo-Cy5 . 5 at a concentration of 8 . 4 nM to obtain the desired concentration of 4 . 2 nM . Within the first 30 minutes , we recorded at a time interval of 5 minutes , afterwards at a time interval of 10 minutes to obtain more densely sampled measurements at the beginning of the experiment where the signal changes were strongest . Given an average flux FEpoR* , ItoM of Epo-EpoR complexes from the plasma membrane to the cytosol of 0 . 8 nM/min ( Fig 5B ) , and the average cell volume of 5 . 47 pl obtained from stack segmentations , the average number of Epo molecules internalized by a single cell will be about 440 molecules per minute . Therefore , within the experimental duration of 300 minutes , given the amount of 40 . 000 cells per well , about 3% of the total amount of Epo-Cy5 . 5 will be internalized in cells . It was shown that a fraction of the amount of Epo , which was secreted after internalization by cells , was still intact and could stimulate other cells [12] . Therefore , the fraction of Epo removed from the medium will be effectively less than 3% . This justifies the model assumption of constant Epo-Cy5 . 5 concentrations in the medium . We developed custom graphical user interface-based software in MATLAB ( The Mathworks , Natick , MA , USA ) for segmentation of cellular compartments from image stacks ( S1 Fig; for details , see S1 Text ) . MyrPalm-mCherry signals were used to segment the plasma membrane region of interest ( ROI ) , EpoR-GFP and Epo-Cy5 . 5 signals were used to define EpoR or EpoR-Epo vesicles . To obtain observables that were proportional to variable concentrations , fluorescence intensities were normalized by cell volumes ( for details , see S2 Text ) . Absolute volumes were calculated by multiplying voxel numbers and voxel volumes of 0 . 29x0 . 29x0 . 7μm3 . All ODE models were implemented with the MATLAB toolbox PottersWheel that was used for parameter calibrations ( http://www . potterswheel . de ) [87] . Model analysis and simulations were performed with custom MATLAB scripts . As a measure for the goodness of fit , we used the corrected Akaike information criterion ( AICcorr ) . Model equations can be found in S1 , S2 , S4 and S5 Tables , and parameter estimates in S6 and S7 Tables ( for details , see S2 Text ) . To test for linear correlation , we calculated Pearson correlation coefficients . | Cell surface receptors translate extracellular ligand concentrations to intracellular responses . Receptor transport between the plasma membrane and other cellular compartments regulates the number of accessible receptors at the plasma membrane that determines the strength of downstream pathway activation at a given ligand concentration . In cell populations , pathway activation strength and cellular responses vary between cells . Understanding origins of cell-to-cell variability is highly relevant for cancer research , motivated by the problem of fractional killing by chemotherapies and development of resistance in subpopulations of tumor cells . The erythropoietin receptor ( EpoR ) is a characteristic example of a receptor system that strongly depends on receptor transport processes . It is involved in several cellular processes , such as differentiation or proliferation , regulates the renewal of erythrocytes , and is expressed in several tumors . To investigate the involvement of receptor transport processes in cell-to-cell variability , we quantitatively characterized trafficking of EpoR in individual cells by combining live-cell imaging with mathematical modeling . Thereby , we found that EpoR dynamics was strongly dependent on rapid receptor transport and turnover . Interestingly , although transport processes largely differed between individual cells , receptor concentrations in cellular compartments were robust to variability in trafficking processes due to the correlated kinetics of opposing transport processes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"cell",
"physiology",
"medicine",
"and",
"health",
"sciences",
"vesicles",
"endocrine",
"physiology",
"hormone",
"transport",
"membrane",
"receptor",
"signaling",
"cellular",
"structures",
"and",
"organelles",
"intracellular",
"receptors",
"proteins",
"endocrinology",
"me... | 2017 | Correlated receptor transport processes buffer single-cell heterogeneity |
The precision with which visual information can be recalled from working memory declines as the number of items in memory increases . This finding has been explained in terms of the distribution of a limited representational resource between items . Here we investigated how the sensory strength of memoranda affects resource allocation . We manipulated signal strength of an orientation stimulus in two ways: we varied the internal ( sensory ) noise by adjusting stimulus contrast , and varied the external ( stimulus ) noise by altering the within-stimulus variability . Both manipulations had similar effects on the precision with which the orientation could be recalled , but differed in their impact on memory for other stimuli . These results indicate that increasing internal noise released resources that could be used to store other stimuli more precisely; increasing external noise had no such effect . We show that these observations can be captured by a simple neural model of working memory encoding , in which spiking activity takes on the role of the limited resource .
The fidelity with which items are stored in visual working memory ( VWM ) depends on the total number of elements in a visual scene . As the number of items increases , their representations in memory become increasingly variable , leading to less precise recall of each item [1–3] . These findings are consistent with a resource model of VWM , which describes a limited representational medium that is allocated to objects within a visual scene [1 , 4–11] . Consistent with the resource concept , rather than an even storage resolution for all remembered objects , it has been shown that resources can be flexibly distributed . Both external ( e . g . visual salience ) and internal ( e . g . behavioral relevance ) factors can lead to enhanced storage resolution of an object , but with a corresponding cost to the fidelity of other items held in memory [12–19] . Note that some authors argue there is a fixed upper limit on the number of stimuli that working memory resources can be allocated to ( e . g . [20] ) , but this is not a topic of the present study . A recently proposed neural resource model [21 , 22] has provided a biological account of behaviorally observed VWM limitations that is consistent with neurophysiological findings [23–25] . In this neural model , items are encoded in a population of tuned neurons that fire stochastically according to a Poisson process . The total activity in the neural population is constant across changes in set size , as the result of a global normalization mechanism ( see [11] for further discussion and e . g . [26 , 27] for neurophysiological evidence of analogous mechanisms ) . Therefore , increasing the set size leads to a decrease in the activity of neurons encoding each item , providing a biological basis for limited resources . As the neural signal representing each object decreases , internal representations become dominated by random noise in spiking activity , leading to a decrease in memory precision . Neural noise , an inherent feature of neural processing [28] , therefore determines the fidelity of internal representations and represents the limiting factor on VWM performance . Since storing each additional element in VWM takes resources away from existing items , it is of importance to understand what determines an object’s share of resource . One candidate as a limiting factor is the strength of the sensory stimulation , which determines the extent to which noise in the visual system can corrupt the stimulus representation . The importance of the visual system’s internal noise has long been recognized ( e . g . , [29] ) and many studies have aimed to characterize its effects on detection and discrimination performance [30–34] . In a recent study [35] , using an analogue report task with a single item , internal noise was modulated by varying the contrast of the orientation stimulus to be reported . Results showed precision declined as contrast decreased . A population coding model successfully accounted for the detailed pattern of errors by decreasing the total activity in the neural population . One intuitive prediction arising from this model is that , in the case of multiple items , the neural resource ( spiking activity ) released when storing a low contrast item could be used to store other items . Often , efficient visual processing is not only hampered by internal noise in the visual system , but also by stimulus uncertainty ( i . e . external noise ) which is intrinsic to natural circumstances . The complexity of our visual environment is likely to preclude an even distribution of resources , but it is less clear how exactly stimulus variance affects resource allocation . It has been demonstrated that increasing the external ( stimulus ) noise leads to poorer performance in perceptual tasks [36–39] . For example , judging the average feature value from an array becomes less accurate as uncertainty increases , i . e . as the variance of elements in the relevant feature dimension increases ( e . g . [39] ) . A recent study [40] demonstrated that external noise had an opposing effect to internal noise on perceptual biases , consistent with a model of efficient coding [41] . The present study aimed to investigate resource distribution among memoranda with differing sensory strengths and variabilities . Using a cued recall task we looked for changes in recall precision while manipulating the stimulus in order to vary the influence of either internal ( Experiment 1 ) or external ( Experiment 2 ) noise . We found similar effects of changing internal or external noise on recall of the noisy stimuli themselves . However , the consequences for other stimuli in memory differed , indicating that only changes to internal noise released neural resources that could be used to store other stimuli more precisely .
To investigate working memory resource allocation among stimuli affected to different degrees by internal noise , we presented participants with pairs of orientation stimuli of varying contrasts ( Fig 1a ) . One stimulus on each trial had fixed high contrast while the contrast of the other varied as a percentage of each participant’s detection threshold . To capture performance on this task we calculated the mean absolute deviation of recall responses from the mean of the response distribution . Fig 2a depicts these values for low noise ( blue circles ) and variable noise ( red circles ) stimuli . Error distributions for each stimulus and contrast level are shown in Fig 3a & 3b . All distributions , except for the 0% and 75% contrast variable-noise stimuli , displayed significant central tendency ( V > 63 . 8 , p < 0 . 001 ) . Overall , participants performed better when recalling low noise than variable noise stimuli ( F ( 1 , 8 ) = 839 . 7 , p < 0 . 001 , η p 2 = 0 . 99; ANOVA , contrast × probed stimulus ) and when total noise in a trial decreased ( F ( 3 , 24 ) = 14 . 3 , p < 0 . 001 , η p 2 = 0 . 64 ) . A significant stimulus contrast × probed stimulus interaction was observed ( F ( 3 , 24 ) = 37 . 4 , p < 0 . 001 , η p 2 = 0 . 82 ) consistent with a benefit on variable noise stimulus recall and detriment on low noise stimulus recall as contrast of the variable noise stimulus increased . To test this statistically , we compared recall performance on trials when both Gabors had high ( 400% ) contrast with trials where contrast of one Gabor was zero . Recall of variable noise stimuli improved with increasing contrast ( t ( 8 ) = 14 . 2; p < 0 . 001 ) . Critically , recall performance for low noise stimuli worsened as the contrast of the variable noise stimulus increased ( t ( 8 ) = 3 . 2 , p = 0 . 013 ) . These results indicate that increasing the internal noise associated with one stimulus has a beneficial effect on recall of the other stimulus , consistent with a transfer of working memory resources from the noisier to the less noisy stimulus . The goal of this experiment was to explore working memory resource allocation among stimuli with differing amounts of external noise . On each trial we presented participants with two composite stimuli each consisting of a set of Gabors of differing variance . For one , low noise stimulus , within-set variance was always zero ( all Gabors were parallel ) , while orientations for the other , variable noise stimulus were sampled from distributions of different widths . Participants were required to report from memory the average orientation of a probed stimulus . Fig 2b plots performance for variable noise ( red circles ) and low noise ( blue circles ) stimuli . Error distributions for each stimulus and contrast level are shown in Fig 3c & 3d . All distributions , except for the zero precision ( random ) variable-noise stimulus , displayed significant central tendency indicating better-than-chance performance ( V > 122 . 2 , p < 0 . 001 ) . Recall was more precise for low noise than variable noise stimuli ( F ( 1 , 9 ) = 201 . 5 , p < 0 . 001 , η p 2 = 0 . 96 ) , and when the total amount of noise on a trial decreased ( F ( 3 , 27 ) = 28 . 7 , p < 0 . 001 , η p 2 = 0 . 76 ) . A significant stimulus contrast × probed stimulus interaction was again observed ( F ( 3 , 27 ) = 16 . 37 , p < 0 . 001 , η p 2 = 0 . 65 ) . Next , we compared performance on trials with contrasting amounts of noise , i . e . variable noise stimuli in which all orientations were alike ( κ = ∞ ) versus those drawn from a uniform distribution ( κ = 0 ) . A decrease in the noise level of the variable noise stimulus increased recall performance for that stimulus ( t ( 9 ) = 13 , p < 0 . 001 ) , but had no effect on recall of the other , low noise stimulus ( t ( 9 ) = 1 . 52 , p = 0 . 16 ) . In contrast to the findings of Exp 1 , therefore , changing the amount of external noise for one stimulus did not influence recall of the other , consistent with a fixed resource allocation in the presence of changing stimulus noise . We examined whether participants’ estimates were biased towards particular elements of composite stimuli . To this end , we distinguished the four elements in each set by location relative to fixation: horizontal near , horizontal far , vertical above , vertical below . We observed no significant differences in the mean absolute deviation of responses around the orientations of the different component elements ( location × noise level ANOVA; location: F ( 3 , 27 ) = 0 . 8 , p = 0 . 51; interaction: F ( 9 , 81 ) = 1 . 7 , p = 0 . 10 ) , indicating that participants were not systematically biased towards particular elements , and suggesting that participants sampled all the elements on a roughly equal basis to estimate the average orientation of the stimulus .
Recent research has established that , as the working memory resource available to store an item decreases ( e . g . as the number of items increases ) , recall of that item becomes more noisy . Here we address the converse question: does storing a stimulus with a greater level of noise require less resource ? We considered two methods of manipulating the noise associated with a stimulus: ( 1 ) decreasing the sensory strength of the stimulus , which increases the influence of internal noise , and ( 2 ) increasing the intrinsic variability of a stimulus , a modulation of external noise . We found that both noise manipulations had similar effects on the fidelity with which a stimulus could be recalled . However , they differed in their effects on recall of other items in memory . We found that increasing internal noise in one stimulus , by decreasing its contrast , had a favourable effect on recall of another ( low-noise ) stimulus . Increasing external noise , by raising internal variability of a stimulus , had no effect on memory for another stimulus . These results indicate that increases in internal , but not external , noise reduce the resource required to store a stimulus . To account for these patterns of performance , we fit a variant of the neural resource model [21] to the behavioral data . In this model , stimulus features are encoded in spiking activity of a neural population . The total population activity is fixed , representing a finite resource that can be distributed between memoranda . This model has previously been shown to successfully reproduce the patterns of human recall error observed with changes in set size and shifts in stimulus priority [21] . When the contrast of a stimulus decreased , the firing rate associated with that stimulus in the model decreased as well , matching typical behavior of visual neurons in the brain . As a consequence , the representation was more influenced by stochasticity in spiking activity , and this had the consequence of poorer recall performance for that stimulus . Furthermore , due to normalization , the reduced activity devoted to the low contrast stimulus had the contingent effect of enhancing activity for another stimulus stored alongside it . In other words , dedicating less of the fixed neural activity to encoding the weak signal released resources that could be used to encode another signal more precisely . This model successfully captured not only the trade-off in precision between items of different sensory strengths ( Fig 2 ) , but also the detailed pattern of response errors in recall of each stimulus ( Fig 3 ) . Next , we used the same population coding model to fit the observations under different levels of stimulus variability ( external noise ) . Previous studies indicate that forming ensemble representations is a fundamental feature of the visual system [42–45] , and that extracting summary statistics , such as averages , from a visual scene occurs automatically [46] . Here , we hypothesised that the averaging process would introduce variance in the stored estimate that was proportional to the variance in the samples , but have no effect on the overall firing rate of neurons encoding the stimulus . Storage of the calculated average orientation in working memory was treated identically to storing a single stimulus orientation . With these assumptions , the neural resource model successfully captured behavioral observations from the experiment . Increasing levels of external noise decreased the precision with which a stimulus could be recalled , but did not necessitate unequal allocation of neural activity , meaning that there was no contingent effect on other , simultaneously-stored stimuli . Note that at least one prediction of this account , that an averaged stimulus is represented with the same level of activity regardless of the variability in its component values , would be open to testing using single-cell or imaging techniques . The two experiments differed slightly in the placement of stimuli in the visual field , in that the orientation stimuli in Exp 2 were presented on average a little more peripherally , and over a broader range of eccentricity , than the stimuli in Exp 1 . Given that eccentricities of the two memory items presented on each trial were always equal , we are not aware of any reason why stimulus eccentricity would influence resource distribution between them , nor of any related mechanism that could account for the qualitative difference in results between Exps 1 and 2 . Previously it has been suggested that errors in working memory tasks can be explained by a doubly stochastic model , in which there is variability in the precision with which items are represented [5–7 , 47] . Although the population coding model has only one source of stochasticity , a mathematical similarity between variable precision and the population coding model has been highlighted previously [21] , and may explain the success of variable precision models in reproducing empirical error distributions . Because a process for generating variability is not specified in these models , however , they make no clear predictions about the effects of stimulus strength on performance: it is therefore difficult to see how they could account for the present results in anything but an ad hoc fashion . At a more abstract level , it has been suggested that working memory can be viewed as a fixed capacity information channel [48]: this accounts for decreases in precision with set size and , with an appropriate choice of loss function , even predicts the specific error distributions observed in recall tasks [49] ( although not currently swap errors; see below ) . Like the variable precision model , this model may be best seen as operating at a different conceptual level rather than being a competitor to the neural resource model . Unlike the neural resource model , this model cannot offer an explanation as to why less information is stored about a weaker stimulus , but it may be able to account for the consequence that more information is stored about other stimuli . Swap errors , in which a wrong ( i . e . unprobed ) item is reported , are an important component of working memory performance at higher set sizes and when spatial confusability is high [4 , 50 , 51] . However , they are typically found to be vanishingly rare when , as here , only two well-separated items are presented for recall . A recent extension to the neural resource model based on conjunctive coding has demonstrated that swap errors can be explained using the same neural mechanism as response variability [22] . Future work using larger stimulus arrays could examine the effects of stimulus strength on swap errors . We considered the possibility that the precision trade-off between low and high contrast stimuli in Exp 1 arose from biasing of visual attention towards the higher contrast item due to its greater perceptual salience . We repeated the experiment with a sequential mode of presentation , such that only one stimulus was visible on the screen at a time . As with simultaneous presentation , the results showed that recall precision for a stimulus of fixed high contrast varied with the contrast of the other stimulus presented on the same trial , confirming that our effects are due to competition for memory , rather than attentional , resources . This is consistent with previous results demonstrating that resources are distributed between memory stimuli similarly when presented sequentially as when presented simultaneously [14] . The seemingly continuous decline in recall precision of low noise stimuli with increasing variable noise contrast could reflect a mixture of trials in which ( 1 ) the variable noise stimulus was not encoded and the low noise stimulus was stored with high precision , and ( 2 ) both stimuli were encoded and precision of the low noise stimulus was thereby reduced , according to resource principles . Although this alternative account of the data would be no less consistent with allocation of a limited memory resource , we nonetheless addressed this possibility by fitting a number of alternative models that incorporated encoding failures in different ways ( see Supporting Information ) . Consistent with results of [35] , we found that the neural resource model had a consistent advantage in accounting for the data in each case . Whether or not there exists a threshold for conscious perception is a debate with a lengthy history and a substantial existing literature , which we will not attempt to review here . Threshold models of various kinds have typically not fared well in comparison to signal detection theory in capturing psychophysical performance [52] , but there is continuing disagreement in the field [53–55] . While the present results of model comparison , like those of [35] , are consistent with a graded view of perception , in which even the weakest signals are ( weakly ) encoded , we do not imagine the present study can provide a definitive resolution to this question . In summary , using a cued recall task and manipulating the signal strength of memoranda , we found that changing levels of internal , but not external noise , frees working memory resources , which can then be allocated to other items in the visual scene . These results are consistent with a neural resource model of working memory , in which representational fidelity is limited by a fixed total activation in neural populations encoding the stimuli .
The study was approved by the Cambridge Psychology Research Ethics Committee . Thirty-one participants ( 10 males , 21 females; aged 18–33 ) took part in the study after giving informed consent in accordance with the Declaration of Helsinki . All participants reported normal color vision and normal or corrected-to-normal visual acuity . Stimuli were presented on a 21-inch gamma-corrected CRT monitor with a refresh rate of 60 Hz . The monitor was fitted with a neutral density filter to decrease the luminance range to the level of human detection thresholds . Participants viewed the monitor in a dark room at a distance of 50 cm , with head stabilized by a forehead and chin rest . Eye position was monitored online at 1000 Hz using an infrared eye tracker ( SR Research ) . Stimulus presentation and response registration were controlled by a script written in Psychtoolbox 3 . 0 . 14 ( Pelli , 1997 ) and run using Matlab 2016b ( The Mathworks Inc . ) . Ten participants ( 3 males , 7 females; aged 20–29 ) took part in Experiment 1 . In this experiment , we tested how internal noise affects working memory performance by manipulating stimulus contrast . Visual stimuli consisted of randomly oriented Gabor patches ( wavelength of sinusoid , 0 . 75° of visual angle ( 1 . 33 cpd ) ; s . d . of Gaussian envelope , 0 . 75° ) presented on a grey background . A central fixation dot and two circles ( white , 2° radius ) , located 5° to the left and to the right of the fixation dot were present throughout each trial . The circles served as placeholders for the Gabor patches . We first obtained a detection threshold for each participant using the adaptive Psi method [56] . On each trial ( 100 in total ) a Gabor patch of random orientation was presented for 200 ms within one of the two placeholder circles , randomly chosen; participants reported the side on which they saw the stimulus by a mouse click in the area inside the placeholder . Detection threshold was determined as the contrast level corresponding to 75% correct performance , estimated by fitting a cumulative normal function to the data . Contrast level on each trial was chosen to maximize the expected gain in information of the fitted psychometric parameters . The main part of Experiment 1 was a cued recall task that tested memory for orientation ( Fig 1a ) . Each trial started with presentation of the central fixation dot ( gray ) and placeholders . Once a stable fixation was recorded , the fixation dot turned white for 500 ms , followed by presentation of a memory display consisting of two randomly oriented Gabor patches for 200 ms . The contrast level of one patch ( randomly chosen ) was always 400% of the previously obtained detection threshold ( low noise stimulus ) . The contrast level of the other Gabor was chosen at random from {0% , 75% , 100% , 150% , 400%} of detection threshold ( variable noise stimulus ) . After a 1000 ms blank retention interval , one of the two stimuli was randomly cued for recall . The cue consisted of a second , larger circle drawn around one of the placeholders and overlaid with a randomly-oriented bar . Using a mouse , participants adjusted the bar orientation to match the remembered orientation corresponding to the cued location . Once participants had made their response , a second cue was presented at the location of the variable noise patch and participants indicated how confident they were that a patch had been presented at that location by clicking on one of five buttons labelled {0% , 25% , 50% , 75% , 100%} . Trials on which gaze deviated more than 2° from the fixation dot before the cue was presented were restarted with new random orientations . Participants completed 400 trials . One participant was excluded from analysis because data indicated poor comprehension of the confidence task instructions ( uniform distribution of confidence ratings across all contrasts ) . Eleven participants ( 4 males , 7 females; aged 18–33 years ) took part in Experiment 2 . In this experiment we examined how external noise affected working memory performance by manipulating stimulus variability . Except where indicated , the procedure and stimulus timing were identical to Exp 1 . Stimuli consisted of randomly oriented Gabor patches ( wavelength of sinusoid , 1 . 5° of visual angle ( 0 . 67 cpd ) ; s . d . of Gaussian envelope , 0 . 9° ) assembled in sets of four and presented on a grey background . Throughout the trial a central fixation dot and two placeholder circles ( white , 3 . 5° radius ) , located 6° to the left and right of the fixation dot , were present . On each trial , two sets of Gabor patches were presented for 1000 ms , one in each placeholder circle ( Fig 1b ) . Participants were required to memorize the mean orientation of each set . The four patches within each set were positioned with equal spacing on an invisible circle ( radius 2° ) , centered inside the placeholder and rotated randomly from trial to trial . The variability of one set ( left or right , randomly selected ) was fixed at zero , i . e . all constituent Gabor patches had the same orientation ( low noise stimulus ) . The variability of the other set ( variable noise stimulus ) was determined by sampling orientations from a von Mises distribution with randomly-assigned mean and concentration ( inverse variability ) chosen from {0 , 1 , 2 , 5 , ∞} . Note that a von Mises with concentration of zero is equivalent to a uniform distribution , i . e . patches had random orientations , while an infinite concentration meant that all patches had the same orientation . Cue and response were the same as in Exp 1 . Confidence ratings were not obtained in Exp 2 . Participants completed 400 trials with two sets of Gabors as described above . On an additional 40 trials ( randomly interleaved ) , to obtain a measure of baseline performance , a single low noise set was presented ( randomly left or right ) and cued for recall . One participant was excluded due to reporting on debriefing that they had ignored high noise stimuli . Orientations were analysed and are reported with respect to the circular parameter space of possible values , i . e . the space of possible orientations [−90° , 90° ) was mapped onto the circular space [−π , π ) radians . Recall variability was assessed as the mean absolute deviation of recall errors from the mean of the error distribution . Tests of central tendency were performed using the V test [57] on pooled data within each condition . When testing model fits , estimates of the parameters were obtained separately for each subject over all experimental conditions using a nonlinear optimization algorithm ( fmincon in MATLAB ) . We modeled orientation and contrast information presented at each stimulus location as providing input to an independent subpopulation of M neurons ( Fig 5 ) . The unnormalized response of the ith neuron responding to the jth stimulus was defined as the product of a von Mises ( bell-shaped ) orientation tuning function and a Naka-Rushton contrast response function: f i j ( θ j , c j ) = g i j ( θ j ) h j ( c j ) , ( 1 ) g i j ( θ j ) = 1 I 0 ( κ ) exp ( κ cos ( θ j - φ i j ) ) , ( 2 ) h j ( c j ) = c j α σ α + c j α , ( 3 ) where θj is the orientation of the jth stimulus , cj is its contrast , φij is the neuron’s preferred orientation , κ is a scale parameter for the tuning function , σ and α are parameters of the contrast response function , and I0 ( ⋅ ) is the modified Bessel function of the first kind with order zero . Preferred orientations were evenly distributed throughout the circular space of possible values . Normalization operated over the entire population of neurons , such that the post-normalization output of a neuron ( its firing rate ) was given by: r i j = γ M f i j ( θ j , c j ) ∑ k ℓ f k ℓ ( θ ℓ , c ℓ ) , ( 4 ) where γ is the population gain ( i . e . summed population activity ) . Assuming that the distribution of tuning curves provides a dense uniform coverage of the orientation space ( valid for large M ) , the summed activation of the population is independent of stimulus orientation and Eq 4 simplifies to: r i j = γ M g i j ( θ j ) h j ( c j ) ∑ ℓ h ℓ ( c ℓ ) . ( 5 ) Spiking activity was modeled as a homogeneous Poisson process , such that the probability of a neuron generating n spikes in time T was: Pr[ nij ]= ( rijT ) nijnij ! exp ( −rijT ) . ( 6 ) Recall of the orientation of a probed item p was modelled as maximum a posteriori ( MAP ) decoding of feature information from the population’s spiking activity , n , over a decoding period Td . Assuming a uniform prior , this is equivalent to maximizing the likelihood: θ ^ M A P = arg max θ p Pr [ n | θ p ] . ( 7 ) If two or more orientations tied for the maximum , the decoded value was sampled at random from the tied values . The output of the model was given by θ ^ = θ M A P ⊕ β , where β is a response bias term , and ⊕ indicates addition on the circle . For simplicity we fixed the decoding period Td to 1 s ( changing this value would merely result in a corresponding change to the estimated gain parameter γ , e . g . setting Td = 0 . 1 s would multiply the gain by 10 ) . In Exp 2 , each stimulus consisted of a set of Gabor patches with varying orientation , and participants were required to store in memory the mean orientation of the patches . We assumed that the concentration ( inverse variability ) of the estimated mean was directly proportional to the concentration of the distribution from which the orientations were sampled , i . e . θ j ∼ VM ( μ j ′ , s κ j ′ ) , ( 8 ) where μ j ′ and κ j ′ are the mean and concentration of the sampled distribution , s is a constant of proportionality , and VM ( μ , κ ) is a von Mises distribution with mean μ and concentration κ . In order to fit the model to data , instead of relying on Monte Carlo simulation , we used a number of previously-obtained analytical results that provided a direct way of estimating the distribution of errors predicted under the model , on the assumption M → ∞; details of these methods can be found in Bays ( 2016 ) ; code is available at www . bayslab . com/code/JN14 . In fitting the data from Exp 1 , the model had five free parameters: κ , γ , α , σ , and β . In Exp 2 , stimulus contrasts were fixed and identical for all stimuli , making parameters σ and α unnecessary , but the input orientations θj were now random variables controlled by the proportionality parameter s; the model therefore had four free parameters: κ , γ , β and s . | Investigations of visual short-term memory typically involve memorising clearly visible objects with elementary features , such as monochromatic disks or oriented bars . Results of such studies indicate that memory is allocated like a limited resource , i . e . shared out between objects . However , in daily life we are often confronted with visual features that are difficult to make out , like when an object is in shadow , or poorly-defined , like the color of a variegated leaf . Here we asked whether these kinds of features occupy as much memory resource as simple highly-visible objects . Our results demonstrate that reducing the sensory strength of a stimulus makes the quality of recall worse , but also takes up less resource so other objects can be remembered more precisely . Increasing the variability within a stimulus worsens recall , but has no effect on how other objects are remembered . These findings can be explained by considering how visual information is stored in populations of neurons: only the manipulation of sensory strength changes the amount of spiking activity dedicated to a stimulus . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"engineering",
"and",
"technology",
"signal",
"processing",
"social",
"sciences",
"neuroscience",
"learning",
"and",
"memory",
"noise",
"reduction",
"cognitive",
"neuroscience",
"cognition",
"computational",
"neuroscience",
"memory",
"vision",
"neuronal",
"tuning",
"codin... | 2018 | Internal but not external noise frees working memory resources |
Plant viruses move through plasmodesmata to infect new cells . The plant endoplasmic reticulum ( ER ) is interconnected among cells via the ER desmotubule in the plasmodesma across the cell wall , forming a continuous ER network throughout the entire plant . This ER continuity is unique to plants and has been postulated to serve as a platform for the intercellular trafficking of macromolecules . In the present study , the contribution of the plant ER membrane transport system to the intercellular trafficking of the NSm movement protein and Tomato spotted wilt tospovirus ( TSWV ) is investigated . We showed that TSWV NSm is physically associated with the ER membrane in Nicotiana benthamiana plants . An NSm-GFP fusion protein transiently expressed in single leaf cells was trafficked into neighboring cells . Mutations in NSm that impaired its association with the ER or caused its mis-localization to other subcellular sites inhibited cell-to-cell trafficking . Pharmacological disruption of the ER network severely inhibited NSm-GFP trafficking but not GFP diffusion . In the Arabidopsis thaliana mutant rhd3 with an impaired ER network , NSm-GFP trafficking was significantly reduced , whereas GFP diffusion was not affected . We also showed that the ER-to-Golgi secretion pathway and the cytoskeleton transport systems were not involved in the intercellular trafficking of TSWV NSm . Importantly , TSWV cell-to-cell spread was delayed in the ER-defective rhd3 mutant , and this reduced viral infection was not due to reduced replication . On the basis of robust biochemical , cellular and genetic analysis , we established that the ER membrane transport system serves as an important direct route for intercellular trafficking of NSm and TSWV .
Plasmodesma-mediated macromolecular trafficking plays important roles in plant growth and development [1–3] and in plant–pathogen interactions [4–6] . Structurally , a plasmodesma is composed of the plasma membrane with a central , modified appressed endoplasmic reticulum ( ER ) , the desmotubule [7] . Besides the long-established cell-to-cell transport of small molecules via plasmodesmata , macromolecules such as proteins and RNAs have been shown in the last two decades to traffic between cells through plasmodesmata ( PD ) . Such macromolecular trafficking is crucial for viral infection [4–6] , plant defense [8 , 9] , and developmental regulation [1–3] . Plant viruses need to move within and between cells to establish systemic infection . To accomplish this task , the plant virus encodes a movement protein ( MP ) to facilitate intracellular trafficking of the viral genomes from the replication site to PD and to assist the spread of the viral replication complexes or viral particles between plant cells through PD [5 , 6 , 10–13] . Plant viruses not only utilize viral-encoded MPs or other viral components for viral intra- and intercellular movement , but also co-opt host cell transport machineries for their movement [13–17] . The cytoskeleton and membrane transport systems of cells are important for intracellular movement of vertebrate viruses ( reviewed in [16] ) , essential for organellar trafficking within plant cells [18 , 19] and involved in the intercellular trafficking of macromolecules [20 , 21] . In the case of the best-studied plant virus , Tobacco mosaic virus ( TMV ) , the ER membrane is important for its association with the viral replication complexes ( VRC ) and MP granules , whereas microtubules and microfilaments facilitated their movement on the ER ( reviewed in [22] ) . The ER membrane also serves as an important platform for anchoring several other viral MPs , which are required for intracellular movement and viral spread [23–27] . The ER-to-Golgi secretory pathway is further involved in PD targeting and intercellular trafficking of several viruses [28–33] . Microfilaments and different myosin motors also participate in the intra- or intercellular movement of diverse MPs or viruses [28–30 , 34–40] . In addition , the endocytic pathway has also been shown to influence the movement of three viruses [41–43] . We used Tomato spotted wilt topspovirus ( TSWV ) as a model to study the mechanism of intercellular movement of tospoviruses and multipartite negative-strand RNA plant viruses . TSWV , the type member of Tospovirus which is the only genus containing plant-infecting negative-strand RNA viruses in the family Bunyaviridae [44–46] , causes severe diseases in many economic important crops and is listed as one of the most devastating plant viruses worldwide [47] . Its nonstructural protein NSm is not encoded by any of the animal-infecting members of Bunyaviridae and is thought to be the result of evolutionary adaptation of tospoviruses to infect plants [48] . It has the typical characteristics of a plant viral MP , including a short period of expression early in systemic infection , association with nucleocapsid aggregates in the cytoplasm [49] , localization to PD [50] , presence in the cell wall and membrane fractions of cells [49–51] , induction of tubule structures [52 , 53] , increasing the size exclusion limit of PD [50 , 54] , binding of nucleic acids [55] , complementation of cell-to-cell and long-distance movement of a movement-deficient TMV [51 , 53] or Cucumber mosaic virus ( CMV ) vector [56] by heterologous expression of NSm and interactions with two host trafficking proteins [55 , 57] . Recently , the biochemical properties of TSWV and other tospoviruses NSm related to membrane association have been characterized [51 , 58] . Although these findings have partially elucidated the requirements and mechanism for TSWV movement , little is known about the involvement of host cell transport systems for intercellular movement of the NSm MP and TSWV . The plant ER is a unique structure that is interconnected among cells via the ER desmotubules in the plasmodesma across the cell wall , forming a continuous ER network throughout the entire plant [18 , 59] . In the present study , using in vitro and in vivo systems to characterize membrane association properties of movement proteins , we revealed that TSWV NSm was physically and tightly associated with the ER membrane and trafficked from cell to cell . Taking advantage of these features of TSWV NSm , we investigated the contribution of the plant ER membrane transport system to the intercellular movement of NSm MP and TSWV . With robust biochemical , cellullar and genetic evidence , we demonstrated that the plant ER membrane transport system serves as an important direct route for intercellular trafficking of NSm and TSWV . Our findings have important new implications for mechanistic studies of the intercellular trafficking of tospoviruses and other multipartite negative-strand RNA plant viruses .
Because several studies have reported that TSWV NSm is associated with membranes [49 , 51 , 58] , we used various transmembrane ( TM ) predicting tools to determine the sequence in TSWV NSm that might enable insertion into or association with membranes . The predicted outcome ( S1 Table ) varied according to the methods used . MPEx [60] and DAS-Tmfilter [61] predicted two TM regions , whereas TMpred , and TopPred identified only one ( S1 Table and S1 Fig ) . TMHMM and ΔG prediction algorithms failed to predict any membrane-spanning domain but identified two hydrophobic regions that encompassed residues 127–158 , 161–189 ( TMHMM [62] ) and 127–153 and 163–192 ( ΔG Prediction [63] ) ( S1 Table and S1 Fig ) . To address whether NSm can insert into biological membranes , the two hydrophobic regions identified by the ΔG Prediction was tested using an in vitro experimental system that accurately reports the integration of TM helices into microsomal membranes and has been proved to characterize the membrane-spanning capacity of several plant viral MPs [64–67] . This system uses ER-derived microsomal membranes and provides a sensitive way to detect the insertion or translocation of hydrophobic regions through the Sec translocon [68] . The system is based on the cotranslational glycosylation performed by the oligosaccharyltransferase ( OST ) enzyme . OST adds sugar residues to consensus sequences after the protein emerges from the translocon channel . The glycosylation of a protein region translated in vitro in the presence of microsomal membranes therefore indicates the exposure of this region to the OST active site on the luminal side of the ER membrane . In our experimental assay ( Fig 1A ) , a hydrophobic segment to be assayed ( HR-tested ) replaces the second TM segment from the integral membrane protein Lep . The glycosylation acceptor site ( G2 ) located at the beginning of the P2 domain will be modified only if the HR-tested segment inserts into the membrane , while the G1 site , embedded in an extended N-terminal sequence is always glycosylated [69 , 70] . We found that HR1 and HR2 of NSm inserted 27 . 04 ± 4 . 53% and 14 . 24 ± 3 . 99% , respectively , of the molecules into the membrane . The nature of the cytosolic/luminal domains was further examined by proteinase K ( PK ) digestions . Treatment with PK degrades domains of membrane proteins that protrude into the cytosol , but membrane-embedded or luminally exposed domains are protected . The addition of PK to the Lep-derived translation mixtures ( Fig 1A , lanes 6 and 9 ) generated a residual band originating from the HR2-containing construct , which corresponded to the protected , glycosylated HR1-P2 fragment . These results suggest that NSm hydrophobic regions insert more efficiently than the corresponding regions of TMV and PNRSV MPs [65 , 66] , but not as well as the membrane-spanning capacity required for TM disposition that has been observed for plant viruses with several small integral MPs [64 , 67] . The microsomal in vitro system closely mimics the conditions of in vivo membrane protein assembly . However , HR-tested sequences are not analyzed in their native context . To further characterize the membrane-association of full-length TSWV NSm protein , we prepared subcellular fractions ( 30 , 000 × g pellet [P30] , 30 , 000 × g supernatant [S30] , 100 , 000 × g pellet [P100] and 100 , 000 × g supernatant [S100] ) from N . benthamiana plants that were infected with TSWV or that transiently expressed the NSm protein . The ER luminal marker ( luminal binding protein , BiP ) was localized in both the microsomal and soluble fractions ( Fig 1B ) . The vacuolar H-ATPase ( V-H-ATPase ) subunit E was present primarily in the microsomal P30 fraction , whereas soluble phosphoenolpyruvate carboxylase ( PEPC ) was found exclusively in the soluble fractions ( Fig 1B ) . Immunoblotting showed that NSm was mainly present in the P30 membrane fraction ( Fig 1B ) . These results agreed with earlier data suggesting that TSWV NSm is a membrane-associated protein [49 , 51 , 58] . Alkaline extraction ( 0 . 1 M Na2CO3 ) , salt ( 1 M KCl ) and 4 M urea treatments have been used to distinguish between peripheral and integral membrane proteins [65 , 71] . After these chemical treatments , the TSWV NSm MP was recovered from the P30 fractions ( Fig 1C ) upon centrifugation of both TSWV-infected or transiently expressing NSm plant extracts , indicating that NSm protein is tightly associated with cellular membranes . Next , we washed the membranes with 7 M urea , a treatment that should release all polypeptides from the membrane , except for truly integral membrane proteins [65] . As shown in Fig 1D , 46 . 1% of NSm protein was detected in the soluble fraction . In control experiments , 11 . 8% of the small integral MPs TGB2 and 26 . 6% of the TGB3 of Potato virus X ( PVX ) [27 , 72 , 73] were detected . These results suggested that the amount of protein that was released from the membrane fraction for NSm by 7 M urea was lower than for peripheral membrane proteins [65] , but more than for truly integral membrane proteins ( Fig 1D ) . To further investigate whether NSm is physically associated with membranes , we analyzed the membrane association capacity of NSm using a Triton X-114 partitioning assay , in which aqueous and organic phases are formed , and integral membrane proteins should be partitioned into the organic phase [74] . We treated the pellet fraction ( P30 ) from TSWV-infected or NSm-transient-expressing plant materials with the non-ionic detergent Triton X-114 and as the control used TMV MP , which was recently shown to be a peripheral membrane protein [66] . After the Triton X-114 treatment , most of the TMV MP separated into the aqueous phase , as expected for a peripheral protein , whereas NSm from both types of plant samples remained preferentially in the organic phase ( Fig 1E ) . Thus , Triton X-114 partition results clearly showed that NSm is physically associated with membranes . Considering all the membrane-association data together , we conclude that TSWV NSm is neither a peripheral membrane protein nor a canonical integral membrane protein; instead , the protein is strongly and physically bound to cellular membranes , but its hydrophobic regions probably do not span the lipid bilayer . Previous studies have shown that NSm is localized in the cytoplasm and PD [50 , 58] . To identify the specific membranous structures in which NSm is localized in living cells , we transiently expressed recombinant NSm-YFP ( yellow fluorescent protein ) in leaf epidermal cells of N . benthamiana by agroinfiltration . Using confocal laser scanning microscopy ( CLSM ) , we detected NSm-YFP in a structure that was reminiscent of the ER network ( Fig 2A ) . We also detected NSm-YFP near the cell walls in a punctate pattern reminiscent of PD localization ( Fig 2D ) . Co-expression of NSm-YFP with the cortical ER marker mCherry-HDEL confirmed NSm localization to the ER ( Fig 2A–2C ) . We also checked the localization of NSm-YFP for Golgi bodies ( using marker Man49-mCherry ) , nucleus ( H2B-mRFP ) and chloroplasts after agroinfiltration of N . benthamiana leaves . CLMS showed that NSm-YFP did not co-localize with Golgi bodies ( S2A–S2C Fig ) , nucleus ( S2D–S2F Fig ) or chloroplasts ( S2G–S2I Fig ) . To verify ER localization of NSm independently , we used sucrose gradient fractionation to analyze membrane association patterns of NSm with materials derived from transiently expressed NSm in N . benthamiana . To verify the association of NSm with the ER , sucrose gradients fractions with or without MgCl2 were prepared from N . benthamiana leaves transiently expressing NSm . The ER marker BiP , Golgi marker Arf1 and cytoplasmic soluble protein marker PEPC in the fractions were detected by Western blot . In the absence of MgCl2 , ribosomes become dissociated from the ER , causing a shift in the migration of the ER in the gradient [75] . If the protein is associated with the ER , it will have the same shift as the ER marker BiP in the gradient . Immunoblots of the gradient fractions in the presence or absence of MgCl2 revealed that TSWV NSm protein shifted the same as the ER marker BiP; the Golgi marker ( Arf1 ) and the cytosol protein ( PEPC ) did not shift as the ER did ( Fig 2J ) , These results thus provide additional evidence for ER localization of NSm . To verify whether the subcellular localization of NSm-YFP in the cell walls represents localization to PD , NSm-YFP and TMV MP , which has been well established to have a PD localization function [76] , with red fluorescent protein ( RFP ) fused at its C-terminus , were co-expressed in N . benthamiana leaf epidermal cells . The merged images of NSm-YFP and TMV MP-mRFP localization patterns demonstrated that NSm-YFP localized with the PD as did the TMV MP-mRFP marker ( Fig 2D–2F ) . To further validate the PD localization of NSm , we plasmolyzed the cells to separate the cytoplasmic ER from the PD membranes . As shown in Fig 2G–2I , NSm remained on the PD after the cytoplasmic ER was separated from the cell wall by plasmolysis . Collectively , these results established that NSm is physically associated with ER membranes and recruited to the PD . Thus , we next tested for its potential role in cell-to-cell trafficking . To test whether NSm functions in cell-to-cell trafficking , we bombarded epidermal cells on intact leaves of N . benthamiana plants with a DNA construct that expresses an NSm-GFP fusion protein . As a control , we bombarded separate leaves with a DNA construct expressing a GFP-GFP fusion protein . As shown in Fig 3B and Table 1 , GFP-GFP ( 54 . 0 kDa ) was restricted to single cells in a total of 53 fluorescent loci expressing the fusion protein at 22 h post-bombardment . At the same time , NSm-GFP ( 60 . 9 kDa ) was trafficked into neighboring cells in 58 of the 110 fluorescent foci expressing the fusion protein ( Fig 3C and Table 1 ) . We also analyzed cell-to-cell movement of NSm-GFP and GFP-GFP over 0 h to 48 h post bombardment in epidermal cells of N . benthamiana . As shown in S2 Table , starting at 9–10 h post bombardment , cell-to-cell movement of NSm-GFP was detected with 16 . 3% of the foci expressing the fusion protein . From 10 to 21 h post bombardment , NSm-GFP continued to move from the initial site , reaching 34 . 5% of the foci at 14–15 h , 40 . 4% at 19–20 h and 41 . 9% at 26–27 h . At 48 h post bombardment , the cell-to-cell movement of NSm-GFP had decreased to 27 . 4% of the foci ( S2 Table ) . Parallel observations showed that GFP-GFP remained in the initial cell . To test whether the addition of NSm can facilitate the cytoplasmic soluble GFP-GFP protein to traffick into neighboring cells , leaves of N . benthamiana were agroinfiltrated to express NSm or an empty vector , and 12 h later , bombarded the agroinfiltrated leaves with GFP-GFP . After an additional 24 h , we checked bombarded foci for cell-to-cell movement of GFP-GFP . As shown in S3 Table , GFP-GFP did not move from the initially bombarded cell in the presence of NSm . To further verify the cell-to-cell movement of NSm , a plant binary vector was constructed to harbor two gene cassettes to express NSm-GFP and mCherry-HDEL simultaneously ( S3A Fig ) . The agrobacterium containing the plant binary construct was infiltrated at OD260 = 0 . 004 to express it in a single epidermal cell of N . benthamiana . As shown in S3B Fig ( upper panel ) , the red fluorescence signals from mCherry-HDEL were mostly observed in one to two cells ( primary expression cell ) . Green fluorescence signals from NSm-GFP overlapped with red signals in the primary expression cell , whereas green signals were also observed in cells surrounding the primary cell ( secondary expression cells ) . As a control , we also made a plant binary construct to co-express GFP-GFP and mCherry-HDEL ( S3A Fig ) . When agrobacterium containing this construct was delivered into plant cells using the same strategy for delivering NSm-GFP , both the green fluorescence signals from GFP-GFP and the red fluorescence signals from mCherry-HDEL remained in the primary expression cell; no GFP-only cells were found surrounding the primary expression cell ( lower panel in S3B Fig ) . Thus , these data provide strong evidence that NSm-GFP indeed moves from cell to cell . As noted already , the plant ER is interconnected among the cells via the desmotubule of the PD , forming a continuous ER network throughout the plant [18 , 59] . Because NSm is physically associated with the ER , it may move from cell to cell through the PD along the continuous ER network among cells . If so , the trafficked protein is expected to reside in the ER in neighboring cells . To test for this case , we expressed mCherry-HDEL in epidermal leaf cells in N . benthamiana by agroinfiltration . After 12 h post agroinfiltration , we bombarded N . benthamiana leaf cells that labeled the ER network with the DNA construct expressing NSm-GFP . After additional 22 h , we checked the cell-to-cell movement of NSm-GFP along the ER membrane . We analyzed a leaf region containing three layers of cells: the initially bombarded cells showing strong fluorescence ( Cell 1 ) , the immediately adjacent cells with weaker fluorescence ( Cell 2 ) , and the third layer of cells emitting weakest fluorescence ( Cell 3 ) ( Fig 3A , 3D and S4A Fig ) . Merging of the GFP and mCherry channels and a fluorescence spectra analysis showed that NSm-GFP co-localized with mCherry-HDEL in all cells ( Fig 3E–3Q and S4 Fig ) . This pattern is consistent with the hypothesis that NSm-GFP trafficks along the ER network between cells . The following experiments tested this hypothesis directly . If NSm trafficked along the ER membrane network from cell to cell , disruption of its physical association with ER should inhibit trafficking . We first tested whether mutating the predicted hydrophobic region ( residues 127 to 192 , which includes HR1 to HR2 ) of NSm ( Fig 4A ) would affect its physical association with the membrane . We designed two mutants with a respective aspartate substitution at amino acids 133–135 ( IVI ) or 177–179 ( FVF ) , roughly in the middle of HR1 to HR2 ( Fig 4A ) . N . benthamiana leaves were agroinfiltrated with these two mutants , then the membrane fractions prepared from these leaves were treated with Triton X-114 . As shown in Fig 4B , after the Triton X-114 treatment , approximately 17% of NSm133-135D and 38% of NSm177-179D mutant proteins partitioned into the aqueous phase , whereas 100% of the wild-type NSm remained in the organic phase , suggesting that the physical membrane integration capacity of the two mutants was altered to a certain degree . To test the cell-to-cell trafficking function of these two mutants , we examined single cells of N . benthamiana leaves bombarded with one or the other mutant . Both mutants remained in single cells ( Fig 4C–4E and Table 1 ) . Thus , full membrane integration is critical for NSm to traffic from cell to cell , providing another piece of critical experimental evidence to support the hypothesis that NSm traffics intercellularly along the ER network . The following experiments further tested this hypothesis using complementary approaches . Our confocal colocalization analysis showed that NSm formed punctuate structures on the ER ( Fig 2A–2C ) . We reasoned that NSm might contain specific signals for sorting it to the ER . Because di-acid or di-hydrophobic motifs have been shown to play critical roles in ER sorting [77 , 78] , we generated two NSm mutants with alanine substitutions at di-hydrophobic and di-acid amino acids 4–5 ( FF ) and 230–232 ( DKD ) . We agroinfiltrated N . benthamiana leaves to transiently express these two mutants , and then analyzed their subcellular localization by confocal microscopy . Colocalization analysis showed that mutant NSm4A/5A formed a smooth structure that aligned well with the ER membrane , but no longer formed punctate structures ( Fig 5A–5C ) , indicating that the mutations blocked NSm sorting to the cortical ER . Mutant NSm230A/232A was present as punctate structures in the cells , but was not localized to the ER ( Fig 5D–5F ) , indicating that it was mis-sorted to other subcellular localizations . We also used sucrose gradient fractionation with and without MgCl2 to identify the cellular fraction that contained the mutant NSm4A/5A and NSm230A/232A . The immunoblot of gradient fraction showed that NSm4A/5A was distributed mainly in fractions 6–8 with MgCl2 and 5–7 without MgCl2 ( S5B Fig ) , whereas NSm230A/232A was distributed in fractions 2–4 whether MgCl2 was present or not ( S5C Fig ) . As shown in above , in the absence of MgCl2 , ribosomes became dissociated from the ER , and the NSmWT had the same shift as the ER marker in the gradient ( Fig 2J ) . However , neither NSm4A/5A or NSm230A/232A had the same shift as the NSmWT ( S5 Fig ) . These results suggest that the subcellular localization of the two NSm mutants differed from the ER location of the NSmWT . Via biolistic bombardment , mutants NSm4A/5A and NSm230A/232A were then expressed separately in single cells of N . benthamiana leaves . Neither NSm4A/5A nor NSm230A/232A could move cell to cell ( Fig 5G–5I and Table 1 ) . Thus , sorting of NSm to particular sites on the ER is necessary for its intercellular trafficking . If NSm trafficks along the ER membranes , disruption of the ER network should negatively affect this trafficking . We thus tested the effect of brefeldin A ( BFA ) , a pharmacological drug that at high concentrations can disrupt the integrity of the ER network [79 , 80] , on the structural integrity of the ER network in N . benthamiana leaves expressing mCherry-HDEL . At 3 h post-treatment with 20 μg/mL BFA , the sheet structure of the ER clearly changed ( S6A and S6B Fig ) . Within 6–12 h post-treatment , the ER network was severely disrupted ( S6C–S6F Fig ) . To test if the disruption of ER membrane by BFA treatment affects cell-to-cell movement of NSm , the bombarded leaf tissues of N . benthamiana transiently expressing NSm-GFP were treated with 20 μg/mL of BFA or DMSO ( as a control ) 6 h post bombardment for 12 h , and checked for cell-to-cell trafficking of NSm-GFP by confocal microscopy . As shown in Table 2 , 37 . 9% of the foci showed multicellular fluorescence from the NSm-GFP fusion protein in DMSO-treated control leaf tissues , in contract to a much lower percentage ( 10 . 9% ) of such foci observed in BFA-treated leaves . Thus , BFA treatment significantly inhibited cell-to-cell trafficking of NSm-GFP . To test whether this inhibition is the result of a general disruption of transport through PD , we assayed cell-to-cell trafficking of GFP , which moves freely between leaf epidermal cells [81 , 82] . As shown in Table 2 , cell-to-cell trafficking of GFP , which most likely occurred through the microchannels formed between the plasma membrane and the ER with PD , was not affected by BFA treatment in comparison with the DMSO treatment ( Table 2 ) . Thus , BFA treatment did not affect general transport through PD , but specifically disrupted cell-to-cell trafficking of NSm . These combined results provide pharmacological evidence that NSm trafficked along the ER membrane , rather than via the cytoplasmic channels among cells through PD . Because BFA at low concentrations can block transport from the ER to the Golgi apparatus [83] , we used 2 . 5 μg/mL BFA to examine whether the ER-to-Golgi secretion pathway is involved in the cell-to-cell movement of NSm-GFP . Although sufficient to cause the Golgi marker Man49-GFP to retreat back into the ER ( S7A–S7F Fig ) , 2 . 5 μg/mL BFA had no effect on cell-to-cell movement of NSm-GFP after biolistic bombardment ( S4 Table ) , strongly suggesting that NSm does not enter the ER-to-Golgi early secretion pathway . Next , we investigated the contribution of cytoskeleton transport components , microfilaments , myosin motors and microtubules , to the intercellular movement of TSWV NSm . To examine the involvement of microfilaments , we treated N . benthamiana leaves expressing an actin marker with 5 μM latrunculin B ( LatB ) , which disrupted the actin filaments within 5 h ( S8A–S8F Fig ) . However , treatments with LatB longer than 6 h also disrupted the ER membrane structure ( S8G–S8I Fig ) . Thus , we chose another strategy to conduct this experiment . In the time course of cell-to-cell movement established for NSm-GFP , the NSm-GFP moved intercellularly 16 . 3% of the initial foci expressing fusion proteins by 10 h post bombardment and 34 . 5% of the foci by 15 h , indicating approximately 18% of the foci underwent cell-to-cell movement between 10 and 15 h . This time frame is thus appropriate for testing the effect of LatB . By adding LatB at 10 h , we could observe the effect of the drug on NSm intercellular movement at 15 h before LatB affects ER structure . As shown in S4 Table , cell-to-cell trafficking of NSm-GFP in the LatB-treated leaf cells did not differ significantly from that in the DMSO-treated control , suggesting that actin filaments are not involved in NSm movement . We then investigated the contribution of myosin motors to NSm movement by expressing the Golgi marker in N . benthamiana leaves and treated them with BDM , a myosin inhibitor [84] . We found that 100 mM 2 , 3-butanedione monoxime ( BDM ) was sufficient to inhibit intracellular movement of the Golgi bodies ( S1 Movie ) within 6 h without any significant disturbance of ER structure ( S9A–S9F Fig ) as seen with confocal microscopy . We thus treated N . benthamiana leaves bombarded to express NSm-GFP for 6 h with 100 mM of BDM or PBS buffer ( as a control ) . At 5 h post treatment , we checked for cell-to-cell trafficking of NSm-GFP . As shown in S4 Table , there was no significant difference between the BDM-treated and PBS-treated leaf cells in cell-to-cell trafficking of NSm-GFP , suggesting that myosin motors are not involved in the intercellular movement of NSm . Finally , we investigated the contribution of microtubules to NSm movement . We previously reported that 20 μM oryzalin efficiently disrupts the microtubule filaments [85] , and it had no effect on ER membrane structure within 6 h ( S9G–S9L Fig ) . We then delivered NSm-GFP into N . benthamiana leaf cells via biolistic bombardment . After 10 h , we treated the bombarded leaves with 20 μM of oryzalin or DMSO . When we checked for cell-to-cell trafficking of NSm-GFP 5 h later , we did not observe any significant differences after the oryzalin and control treatments in the cell-to-cell movement of NSm-GFP ( S4 Table ) , suggesting that microtubules are not involved in NSm movement . Taken together , these data suggest that the ER-to-Golgi secretion pathway and cytoskeleton transport systems are not involved in the intercellular movement of NSm . We finally tested the ER-based NSm cell-to-cell trafficking hypothesis and its biological significance by a genetic approach . The ER network is formed by homotypic fusion of membrane tubules . In mammalian cells , a class of dynamin-like , membrane-bound GTPases called atlastins are involved in the generation of the tubular ER network [86 , 87] . In plants , RHD3 ( ROOT HAIR DEFECTIVE 3 ) , an analogue of the mammalian atlastin , mediates the generation of the tubular ER network . Knockout of the RHD3 gene leads to a nonbranched ER network in Arabidopsis thaliana [88 , 89] . Using the Arabidopsis rhd3 mutant to investigate whether the altered ER network structure affects cell-to-cell trafficking of NSm , we first confirmed the altered morphology of the ER network in the rhd3-8 mutant line . A nonbranched ER network labeled by mCherry-HDEL was clearly observed in the rhd3-8 mutant but not in the wild-type ( WT ) Col-0 ( Fig 6A and 6B ) . We then used biolistic bombardment to produce NSm-GFP in leaves of the WT and the rhd3-8 mutant plants . As shown in Fig 6C , 6D and Table 3 , cell-to-cell trafficking of NSm was significantly reduced in the rhd3-8 mutant , compared with that in the WT plants . Specifically , 96 of the 178 ( 53 . 9% ) fluorescent foci showed NSm trafficking in WT plants , whereas only 60 of the 171 ( 35 . 1% ) loci did in the rhd3-8 mutant at 21 h post-bombardment . Furthermore , NSm-GFP moved to 4–6 cells in WT plants in contrast to only 2–4 cells in the rhd3-8 mutant plants ( Table 3 ) . Importantly , cell-to-cell diffusion of GFP did not differ significantly between WT and rhd3-8 plants ( Table 3 ) , indicating that the altered ER structure did not impact general transport through PD microchannels . Next , we analyzed TSWV infection in WT and rhd3 mutant plants by mechanically inoculating leaves of these plants with equal amounts of TSWV virions followed by monitoring disease development and viral accumulation over time . As shown in Fig 7A–7C and S5 Table , TSWV disease development was much slower in the infected rhd3-8 mutant plants than in the infected WT plants . In contrast to the 100% infection and typical cell death symptoms at ∼15 days post-inoculation in the WT plants , disease symptoms were delayed by 3–5 days and attenuated in the rhd3 mutant plants ( Fig 7A–7C and S5 Table ) . Immunoblots of viral accumulation in systemically infected leaves , probed with a monoclonal antibody against the TSWV nucleocapsid , at 15 days postinoculation showed that TSWV accumulation was significantly lower in the rhd3-8 mutant plants than in the WT plants ( Fig 7D ) . To determine whether the reduced virus infection resulted from delayed cell-to-cell movement or from reduced viral replication or both , protoplasts were isolated from WT and rhd3-8 mutant and transfected with purified particles of TSWV . At 24 h post transfection , protoplasts were harvested , and the accumulation of viral genomic RNAs or proteins of TSWV was examined by quantitative real-time RT-PCR or immunoblot . As shown in S10A Fig , replication of the TSWV M ( left panel ) or S ( right panel ) RNA segment in rhd3-8 was comparable to that in WT . The expression level of TSWV nonstructural protein NSm ( right upper panel ) or NSs ( right middle panel ) , which was not present in the purified viral particles , in the rhd3-8 mutant was also similar to that in WT ( S10B Fig ) , suggesting that the reduced virus infection in rhd3-8 is due to delayed cell-to-cell movement of NSm . Together , these results provided genetic evidence to support the hypothesis that TSWV NSm employs the ER membrane network as a route for cell-to-cell trafficking . Furthermore , the results indicate that this route is important for viral infection .
Previous studies have demonstrated that different plant viruses may use different host cell transport machineries to move from one cell to another through PD [13–17 , 22] . In the present study , we confirmed that TSWV NSm is physically associated with the ER membrane , revealed that the NSm moves between cells through the PD , and evaluated the contribution of the ER membrane transport system to the intercellular movement of NSm and TSWV . We have obtained comprehensive biochemical , cellular and genetic evidence that the ER membrane transport system is critical for cell-to-cell trafficking of NSm and the virus . A recent study suggests that NSm from four tospoviruses are peripherally associated with membranes [58] , based on results from a treatment with 7 M urea . We observed that TSWV NSm is still tightly associated with the membrane after treatment with 4 M urea , consistent with a previous suggestion that GNRV NSm is also tightly associated with or even integrated into the membrane [90] . Although 7 M urea is a strong treatment that should release all peripheral proteins from the membrane [65] , it only extracted half of the NSm molecules from the membrane fraction . Our in vitro membrane insertion experiment clearly showed that TSWV NSm HRs can integrate to a certain extent into the membrane . Triton X-114 partition analysis further showed that NSm from transient expression or virus infection is physically and tightly associated with membrane . These results suggest that TSWV NSm is not a peripheral membrane protein as recently reported [58] . However , the strength of the membrane association with NSm is not as strong as for a canonical integral membrane protein that span the membrane . The 7 M urea treatment extracted more NSm protein molecules than it did for PVX TGB2 and TGB3; the insertion capability of NSm was not as good as that of previously reported , membrane-spanning proteins [64 , 67 , 69] . All these data suggest that the membrane insertion capability of NSm lies between the peripheral membrane protein and the integral membrane protein . The new conclusion that could fit all previous data and our results in this study is that NSm strongly and physically associates with the membrane , but probably does not span the lipid bilayer . Thus , tospovirus NSm may represent a new subcategory of movement protein with a distinct membrane association property in 30K superfamly . NSm-GFP ( 60 . 9 kDa ) expressed in a single cell of N . benthamiana moved efficiently from one cell to another , which is supported by the fact that TSWV NSm can increase the size-exclusion limit ( SEL ) of the PD [50 , 54] . However , the dilated SEL of the PD itself is not sufficient to support cell-to-cell movement of cytoplasmic proteins such as GFP-GFP ( 54 . 0 kDa ) in the presence of NSm . As we mentioned earlier , NSm is physically and tightly associated with the ER membrane ( Figs 1 and 2 ) . Mutations in NSm that disrupted its association with the ER membrane or caused mis-sorting to other subcellular localization inhibited cell-to-cell trafficking , suggesting that targeting onto the ER is a critical step for cell-to-cell movement . We observed that NSm forms inclusion bodies on the ER . NSm predominantly targets the ER where it likely accumulates within the plant cell . The NSm inclusions were easily detected when NSm was transiently expressed after agroinfiltration . However , the NSm inclusion bodies were not observed in cells from which NSm trafficked to adjacent cells in the biolistic bombardment assay , suggesting that the inclusion bodies probably are due to NSm accumulation on the ER . Although we do not yet know whether NSm inclusion bodies have any other biological function for TSWV , these inclusion bodies are a reliable indicator of the ER targeting of NSm . The NSm4A/5A mutant that did not induce ER inclusions was not in the same sucrose gradient fraction as the ER , and it was defective in cell-to-cell trafficking , further supporting that targeting the ER is an important step for its intercellular movement . The plant ER is interconnected among cells via the ER desmotubule in the plasmodesma of the cell wall [18 , 59] . Pharmacological disruption or a genetic defect in the ER network inhibited NSm-GFP trafficking but not GFP diffusion , which occurs through the cytoplasmic channels . The defect in the ER network strongly delayed cell-to-cell spread of TSWV . These results demonstrated that an intact ER membrane transport system is critical for intercellular movement of viral MP and TSWV . The plasmolysis assay showed that NSm remained on the PD after the cytoplasmic ER was separated from the cell wall ( Fig 2G–2I ) . The PD localization of NSm and its physical association with the ER strongly suggest that NSm and TSWV may move between cells through PD via the desmotubule ER . Compared with the discrete ER confined within each cell in animals , the plant ER forms a continuous membrane network throughout the entire plant [18 , 59] . Such a unique ER membrane network structure in plant provides an effective continuous transport route to deliver NSm and TSWV to neighboring cells or to remote growing tissues . Although molecular genetics of the animal-infecting members of the Bunyaviridae have been facilitated by the rescue of infectious cDNA clones , a reverse genetics system remains elusive for all members of the multipartite negative-strand plant RNA viruses . Despite discovering the NSm mutant that is defective in its ER association or ER sorting , without infectious cDNA clones of TSWV , it is difficult for us to use this system to evaluate the contribution of the ER-membrane association of NSm to the intercellular trafficking of the virus . We have also tested the BFA effects on the intercellular movement of TSWV; however , a high concentration of BFA caused severe cell death in infiltrated leaves of N . benthamiana or N . tobacum cv . Samsun after 30 h of treatment . Chemical treatments have been widely used in the studies of virus movement [28 , 37 , 85] . But as we mentioned in the results , prolonged incubation with chemicals has nonspecific effects on the plant , especially when drugs must be used for days to see any effects on virus infection and thus greatly interferes with the interpretation of results . Thus , with this system , evaluating the contribution of the ER membrane network to intercellular movement of virus is also difficult . We were finally able to find an rhd3 mutant that is defective in the generation of the tubular ER network . This mutant with nonbranched ER gives a clean background and consistent outcome and provides us a powerful genetic system to investigate the contribution of the ER membrane network to the intercellular movement of TSWV . By using this system , we were able to identify that virus infection was strongly delayed in the rhd3 mutant . The protoplast assay further revealed that the reduced viral infection was not due to defective replication . Thus , we established that the ER membrane transport system is important for TSWV to establish infection of the plant . Although NSm accumulated on the ER membrane , we found that the ER-to-Golgi secretion pathway is not involved in NSm movement . Subcellular localization showed that NSm was not exported onto the Golgi body . Treatment with a low concentration of BFA that causes the redistribution of a Golgi marker back to the ER did not affect the intercellular movement of NSm . Although cytoskeleton transport systems are important for the intercellular movement of many MPs and viruses [28 , 30 , 36–40] . Our pharmaceutical treatment analysis showed that microfilaments , myosin motors and microtubules are not involved in the cell-to-cell movement of TSWV NSm . However , we showed previously that TSWV nucleocapsids , the major component of the VRC , move intracellularly along the ER and actin microfilaments [85] . The actin microfilaments and myosin XI-K were found to be responsible for the intracellular movement of the TSWV nucleocapsid and for the virus to infect the plant [85] . Studies of the classical virus TMV have demonstrated that microfilaments and myosin motors are important for intracellular movement of viral VRC [40 , 91] , whereas the ER membrane to which the VRC and MP are anchored have been proposed to play essential roles in the cell-to-cell movement of the MP and virus by diffusing between cells through the PD [92] . In the present study , we demonstrated that the ER membrane transport system is critical for intercellular trafficking of the NSm MP and TSWV . Based on our previous and current findings , a new picture is emerging for TSWV that microfilaments and myosins are involved in the intracellular trafficking of viral VRC , whereas the ER membrane transport system , where both NSm and VRC associate , provides a critical direct route for NSm and the virus to move from cell to cell . Tospovirus NSm is predominantly membrane associated [49 , 51 , 58] . The VRC of Fig mosaic virus , another multipartite negative-strand plant RNA virus , was also found to traffic intracellularly along the ER and actin microfilaments [80]; however , little has been known about the mechanism of intercellular movement of multipartite negative-strand RNA plant viruses . Our findings have important new implications to guide future mechanistic studies on intercellular trafficking of tospoviruses and other multipartite negative-strand plant RNA viruses . In summary , we showed in this study that the ER membrane transport system is important for the intercellular movement of NSm and TSWV . Because NSm localized on both the ER and PD , the virus likely recruits essential host factors to facilitate its cell-to-cell movement via the desmotubule ER . Identifying the host proteins within the ER membrane transport system that interact with the viral proteins and enable virus movement will be the focus of our future investigations .
The NSm gene was amplified from total RNA isolated from TSWV YN-infected tomato by RT-PCR . The ER markers mCherry-HDEL and YFP-HDEL , the Golgi markers Man49-mCherry and Man49-GFP [93] were obtained from the Arabidopsis Biological Resource Center ( ABRC ) . H2B-mRFP was constructed by fusing H2B and mRFP by overlap PCR . All constructs and primers used in this study are listed in S6 Table . For expression of Lep-derived constructs harboring HR from NSm protein , the Lep sequence carried one glycosylation acceptor site in positions 3–5 of an extended sequence of 24 residues previously described [70] . Oligonucleotides encoding the HR1 and HR2 sequences were introduced replacing H2 TM segment from Lep . Tested sequences were constructed using two double-stranded optimized oligonucleotides with 5′ phosphorylated overlapping overhangs at the ends . Pairs of complementary oligonucleotides were first annealed at 85°C for 10 min followed by slow cooling to 30°C , after which the two annealed double-stranded oligonucleotides were mixed , incubated at 65°C for 5 min , cooled slowly to room temperature and ligated into the vector . NSm-derived segment inserts were confirmed by sequencing of plasmid DNA . Six- to eight-week-old plants of N . benthamiana or A . thaliana were used for all transient expression analyses and virus inoculations . Homozygous seeds of the rhd3-8 mutant of A . thaliana ( SALK_025215 ) were a gift from Dr . Junjie Hu ( Nankai University , Tianjin , P . R . China ) . TSWV was maintained in N . rustica plants , and sap from fresh systemically infected leaves were used as inocula . Agrobacterium tumefaciens cells ( GV3101 ) containing various NSm constructs and organelle markers were treated with infiltration buffer ( 10 mM MgCl2 , 10 mM MES , pH 5 . 9 , and 150 μM acetosyringone ) for 3 h at room temperature before infiltration ( OD600 = 0 . 2 ) of the abaxial side of N . benthamiana leaves . All agroinfiltrated or virus-inoculated plants were grown in growth chambers ( model GXZ500D , Jiangnan Motor Factory , Ningbo , P . R . China ) at 25°C with 16 h light/8 h dark for N . benthamiana and 8 h light/16 h dark for A . thaliana . Plasmolysis was carried out by infiltrating leaves with 10% NaCl; cells were immediately examined microscopically for the separation of the plasma membrane from the cell wall . In vitro translation and membrane insertion assays for NSm HRs were done in the presence of reticulocyte lysate , [35S]Met/Cys , and dog pancreas microsomes as described previously [69] . Translation membranes were collected by ultracentrifugation and analyzed by SDS-polyacrylamide gel electrophoresis ( SDS-PAGE ) ; gels were visualized on a Fuji FLA3000 phosphorimager with ImageGauge software . For the proteinase K protection assay , the translation mixture was digested with proteinase K for 40 min on ice; the reaction was stopped by adding 1 mM phenylmethylsulfonyl fluoride ( PMSF ) before SDS-PAGE analysis . Fractionation was done as described by Peremyslov et al . [75] . P30 membrane fractions were treated with lysis buffer ( 20 mM HEPES pH 6 . 8 , 150 mM potassium acetate , 250 mM mannitol , 1 mM MgCl2 ) , 0 . 1 M Na2CO3 , 1 M KCl , 4 M urea , 7 M urea or 1% Triton X-114 as described by Peiró et al . [66] . Chemically treated samples were fractionated into the P30 pellet and S30 supernatant . The fractions were analyzed by immunoblotting . Sucrose gradient fractionation in the presence or the absence of MgCl2 was performed as described by Peremyslov et al . [75] . NSm-expressing leaves were ground in lysis buffer . The P30 pellet , prepared as described above , was loaded on the top of 20 to 60% linear sucrose gradients prepared with lysis buffer . Samples were centrifuged for 16 h at 100 , 000 × g at 4°C , and 14 fractions of 1 . 2 mL each were collected from top to bottom . Each fraction was precipitated with trichloroacetic acid ( TCA ) and analyzed by immunoblotting . Confocal images were captured with an inverted Zeiss LSM 710 CLSM and 20× or 63× water immersion objective lenses . YFP and GFP were excited with 488 nm wavelength and emissions at 497–520 nm captured . mRFP and mCherry were excited with 561 nm wavelength and emissions at 585–615 nm captured . For the visualization of chloroplasts , chlorophylls were excited with 488 nm wavelength and emission at 660–720 nm captured . Images were processed using a Zeiss 710 CLSM and Adobe ( San Jose , CA , USA ) Photoshop . Constructs pRTL2-NSm-GFP , pRTL2-GFP-GFP and pRTL2-GFP were coated with 0 . 6-μm gold micro-carriers . Leaves from N . benthamiana or A . thaliana plants were bombarded using a handheld Helios Gene Gun ( Bio-Rad , Hercules , California , USA ) . Bombarded plants were maintained under the same growth conditions and imaged at various times after bombardment . For BFA treatment , N . benthamiana leaves were bombarded with the constructs expressing NSm-GFP or GFP . After 6 h , the bombarded leaves were infiltrated with a low ( 2 . 5 μg/mL ) or high concentration ( 20 μg/mL ) of BFA ( Biyuntian , Haimen , P . R . China ) . Equivalent dilutions of DMSO were used as controls . After 12 h , cell-to-cell trafficking of NSm-GFP was examined with a CLSM . For LatB , BDM and oryzalin treatments , N . benthamiana leaves were bombarded with the construct expressing NSm-GFP . After 10 h , the bombarded leaves were treated with 5 μM LatB ( Sigma , Shanghai , China ) , 100 mM BDM ( Sigma ) or 20 μM oryzalin ( Sigma ) . Equivalent dilutions of DMSO or PBS buffer were used as controls . After another 5 h , cell-to-cell trafficking of NSm-GFP was examined with a CLSM . TSWV particles were purified as described by Kikkert et al . [94] . TSWV virions were isolated at 4°C from systemically infected N . rustica leaves . The virions were homogenized in sterile double-distilled water and stored at −80°C . Before protoplast inoculation , virions were thawed slowly on ice . Arabidopsis protoplasts were prepared as described by Yoo et al . [95] from rosette leaves of 4-week-old Arabidopsis plants . Then 0 . 5–1 × 106 protoplasts were transfected with 2 μg TSWV virions using polyethylene glycol ( PEG3350 ) according to Kikkert et al . with minor modifications [94] . At 24 h post inoculation , protoplasts were harvested and analyzed by real-time PCR and protein immunoblotting . Total RNA from Arabidopsis protoplasts was extracted using an RNAsimple Total RNA kit ( Tiangen , Beijing , P . R . China ) . First-strand cDNA was synthesized using a PrimeScrip RT reagent kit with gDNA eraser ( Takara , Dalian , P . R . China ) , and the cDNA was then amplified using a Power SYBR Green Master Mix ( Life Technologies , Carlsbad , California , USA ) . Primers ( S6 Table ) specific for the TSWV M segment and S segment were used for quantitative analyses of RNA level of TSWV in protoplasts . The qRT-PCR was performed in an ABI 7500 Real-Time PCR system ( Life Technologies ) . Actin 2 served as an internal control to normalize the RNA levels of target gene between samples using a relative quantification method . Western blotting was carried out as described previously [85] . Protein samples were separated by electrophoresis in 10% SDS-PAGE and transferred onto a PVDF membrane . The antigens on the PVDF membrane were detected with antibodies against TSWV NSm , N , NSs , Arf1 , BiP , PEPC , V-H-ATPase , GFP , or HA , followed by AP-coupled goat anti-mouse IgG ( 1:10 , 000 dilution; Sigma ) and 5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium ( NBT/BCIP ) staining ( Sangon Biotech , Shanghai , China ) . | Plant viruses may use different host cell transport machineries to move from one cell to another through plasmodesmata . The contribution of host cell transport systems to the intercellular movement of multipartite negative-strand RNA plant viruses including tospoviruses is poorly understood . We used Tomato spotted wilt tospovirus ( TSWV ) as a model to understand the mechanism of intercellular movement of tospoviruses . In this study , using in vitro and in vivo systems for characterizing membrane proteins , we identified that the TSWV NSm movement protein was physically associated with the ER membrane . NSm expressed in a single leaf cell was able to move into neighboring cells along the ER membrane network . The ER membrane in plants is a unique structure that runs between neighboring cells via the ER desmotubule of the plasmodesmata and forms a continuous network throughout the plant . Taking advantage of TSWV NSm being tightly associated with ER membrane and trafficked between cells through plasmodesmata , we demonstrated here by robust biochemical , cellullar and genetic evidence that the ER membrane transport system of plants serves as an important route for intercellular trafficking of the NSm movement protein and TSWV . Our findings have important new implications for mechanistic studies on intercellular trafficking of tospoviruses and other multipartite negative-strand RNA plant viruses . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"physiology",
"plant",
"anatomy",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"endoplasmic",
"reticulum",
"cell",
"processes",
"microbiology",
"brassica",
"viruses",
"membrane",
"proteins",
"plant",
"scienc... | 2016 | The ER-Membrane Transport System Is Critical for Intercellular Trafficking of the NSm Movement Protein and Tomato Spotted Wilt Tospovirus |
The putative link between gene expression of brain regions and their neural connectivity patterns is a fundamental question in neuroscience . Here this question is addressed in the first large scale study of a prototypical mammalian rodent brain , using a combination of rat brain regional connectivity data with gene expression of the mouse brain . Remarkably , even though this study uses data from two different rodent species ( due to the data limitations ) , we still find that the connectivity of the majority of brain regions is highly predictable from their gene expression levels–the outgoing ( incoming ) connectivity is successfully predicted for 73% ( 56% ) of brain regions , with an overall fairly marked accuracy level of 0 . 79 ( 0 . 83 ) . Many genes are found to play a part in predicting both the incoming and outgoing connectivity ( 241 out of the 500 top selected genes , p-value<1e-5 ) . Reassuringly , the genes previously known from the literature to be involved in axon guidance do carry significant information about regional brain connectivity . Surveying the genes known to be associated with the pathogenesis of several brain disorders , we find that those associated with schizophrenia , autism and attention deficit disorder are the most highly enriched in the connectivity-related genes identified here . Finally , we find that the profile of functional annotation groups that are associated with regional connectivity in the rodent is significantly correlated with the annotation profile of genes previously found to determine neural connectivity in C . elegans ( Pearson correlation of 0 . 24 , p<1e-6 for the outgoing connections and 0 . 27 , p<1e-5 for the incoming ) . Overall , the association between connectivity and gene expression in a specific extant rodent species' brain is likely to be even stronger than found here , given the limitations of current data .
Genes play a major role in the formation of the nervous system and in its continuous function . They specify neuronal cell types , help destine neurons into defined neural circuits , and provide important cues determining their connectivity [1]–[2] . Inspired by Roger Sperry's classical chemo-affinity hypothesis that states that neuronal wiring takes place by selective attachment guided by specific molecular identifiers , a large array of studies have described various gene families that are involved in axonal guidance and in determining their specific targets ( see [3]–[7] for reviews ) . Another central paradigm has posited that a central driving force in determining synaptic connectivity are activity-dependent mechanisms , by which synapses are formed between neurons whose firing tends to be correlated in a self-organizing Hebbian manner ( see [8]–[9] for reviews ) . A third paradigm has recently emphasized the potential role of random axonal outgrowth and location-dependent competition in establishing connectivity [10] . These paradigms are obviously not mutually exclusive and are likely to concur concomitantly , and quantifying the extent of association between gene expression and connectivity may provide global constraints on their relative contribution . A few recent studies have examined the association between gene expression and connectivity on the neuronal level in the worm C . elegans , by studying the relation between a neuron's gene expression and its connectivity to and from other neurons . C . elegans offers a unique opportunity to perform such an investigation , as it is currently the only model organism for which both a large fraction of its synaptic connectivity and gene expression are known on an individual neuronal level . While [11]–[12] have set to predict the formation of synapses in the worm based on the expression pattern of the pertaining genes [13] , aimed to do so while additionally considering their spatial proximity . Overall , these studies have shown that: ( 1 ) neuronal gene expression does contain significant information about its connectivity , but the predictive power it entails is rather moderate , at least with the current available data , and ( 2 ) it is still possible to use this information to identify genes that potentially play part in determining the neural architecture , on a genome scale . Here we aim to significantly go beyond these earlier studies and to investigate the fundamental relation between gene expression and connectivity in a mammalian brain , and to study it at the level of connectivity between different brain regions . A recent study [14] has used the mouse brain data of the Allen mouse brain atlas ( ABA ) [15]–[16] and the accompanying spatial gene expression correlation map tool to study gene expression patterns within the CA1 field . Multiple observations have been made to suggest that gene expression associations between CA1 regions and other sub-cortical brain regions are indicative of direct or indirect projections to or from distinct spatial domains of the CA1 field . In another study [17] , it was shown that a factorization of the hippocampus volume by the local gene expression levels leads to a spatial grouping that agrees with the known patterns of differential connectivity . Inspired by these studies , we set out here to generalize their scope and examine the possibility of using gene expression signatures to predict regional connectivity in a mammalian brain . Presently , as there is no adequate regional gene expression and connectivity data available for a single mammalian species , we therefore fuse data from two species: brain wiring data for the rat brain and regional gene expression data from the mouse brain , to study their relation in a prototypical rodent brain . The rat connectivity atlas [18] available online ( http://brancusi . usc . edu/bkms/ ) provides connectivity information for the anatomical structures of the rat . The Allen mouse brain atlas ( ABA ) [15]–[16] provides gene expression images for the adult mouse brain . Although gene expression during embryogenesis and development would have ideally been more befitting , this data is still lacking on the large scale . Yet , major components of synapses ( such as synaptic boutons and spines ) are undergoing continuous turnover and are actively maintained during adult life ( e . g . , [19]–[20] ) , which raises the possibility that information on synaptic connectivity may also be manifested in adult gene expression . This , coupled with the success of the earlier studies in the worm in predicting connectivity from adult gene expression [11]–[13] , has motivated us to explore this possibility in depth here . The Allen atlas also provides a mapping between image regions and brain structures . By matching the brain structures of the rat connectivity map and the brain structures of the mouse brain we are able to construct a combined gene expression/connectivity atlas of the rodent brain ( Materials and Methods ) . Using the combined atlas we find that gene expression levels in different brain regions contain considerable predictive information on their connectivity ( interestingly , more than the level found in previous studies in the worm ) and identify the genes and functional annotations whose expression is most predictive . Obviously some errors may be introduced in this mapping due to inter-species variations in connectivity and expression levels that may hinder the statistical significance of our results . Hence , importantly , the results presented here are likely to be a lower bound on the actual magnitude of the relationship between gene expression and regional brain connectivity . In parallel to our study , another group demonstrated evidence for a correlation between gene expression and connectivity in the rodent brain by using similar sources for gene expression of the mouse brain and rat connectivity maps [21] .
The combined expression/connectivity atlas of the rodent brain contains 176 brain regions . Each is associated here with three signatures . The first signature is a gene expression vector of size 20 , 936 obtained from processing the Allen Brain Atlas . The other two signatures specify brain region connectivity: one encodes the outgoing connections from each region ( Efferent connectivity ) , and the other encodes the incoming connections to each region ( Afferent connectivity ) . Connectivity is obtained from the BAMS atlas [18] using the nomenclature of [22] , assuming that connections that are not reported do not exist [23] . Similarly to [11] we study the connectivity information contained in gene expression by considering both prediction accuracy and the expression/connectivity correlation . Prediction accuracy measures the extent to which connectivity is predicted given the gene expression data . It is estimated for each region separately via a standard cross validation procedure . The correlation between gene expression and connectivity is a global index that measures how similar are the distances between regions in connectivity terms to their distances in expression terms , for all regions at once . On top of predictability and correlation , we also bring further support to our results by examining the enrichment of connectivity-related predicted genes in various disorders that are believed to be related to alterations in brain connectivity . Connectivity prediction ability was studied using a linear SVM classifier ( see Materials and Methods ) . We first obtain results for outgoing connections: In order to examine each region only once , we consider those 146 regions that do not contain other regions , i . e . , regions that are leaves of the regional hierarchy of ABA ( Figure 1 ( a ) ) . Additionally , all regions that have less than 5 outgoing connections are discarded , resulting in a set of 44 regions A1 , … , A44 . We then fix a region Ai and consider the expression signatures of all other leaf regions B1 , … , B146 . At each of the 5 cross-validation iterations , we train a classifier using 4/5 of the regions and obtain a mapping between gene expression of the target region Bj and the existence of an outgoing connection from Ai to Bj . The learned map is then applied to the remaining 1/5 regions in order to obtain predictions on the test data , unseen during training . These 5 iterations produce predictions to all regions B1 , … , B146 , and the overall prediction performance is quantified using the standard Area Under Curve ( AUC ) measure . A p-value is assigned to each region by performing a standard permutation test ( see Materials and Methods ) . An analogous procedure was applied for predicting incoming connections . The resulting prediction ability for outgoing connectivity is significant ( p<0 . 05 ) for 32 out of the 44 regions ( 73% ) . The average AUC was 0 . 74 over all regions , and 0 . 79 for the significant regions . Significant prediction ability was observed also for the incoming connections . There are 57 regions that are not contained in other regions and which have at least 5 incoming connections . Out of these regions 32 ( 56% ) have statistically significant ( p<0 . 05 ) prediction accuracy . The average AUC is 0 . 73 for all the 57 regions and 0 . 83 for the 32 significant ones . The results for the prediction experiments ( combining incoming and outgoing ) are provided in Table S1 , and the significant regions are portrayed in Figure 1 ( b , c ) . The outgoing and the incoming experiments share 35 brain regions that have at least 5 outgoing and 5 incoming connections , out of which 15 are successfully predicted in both incoming and outgoing sets . In several regions of the hierarchy , the BAMS atlas is more detailed than the Allen Brain Atlas , therefore there are known BAMS connections that exist between substructures of the given leafs of the Allen Brain Atlas . In our study , such connections are eliminated since they arise from localized substructures that might have specific gene expression profiles , not necessarily matching that of the larger structures . This conservative approach is in line with the incompleteness of BAMS [24] , i . e . , the conservative connectivity map is geared to allow for more missing links rather than erroneously including spurious ones . However , for completeness , we also report the results obtained when taking a more liberal approach , which propagates links between BAMS substructures up to regions that have ABA analogs , are also presented in Table S1 . This ‘liberal’ connectivity matrix contains well studied links that do not appear in the conservative connectivity map , such as the projection from the dentate gyrus to Ammon's horn . In this experiment too , there are many regions for which the connectivity prediction is significantly above chance −49% of the efferent regions and 58% of the afferent regions show significant predictability . While this is somewhat lower than the results obtained using the conservative connectivity matrix ( 73% and 56% ) , this drop in performance is expected due to the addition of regions with only few known connections , and the specificity of the connections to and from sub-regions that go beyond the resolution of the maps . Several other alternative choices were also made in order to demonstrate the robustness of the experimental design and results , and are also depicted in Table S1 . When choosing a threshold of 10 connections instead of 5 , the average AUC obtained is similar; When replacing the SVM algorithm with the ensemble algorithm gentleBoost [25] , results remain similar or slightly improve . Interestingly , when using the Nearest Neighbor algorithm as the classifier , the results somewhat deteriorate , suggesting that the connectivity predicting patterns are not metrically related in a trivial manner . To provide further support to the validity of the prediction method in the face of missing connectivity data ( as BAMS is probably not comprehensive [24] ) , we also run simulations on synthetic connectivity graphs where one can carefully control the level of missing information ( Materials and Methods ) . The results show that it is possible to have significantly correct predictions even if a large majority of the connections are missing . Supplementary Table S2 shows predictions for individual connections that were obtained by aggregating the results over individual brain regions . Shown are both connections which are known to exist ( 230 outgoing and 207 incoming ) and newly predicted connections that currently have not been reported in the literature ( 416 outgoing and 390 incoming ) , obtained with the natural SVM detection threshold at zero . Using the connectivity prediction paradigm described above we employ a zero-norm SVM feature selection procedure ( see Materials and Methods ) to select the genes whose expression levels are most predictive of connectivity . For each region , the top 500 genes ( out of 20 , 936 ) are selected , and a list of the 500 most frequently selected genes over all regions is formed , one for predicting the outgoing and one for predicting the incoming connections ( Materials and Methods ) . As can be seen in Figure 2 , many genes are selected repeatedly over the different regions in each of the outgoing and the incoming experiments . Remarkably , 241 genes ( out of the 500 most selected ) are shared by both the outgoing and the incoming lists ( the expected number of shared genes according to the hypergeometric distribution is approximately 12 , p<1e-5 ) . The lists of genes selected are reported in Supplementary Table S3 . Thus , in parallel to our finding that the connectivity of many brain regions is predictable on both the outgoing and incoming side , we also find that many genes are informative of both the incoming and outgoing connectivity . Since the outgoing predictions are based on the gene expression vectors of the target regions , and the incoming predictions are based on those of the source regions , the two sets of experiments use two halves of the data and the intersection of the two gene lists is not a statistical necessity . As a control test , we check whether those genes that show the highest region-to-regions variability are those that get selected as predictive . If this were the case , one could attribute their selection to the increased variability and not to their ability to predict connectivity . To this end , all genes were ranked according to their region-to-regions variability , measured as the mean distance from the average expression value , and put in equally sized bins . Then , the intersection of each bin with the two lists of the most informative genes was computed . As is evident from Figure 3 the selected connectivity-predicting genes are not necessarily those genes with the highest region-to-region variability and the two sets are inherently different . Apparently , a large amount of variability points to the influence of other factors that are not related to connectivity . Having such lists gives as an opportunity to estimate the level of involvement of neural connectivity alterations in different brain disorders . To this end , we assembled from the literature lists of the top 100 genes that have been associated with each disorder examined , and quantified the number of ( both efferent and afferent ) connectivity related genes in each such list – the higher this number is , the more likely it is that connectivity alterations may play a role in the pathogenesis of the said disorder ( Materials and Methods ) . Ranked by this measure ( supp Table S4 ) , the disorders we examined are ( from the most associated to the least associated ) Autism , attention deficit disorder , Schizophrenia , anxiety disorder , major depression , Parkinson's disease , bipolar disorder , Alzheimer's disease , obesity , glioma , and cardiovascular diseases . This ranking order fits fairly well with the prominent role ascribed to neuronal connectivity alterations in schizophrenia and autism . To obtain a rough estimate of the role of neuronal connectivity in these disorders as perceived in the literature , we recorded the number of web documents reported by the Google search engine that contained both the name of the disorder and the term “neuronal connectivity” and compared the latter to the connectivity-involvement measure we computed above . The web frequency count , as collected between March 28 and March 30 , 2010 ( supp Table S4 ) , shows that the disorders examined can be divided to three main groups - high ( schizophrenia and autism ) , low ( obesity , glioma and cardiovascular ) and medium level ( the remaining ones ) . Quite remarkably , the high-frequency group has the highest mean of predicted connectivity related genes ( 15 ) , followed by the medium level group ( 11 . 8 ) and then the low level one ( 3 ) . These differences are statistically significant . Notably , one disorder originally belonging to the medium-level group ( attention deficit disorder ) has a similar number of connectivity-related genes as those in the high level group , possibly suggesting a potential role of connectivity alterations in its pathogenesis . A recent comprehensive meta-analysis of genes associated with Schizophrenia [21] , listing 75 Schizophrenia related genes , has provided us an opportunity to examine our pertaining predictions in light of this gene association data . A random intersection of 500 genes would include less than 1 . 8 genes on average . The list of incoming connectivity genes intersects this list by 7 genes ( p<0 . 002 ) , and the outgoing lists intersects it by 4 genes ( p = 0 . 1 ) . To estimate the global correlation ( i . e . , across all regions ) between gene expression and connectivity we represent each of these two information sources as a square matrix that depicts the correlation in either gene expression or the connectivity profiles between every two regions ( see Materials and Methods ) . Three 146×146 matrices are hence obtained: one based on similarity in gene expression and two for the similarity in incoming and outgoing connectivity profiles . Following previous work [11] , [26] , we compute the Pearson correlation between the lower triangular part of the matrices to evaluate correlation between data sources . The correlation between gene expression and outgoing connectivity is 0 . 26 ( p<1e-7 , empirical p<1e-4 ) and the one to outgoing connectivity is 0 . 23 ( p<1e-6 , empirical p<1e-4 ) , showing again that there is a robust and significant relation between gene expression and regional brain connectivity . We then employ such a correlation test to evaluate the connectivity information content of four different sets of genes of interest ( Materials and Methods ) : an axon guidance list based on [27] , a compilation of presynaptic genes [28] , the list of predictive genes identified in C . elegans [11] , and the list of genes that were found to bear an embryologic imprint [29] . The first two lists represent known gene sets that given their axonal/synaptic function are potentially , likely to be involved in determining and maintaining brain connectivity . The Third set has been previously found to be predictive in the worm . The last set might be correlated with connectivity since developmental relationships are sometimes mirrored in connectivity [30] . For each of these four sets we compute the 146×146 expression similarity matrix and examine its correlation to the original connectivity matrix obtained between the 146 different leaf regions . The results are presented in Table 1 . Quite remarkably , only the genes known to be associated with axon guidance from the literature are significantly correlated with the brain regional connectivity and a significant correlation is absent for the three other groups . It is intriguing to find such an association between axon guidance and connectivity-related genes , even when looking at adult expression data . In addition to the four sets of genes , Table 1 presents the p-values of the connectivity correlation test applied to the lists of genes that were collected for each of the medical conditions mentioned above . These results are similar to the expected ranking , with various brain disorder genes showing an inter-region distribution that is significantly correlated with brain connectivity . To further study which gene annotation groups are informative with respect to connectivity , we also applied the correlation test to individual functional annotation groups . For each of 1 , 616 annotation groups in DAVID [31] that were at least partly expressed in the 20 , 936 genes at hand , we compute its 146×146 expression regional expression similarity matrix and examine its correlation to the original connectivity matrix . The results are summarized in Table 2 for the outgoing connectivity and Table 3 for the incoming connectivity , and are given in full in Table S5 . Reassuringly , the top listed functional annotation groups are generally mostly related to neurogenesis , cell-cell signaling , synaptic activity and axonogenesis ( both tables ) , and to neurotransmitter binding and receptor activity on the incoming side . There were 276 outgoing groups with p-value smaller than 0 . 05 , and 200 incoming groups and the two lists share 156 annotation groups ( 18 expected by random ) . Finally , it is interesting to compare the association we found between expression and connectivity of brain regions in rodents to the linkage previously found for single neurons in nematodes . To this end , we reanalyzed the data used in [11] using the global correlation test and created a list of functional annotation groups that are most correlated with connectivity in C . elegans ( Table S6 ) . A Pearson correlation test reveals that the list of p-values obtained for each functional annotation group in the worm is significantly correlated with the similar list obtained for rodents . For outgoing ( incoming ) connectivity , the correlation value is of 0 . 24 , p-value 1e-5 ( 0 . 27 , p-value 1e-6 ) . Hence , there is a certain similarity in the functional gene groups that are associated with neural/brain connectivity across fairly distant phyla and across neuroanatomical scales .
Our work follows a direction set forth by previous work done for single neurons in C . elegans [11]–[13] . Despite obvious differences in the brain complexity , connectivity type , and the amount and quality of the data , it is interesting to compare the prediction performance obtained here to that of its preceding C . elegans investigation . In the previous study of [11] , the mean Area Under the ROC curve ( AUC ) for the prediction experiments is only about 0 . 6 for both incoming and outgoing connectivity . In our results , the average AUC is markedly higher ( 0 . 73 and 0 . 74 ) . For all 289 genes used in [11] , the correlation between connectivity and expression in the worm was 0 . 176 for outgoing connectivity , and 0 . 075 for incoming connectivity . Looking at all of the 20 thousands plus genes used in this work at once , the equivalent correlations are 0 . 26 and 0 . 23 . Moreover , there is considerable variance in the predictability in different regions and some regions achieve quite high predictive values ( 0 . 83 and 0 . 79 mean AUC values over the significant regions , with maximal AUC values reaching 0 . 99 ) . Our results are further supported by the recent parallel contribution of French and Pavlidis [21] , in which a similar correlation test yields a score of 0 . 22 and 0 . 26 for incoming and outgoing connectivity respectively . The work of [21] is focused on the correlation assay and the authors state that they were unable to perform convincing predictive experiments . Here , in difference , we show that there is a considerable predictive signal . In fact , the prediction capability is considerably stronger than that found in the worm , and many of the brain regions present a marked and highly significant level of predictability . This prediction ability is further used here to select the lists of connectivity-related genes . A predictive test is , in our minds , a more solid foundation for gene selection than a correlation test . This is because a combination of even uninformative features can produce a correlation map that is similar to a given input map , while the separation between train and test data in the prediction experiments is much less prone to this pitfall . The lists of selected connectivity-related genes we obtain are verified here by comparing them to various lists obtained from the literature , again , going beyond the results presented in [21] . Regions of high predictability do not seem to be clustered in specific parts of the hierarchy . While smaller nuclei with many connections and therefore more available data seem somewhat easier to predict , a comparison between a structure's volume and the predictability of its connectivity map shows that regions of all sizes depict good predictability ( Supplementary Figure S2 ) . This might suggest that all regions are potentially of high predictability; however , the quality of the data currently available limits our ability to uncover their true predictability . The correlation between spatial proximity and connectivity is 0 . 11 and 0 . 10 for outgoing and incoming connectivity ( compared to 0 . 26 and 0 . 23 ) . Thus , while in the brain nearby regions are more likely to be connected , this association is significantly lower than the association between gene expression and connectivity . To build the combined rodent brain atlas that contains both expression and connectivity , we rely on available resources that are not fully compatible or complete . Some of the connectivity that is currently absent in the rat atlas may actually exist in the rodent brain . The assumption of conservation of connectivity and expression between mouse and rat , underlying the construction of a combined atlas of a common rodent ancestor , probably holds only partly . Furthermore , the gene expression data was not measured during brain development , as would ideally have been more befitting . Yet , as both connectivity and expression are associated with common factors such as functionality , it is perhaps not surprising that considerable pertaining information can be delineated in adult expression patterns of neurons . As evident , the latter permit a considerable level of connectivity prediction , exhibit significant correlations with the connectivity data , and show a marked overlap between genes that are discriminative for incoming and outgoing connectivity . Finally , strictly speaking , we identify an association and not a causal relation from genes to connectivity . Although this causal direction is expected based on current consensus , it is certainly possible that connectivity in turn affects gene expression – one possible route for such effects may be indeed via activity-dependent mechanisms that shape synaptic formation and maintenance , mentioned earlier [8]–[9] . Despite the above limitations to the quality of the data , we were able to uncover a fairly marked association between gene expression and connectivity . Thus , we are able to make a significant advancement toward the long term goal of inferring the connectome from the genome [32] . Naturally , had our data been richer , for example , alleviating the need to rely on conservation across species , even better results could be expected . However , especially given these limitations , the magnitude of the association found here is truly remarkable , and the large-scale analysis approach presented here will undoubtedly show its continuing value in future studies as more refined data accumulates . This type of analysis is valid for both single neuron connectivity and connectivity between brain regions , and it is likely to be valid for intermediate , mesoscopic scales [24] , [33] . In the nearby future , such efforts can be applied to link between newly established connectivity maps in humans ( e . g . [34] ) with accumulating regional gene expression data in the human brain . Moreover , once the genetic atlas of the developing brain [16] is processed to register gene maps , a distinction can be drawn between genes that are associated in maintaining connectivity and genes that are dominant during the initial formation of brain connectivity . With the future advent of better and more accurate data we might be able to perform the analysis presented here focusing solely on the gene expression of neuronal cells while disregarding other cell types . To gain preliminary experimental insight into the role played by cell type in determining the link between expression and connectivity , we have examined the human data available from two recent papers . The first paper [35] has microarray data collected from the brains of AD patients and controls . In the second paper [36] , care was taken such that the gene expression data was collected from neurons only . Therefore , for a first approximation , we have samples that are glia + neurons and samples that are only neurons . By comparing the two sets of samples we can identify genes that are over-expressed in glia and not over-expressed in neuron samples ( Note that the situation is not symmetric and the opposite list cannot be extracted without further assumptions ) . Working with the mouse homologs of the identified human genes , we find that those genes that tend to be over expressed in glia are less informative than a typical group of the same size . The p-value of this finding is borderline though – 0 . 02 for efferent correlation test and 0 . 17 for the afferent correlation test . Future studies analyzing neuronal vs glial expression data comparatively are hence needed to shed further light on this intriguing question .
Our study has been made possible thanks to the innovative open approach of the Allen Brain project [16] . Gene expression data was obtained from the Allen Mouse Brain Atlas ( ABA ) dataset [15] for gene expression in the adult mouse brain composed of 20 , 936 genes ( http://mouse . brain-map . org/ ) . For each gene a 200 micron 3d volume of gene expression in the mouse brain is available ( a vector of length ∼150 k ) . Some genes have several scans . Scans are available in one of two planes: Coronal and Sagittal . We compiled a dataset of voxel gene expressions based on sagittal scans . When numerous scans exist for a single gene a mean is taken ( maximum was also tried – resulting in only minute , negligible differences in results reported ) . For linking voxels to brain structures we use the structural annotation available at ABA ( http://mouse . brain-map . org/pdf/Allen_Reference_Atlases . pdf ) . It defines a nomenclature of 209 brain structures organized in a hierarchy . The gene expression for each brain structure is computed as the average of all voxels contained within that region . Once more , experiments were also performed by taking the maximal value instead of the mean with little , negligible influence on the connectivity prediction ability and on the results reported . One should note that during the preparation of this work partial results on the developing mouse brain have been uploaded to the ABA website . These results are not complete enough to enable us to run our experiments on a developing brain . For example , there is no mapping currently available between voxels and brain structures . Rat connectivity information is obtained from [18] . To match rat connectivity to mouse gene expression we link the rat nomenclature of [22] and the ABA mouse nomenclature , by creating a mapping between identical terms . The mapping is given in Table S7 . It sometimes occurs that a region is identified in the mouse nomenclatures and at least one of the children of this region is not identified . Even in such cases , we do not perform the analysis on the non-leaf regions . This policy simplifies the framework and minimizes borderline cases , for example , when some of the leaves are identified and some are not . We use a Linear Support Vector Machine ( SVM ) [37] classification with a fixed parameter of C = 1 for prediction . The learned binary labels correspond to the existence or non-existence of a connection between regions . Regions with less than 5 positive examples ( i . e . connections ) are discarded . For each region separately , a balanced 5-fold cross-validation is performed on this data with 80% training and 20% testing . Since each connection ( existing or not ) is tested exactly once , the cross validation procedure produces a connectivity prediction value for each possible connection . We consider the real value which is the signed distance from the learned classifier's separating hyperplane , and use it to compute the Area Under Curve ( AUC ) statistics . To eliminate dependence on the random split used , each such cross-validation experiment is repeated 20 times , and the mean AUC is recorded . In order to evaluate statistical significance , the entire experiment is repeated 1 , 000 times while permuting the labels . To demonstrate the validity of the prediction assay in the face of missing connectivity data we perform the following synthetic data experiment: A random network was created of a similar cardinality as the BAMS network used in our experiments , such that the degrees of the nodes are five times higher than those of the BAMS network ( varying between nodes , similarly to BAMS ) . Synthetic random vectors of “gene expression” were created in such a way that nodes that are connected to a specific node have for a subset of the genes a somewhat similar pattern , randomly varied around a certain central pattern , i . e . , tend to have some genes overexpressed and some genes underexpressed in a similar manner . Then , we run the same protocol as in our prediction assay and measure success by computing the mean AUC obtained from all regions ( the equivalent success in the real data experiments is 0 . 73 ) . This experiment is then repeated when some of the initially given positive connections are held out and marked as ‘non existing’ ( i . e . , incorporating missing data in a controlled manner ) . The results of the simulations for specific missing data values , averaged over many runs are presented below in Supplementary Figure S1 . As can be seen , even for such challenging simulations where the prediction for the full dataset is at 80% , the results degrade nicely with the number of missing connections . In these noisy conditions the results vs the simulated atlas remain well above chance even when only 15% of the connections are retained ( i . e . , ‘known’ , blue-line ) . Moreover , the classifiers learned with the missing data are useful for predicting the complete ( no missing data ) simulated connections ( red-line ) . To examine the correlation between a genetic pattern and a connectivity pattern across all brain structures under investigation , we used an assay similar to the one used by Toledo–Rodriguez et al [26] . This assay was also used in [11] . Given a set of N = 146 structures , we constructed two N×N similarity matrices , S1 and S2 , where S1 ( S2 ) represents the pairwise similarity between the expression data ( connectivity ) of every two brain structures . Pearson correlation is used as a measure of those pairwise similarities for both gene expression and connectivity , both between the vectors of gene expression , and the connectivity vectors . The ( N * N/2 – N ) entries forming the lower triangle of S1 ( S2 ) are concatenated to form a covariation vector v1 ( v2 ) . The Pearson correlation between the two covariation vectors v1 and v2 describes the extent to which similarities in gene expression imply similarities in connectivity and vice-versa . The statistical significance of the resulting correlation is computed using an empiric null hypothesis constructed from repeating the procedure with shuffling . On each repetition the gene expression signatures were shuffled amongst all regions , thus disassociating a region and its gene expression . The p-values are calculated by repeating the shuffling 1 , 000 times and computing the probability to achieve a score equal or higher than the score of the non-shuffled data . Similarly to the prediction assay , for each brain region we take connected regions gene expression as positive examples and non-connected regions as negative examples . This is done once for outgoing connections , and once for incoming connections , where the two experiments are performed independently . At each time , feature ( gene ) selection was performed using zero norm SVM algorithm [38] . Zero norm SVM works by iteratively training an SVM while reweighing the feature vectors until convergence . In order to select a fixed number of features , we have selected the 500 features with the highest weights provided by the zero-norm SVM procedure . This is repeated for each brain structure which has at least 5 connections , i . e . , to 44 regions in the outgoing experiment and to 57 regions in the incoming experiment . To obtain two global lists of selected genes that are informative to either outgoing connectivity or incoming connectivity , the individual lists obtained for each region are combined . This is done by counting for each gene the number of times it was selected across the brain structures in each of the two experiments . The 500 genes that appeared most frequently in the individual outgoing experiments form the list of selected outgoing genes , and similarly for the incoming list . To gain more insight into the nature of the selected genes , we have employed the DAVID functional annotation tools [31] to determine the most prominent annotations in the two lists formed above . The details of this experiment are provided in Supplementary Table S8 . To alleviate potential concerns about the influence of artifacts in the gene expression data on the prediction and gene selection process , we have compared the prevalence of artifacts in the data of selected genes to that of a disjoint sample of genes . 50 genes were sampled randomly from the groups 241 genes that are found to be predictive for both outgoing and incoming connectivity . Another group of 50 genes was sampled from the 1000 most brain active genes that do not appear in either list of predictive genes . For further control , genes that were not highly expressed in the brain were removed from the study since their images are expected to contain less data and therefore fewer artifacts . The results show that for the sample of connectivity predictive genes , 58% of the slices contained local artifacts such as localized stains . The equivalent number for the background group is 57% . The ratio of global artifacts such as folds and scratches are also quite similar between the two groups: 11% and 17% respectively . Overall , we do not observe a tendency for more artifacts in the selected genes in comparison to the general population of brain-expressed genes . Supplementary table S9 contains the raw data of this analysis . The top 100 genes associated with each disorder were extracted from the HuGe database [39] , and the size of the intersection of these lists and the two lists of connectivity genes extracted by the feature selection method above were computed . The expected size of a random intersection is 2 . 5 genes . There were 4 such lists . ( 1 ) Axon guidance genes were obtained from the gene families discussed in [27]: Netrin , Slit , Semaphorin , Ephrin , DCC , UNC5 , Robo , Robo3 , Neuropilin , Plexin and Eph . A total of 86 homologous members of these families were matched in the ABA gene set . ( 2 ) A group of 103 pre-synaptic gene homologs was obtained from a list of 107 genes appearing in [28] . ( 3 ) C . elegans genes were obtained from mouse homologies on the most highly ranked genes shown to be involved in neural connectivity in [11] . ABA homologies of 19 outgoing ( 31 incoming ) were obtained from 30 outgoing ( 53 incoming ) C . elegans genes . ( 4 ) The list of genes which are indicative of embryonic history taken from [29] . 83 such genes were identified within the ABA gene list out of 93 in the original list . In order to compute the significance of the correlation assay results obtained by a group of genes , such as the three literature based gene-lists or the 1 , 616 DAVID groups , we have compared the p-value obtained using the correlation assay with the p-values obtained for 1000 random groups of the same size . This procedure eliminates bias caused by the group size . | Brain connectivity is believed to be associated with gene expression levels in the developing and the adult animal . Recently , this association has been explored in two model animals: the worm C . elegans at the level of single neurons; and the mouse , where specific subpopulations of neurons in the hippocampus were studied . Inspired by these studies , we set out to generalize their scope and examine the possibility of using gene expression signatures to predict regional connectivity in the whole rodent brain . Our results show a higher degree of association between connectivity and expression than shown before , and key genes are identified that are highly predictive of brain connectivity . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genomics",
"computational",
"neuroscience",
"biology",
"computational",
"biology",
"neuroscience",
"genetics",
"and",
"genomics"
] | 2011 | Gene Expression in the Rodent Brain is Associated with Its Regional Connectivity |
Lactoferrin is a multifunctional mammalian immunity protein that limits microbial growth through sequestration of nutrient iron . Additionally , lactoferrin possesses cationic protein domains that directly bind and inhibit diverse microbes . The implications for these dual functions on lactoferrin evolution and genetic conflicts with microbes remain unclear . Here we show that lactoferrin has been subject to recurrent episodes of positive selection during primate divergence predominately at antimicrobial peptide surfaces consistent with long-term antagonism by bacteria . An abundant lactoferrin polymorphism in human populations and Neanderthals also exhibits signatures of positive selection across primates , linking ancient host-microbe conflicts to modern human genetic variation . Rapidly evolving sites in lactoferrin further correspond to molecular interfaces with opportunistic bacterial pathogens causing meningitis , pneumonia , and sepsis . Because microbes actively target lactoferrin to acquire iron , we propose that the emergence of antimicrobial activity provided a pivotal mechanism of adaptation sparking evolutionary conflicts via acquisition of new protein functions .
Genetic conflicts between microbes and their hosts are an important source of evolutionary innovation [1] . Selective forces imposed by these antagonistic interactions can give rise to dramatic bouts of adaptive gene evolution through positive selection . J . B . S . Haldane originally speculated on the importance of infectious disease as an “evolutionary agent” over 60 years ago [2] , and the Red Queen hypothesis later posited that predators and their prey ( or pathogens and their hosts ) must constantly adapt in order to sustain comparative fitness [3 , 4] . More recent studies have demonstrated how evolutionary conflicts progress at the single gene or even single nucleotide level , as molecular interfaces between host and microbial proteins can strongly impact virulence and immunity [5–7] . Host-pathogen interactions thus provide fertile ground for studying rapid gene evolution and acquisition of novel molecular traits [8] . Lactoferrin presents a compelling model for investigating adaptation from an ancestral “housekeeping” function to a specialized immunity factor . Lactoferrin arose from a duplication of the transferrin gene in the ancestor of eutherian mammals roughly 160 million years ago [9] . A fundamental and shared feature of these proteins is the presence of two evolutionary and structurally homologous iron binding domains , the N and C lobes , each of which chelates a single iron ion with high affinity . Iron binding by these proteins can effectively starve microbes of this crucial metal , a protective effect termed nutritional immunity [10 , 11] . Microbes in turn actively scavenge iron from these and other host proteins in order to meet their nutrient requirements [12 , 13] . The importance of iron in human infectious disease is highlighted by genetic disorders of iron overload , such as hereditary hemochromatosis , which render affected individuals highly susceptible to bacterial and fungal infections [14 , 15] . In addition to its role in nutritional immunity , lactoferrin has acquired new immune functions independent of iron binding following its emergence in mammals . Lactoferrin is expressed in a variety of tissues and fluids including breast milk , colostrum , saliva , tears , mucous , as well as the secondary granules of neutrophils and possesses broad antimicrobial activity [16] . Portions of the lactoferrin N lobe are highly cationic , facilitating interaction with and disruption of microbial membranes . Two regions of the lactoferrin N lobe in particular , lactoferricin and lactoferrampin , can be liberated from the lactoferrin polypeptide by proteolytic cleavage and exhibit potent antimicrobial activity against bacteria , fungi , and viruses [17 , 18] . Lactoferrin , as well as lactoferricin alone , can directly bind the lipid A component of lipopolysaccharide ( LPS ) as well as lipoteichoic acid , contributing to interactions with surfaces of Gram-negative and Gram-positive bacteria [19 , 20] . Lactoferrin thus poses a unique challenge for microbes—while its ability to bind iron makes it an attractive target for “iron piracy , ” lactoferrin surface receptors could render cells more susceptible to associated antimicrobial activity . Despite a growing appreciation for lactoferrin’s immune properties , the evolutionary implications of these unique functions remain unclear . In the present study we decipher recent signatures of natural selection acting on lactoferrin in primates as well as modern humans to understand the evolutionary consequences of a newly acquired antimicrobial activity from a distinct ancestral function .
To assess the evolutionary history of lactoferrin in primates , we assembled gene orthologs from publicly available databases and cloned lactoferrin complementary DNA ( cDNA ) prepared from primary cell lines . In total , we compared 15 lactoferrin orthologs from hominoids , Old World , and New World monkeys , representing roughly 40 million years of primate divergence ( Fig 1A and S1 Fig ) . We then used maximum likelihood-based phylogenetic approaches ( performed with the PAML and HyPhy software packages ) to calculate nonsynonymous to synonymous substation rate ratios ( dN/dS ) across this gene phylogeny [21–23] . For our study we included the N-terminal 19 amino acid positions of the full-length lactoferrin protein , which are removed during processing of the mature polypeptide in humans . Our analysis indicated that lactoferrin has evolved under episodic positive selection in the primate lineage , consistent with a history of evolutionary conflict with microbes ( Fig 1A and S1–S7 Tables ) . These findings are also in line with previous genome-wide scans for positive selection in primates which identified the lactoferrin gene ( LTF ) among other candidate loci [24] . We next determined signatures of selection across individual codons in lactoferrin . In total , 17 sites displayed strong evidence of positive selection ( posterior probability >0 . 95 from Naïve Empirical Bayes and Bayes Empirical Bayes analyses in PAML ) , with 13 of the 17 sites found in the N lobe ( Fig 1B and 1C and S1 Fig and S2 , S4 , S5 and S6 Tables ) . This observation was notably dissimilar from a parallel analysis of primate serum transferrin , where sites under positive selection were restricted to the C lobe ( Fig 1B and 1C and S3 Table ) . These results are further consistent with our previous work indicating that rapid evolution in primate transferrin is likely due to antagonism by the bacterial iron acquisition receptor TbpA , which exclusively binds the transferrin C lobe [25–28] . Thus , while lactoferrin and transferrin both exhibit signatures of positive selection in primates , patterns of selection across the two proteins are highly discordant . Evidence of episodic positive selection in primate lactoferrin led us to more closely investigate variation of this gene across human populations . Data from the 1000 Genomes Project revealed six nonsynonymous polymorphisms at greater than 1% allele frequency in humans ( S8 Table ) . Of the 17 sites we identified as rapidly evolving across primate species , amino acid position 47 overlapped with a high frequency arginine ( R ) to lysine ( K ) substitution in the N lobe of lactoferrin in humans ( Fig 2A and S8 and S9 Tables ) . This position is markedly polymorphic between populations; while individuals of African ancestry carry the K47 allele at about 1% frequency , this variant is found in non-African populations at roughly 30–65% allele frequency , with the highest frequencies observed among Europeans ( Fig 2B and S9 Table ) . The presence of R47 in related great apes combined with its high frequency in African populations suggests that R47 is in fact the ancestral allele in humans . Data from the Neanderthal genome browser ( http://neandertal . ensemblgenomes . org ) further revealed lysine to be the consensus residue at position 47 in recently sequenced Neanderthals . The presence of the lactoferrin K47 allele in Neanderthal and non-African human populations and its near absence in Africans suggests one of several intriguing genetic models for the history of this variant , including long-term allelic diversity in hominins , convergent evolution , or introgression from Neanderthals into modern humans . Given the shared variation at position 47 between primate species and among human populations , we sought to determine whether lactoferrin exhibits signatures of positive selection in modern humans . Calculation of pairwise FST between a subset of human populations identified an elevated signal of differentiation between European ( CEU ) and African ( YRI ) populations [29] , consistent with observed differences in allele frequencies between these groups ( S2 Fig ) . The FST at rs1126478 was 0 . 70 ( empirical p-value < 0 . 001 ) , 0 . 30 , and 0 . 03 for CEU-YRI , CEU-CHB , and CEU-FIN , respectively . Single nucleotide variants neighboring rs1126478 also showed signs of elevated FST suggesting that a shared CEU haplotype was driving the signal of differentiation ( S2 Fig ) . We next applied measures of haplotype homozygosity to assess the possibility that the K47 haplotype has been subject to natural selection in humans . Linkage around R47 alleles breaks down rapidly within a few kilobases , while the K47 variant possesses an extended haplotype ( homozygosity of 0 . 5 at 21 , 913 bases ) , consistent with the possibility of an adaptive sweep in this genomic region ( Fig 2C ) . A selective sweep is also consistent with bifurcation plots around position 47 , where the K47 haplotypes possess increased homogeneity relative to R47 haplotypes ( Fig 2D ) . We observed a slight an elevation of the genome-wide corrected integrated haplotype score ( iHS ) for the K47 allele ( -1 . 40136 ) and a depletion of observed heterozygosity ( S2 , S3 and S4 Figs ) . We also examined the patterns of cross population extended haplotype homozygosity ( XP-EHH ) . Consistent with the FST and EHH results , the XP-EHH score was elevated at the K47 position when CEU individuals were compared against YRI ( 1 . 1; p-value: 0 . 129 ) or CHB ( 3 . 1; p-value: 0 . 003 ) ( S5 Fig ) . While XP-EHH between CEU and YRI was moderate , surrounding SNPs less than 3 kilobases away had values as high as 2 . 89 ( rs189460549; p-value: 0 . 01 ) . Genome-wide , the K47 XP-EHH signal is moderate compared to other loci . Next we compared the joint distribution of the p-values from dN/dS analyses [24] with the empirical p-values from the CEU-CHB XP-EHH analyses ( S6 Fig ) . The previous genome-wide rank for lactoferrin , from dN/dS analyses , was 226 before considering the joint distribution and 156 after . The top 20 genes with the greatest change in rank ( dN/dS p-value < 0 . 01 ) include BLK , DSG1 , FAS , SLC15A1 , GLMN , SULT1C3 , WIPF1 , and LTF . This meta-analysis highlights candidate genes that have undergone species-level as well as population-level selection in primates and humans , respectively . By integrating molecular phylogenetic analyses and population genetics approaches , we pinpointed signatures of positive selection associated with an abundant human lactoferrin polymorphism . Signatures of positive selection in the lactoferrin N lobe among diverse primates , including position 47 in humans , led us to more closely investigate evolutionary pressures that have influenced variation in this region . After gene duplication from ancestral transferrin , lactoferrin gained potent antimicrobial activities independent of iron binding through cationic domains capable of disrupting microbial membranes . Two portions of the lactoferrin N lobe in particular , termed lactoferricin ( amino acids 20–67 in full-length protein; 1–48 in mature protein ) and lactoferrampin ( amino acids 288–304 in full-length protein; 269–285 in mature protein ) , have been implicated in these antimicrobial functions [18 , 30] . Phylogenetic analysis revealed that several sites corresponding to lactoferricin and lactoferrampin display signatures of positive selection ( Fig 3A and 3B ) . Notably , positive selection in lactoferricin localized to sites harboring cationic ( lysine , arginine ) or polar uncharged residues ( asparagine ) , which could mediate membrane disruption and regulate antimicrobial activity . Position 47 , which exhibits signatures of selection in humans as well as other primates , also lies within the lactoferricin peptide region . In contrast , hydrophobic tryptophan residues proposed to mediate insertion into microbial membranes are completely conserved among primates , as are cysteine residues that participate in intramolecular disulfide bond formation ( Fig 3A ) . We also observed rapid evolution of the position immediately C-terminal to the pepsin cleavage site in lactoferrampin ( Fig 3A ) , suggesting that the precise cleavage site in this peptide may be variable among species . Notably , the proteases responsible for lactoferrin processing in mucosal secretions and neutrophils remain elusive; identification of such factors will assist in revealing the consequences of genetic variation proximal to cleavage sites . Expanding our phylogenetic analysis to other mammalian taxa , we found that lactoferrin also exhibits signatures of positive selection in rodents and carnivores ( S7 Fig and S10 Table ) . While the specific positions that contribute most strongly to these signatures could not be resolved with high confidence , N-terminal regions corresponding to lactoferricin in primates are absent in several rodent and carnivore transcripts , suggesting that this activity may have been lost or modified in divergent mammals . These observations are further consistent with previous work which identified signatures of positive selection in lactoferrin antimicrobial peptide domains across diverse mammals [31] . Together these results demonstrate that lactoferrin-derived cationic peptides of the N lobe are rapidly evolving at sites critical for antimicrobial action . While rapid evolution of the lactoferrin N lobe may reflect selection for improved targeting of microbial surfaces , it could also represent adaptations that prevent binding by inhibitors encoded by bacteria . For example , pneumococcal surface protein A ( PspA ) is a crucial virulence determinant of Streptococcus pneumoniae , and several studies have demonstrated that PspA specifically binds and inhibits antimicrobial portions of the lactoferrin N lobe [32] . Consistent with an important evolutionary impact for this interaction , numerous sites under positive selection in the lactoferrin N lobe lie proximal to the PspA binding interface [33] , including those corresponding to the lactoferricin peptide ( Fig 3C ) . These data suggest that adaptive substitutions in lactoferrin could negate PspA binding , leading to enhanced immunity against S . pneumoniae or related pathogens . Many strains of pathogenic Neisseria , which cause the sexually transmitted disease gonorrhea as well as acute meningitis , encode lactoferrin binding proteins ( LbpA and LbpB ) which mediate iron acquisition from lactoferrin [34 , 35] . Of four sites identified under positive selection in the lactoferrin C lobe , at least two appear proximal to the proposed Neisseria LbpA binding interface based on recent molecular modeling studies ( S8 Fig ) [36] . One of these , position 589 , also aligns to a region under strong positive selection in transferrin ( position 576 in humans ) which directly contacts the related bacterial receptor TbpA ( Fig 1B ) [28] . These findings suggest that , similarly to transferrin , antagonism by bacterial Lbp proteins may have promoted natural selection in the lactoferrin C lobe . Signatures of selection at distinct lactoferrin-pathogen interfaces thus highlight the diverse conflicts that have arisen during the evolution of this unique immunity factor .
Together our results suggest that the emergence of novel antimicrobial activity in the N lobe of lactoferrin strongly influenced host-microbe interactions in primates , including modern humans ( Fig 4 ) . High disparity in sites under positive selection between the N and C lobes of lactoferrin and transferrin indicate that distinct selective pressures influenced these proteins during primate evolution . We previously demonstrated that primate transferrin has been engaged in recurrent evolutionary conflicts with the bacterial receptor , TbpA [25] . This receptor is an important virulence factor in several Gram-negative opportunistic pathogens including Neisseria gonorrhoeae , Neisseria meningitidis , Haemophilus influenzae , as well as related animal pathogens [26 , 37–39] . Notably , TbpA binds and extracts iron exclusively from the C lobe of transferrin , and signatures of positive selection in transferrin are almost entirely restricted to the TbpA binding interface ( Fig 1 ) [25] . The fact that transferrin family proteins are recurrently targeted by microbes for iron acquisition may have provided the selective advantage for antimicrobial functions that arose in the lactoferrin N lobe . Our results suggest at least two non-mutually exclusive scenarios for evolutionary conflicts involving the lactoferrin N lobe . Positive selection in this region could reflect adaption of lactoferrin for enhanced targeting of variable pathogen surfaces . Lactoferricin is capable of binding the bacterial LPS , which itself is heavily modified in many human-associated bacteria to mediate immune evasion and could provoke counter-adaptations at this interface . Conversely , variation in the lactoferrin N lobe could negate interactions with bacterial inhibitory proteins such as PspA encoded by S . pneumoniae . Lactoferrin binding activity has also been identified in several other important bacterial pathogens including Treponema pallidum [40] , Staphlococcus aureus [41] , and Shigella flexneri [42] , raising the possibility of multiple independent evolutionary conflicts playing out at the lactoferrin N lobe . Iron-loaded lactoferrin could further be viewed as a “Trojan horse , ” where microbes that target it as a nutrient iron source may be more susceptible to antimicrobial peptides . Consistent with this hypothesis , recent work has suggested that Neisseria encoded LbpB recognizes the lactoferrin N lobe , in contrast to its homolog TbpB which selectively interacts with the iron-loaded C lobe of transferrin [35 , 43 , 44] . LbpB binding to the lactoferrin N lobe could thus provide a counter-adaptation with dual benefits by neutralizing lactoferrin antimicrobial activity through negatively charged protein surfaces while simultaneously promoting iron acquisition by its co-receptor , LbpA [43] . These observations point to adaptations involving de novo protein functions on both sides of this molecular interface . It is important to note that many “pathogenic” bacteria that routinely encounter lactoferrin in the respiratory mucosa are generally commensals that rarely cause disease . For example , H . influenzae colonizes a huge proportion of the human population but typically only causes disease in young children who lack a robust immune response . In addition , the dual functions of lactoferrin likely have pleiotropic effects on complex microbial communities in the host mucosa , with inhibition of some members creating new niches for others . Thus , the evolutionary forces acting on lactoferrin and the consequences for positive selection are likely more nuanced than a two-dimensional host-pathogen arms race . Future studies aimed at understanding the functional impact of lactoferrin variation will assist in understanding such complex biological effects . Our results raise the possibility that the lactoferrin K47 variant introgressed into humans from Neanderthals at some point after the out-of-Africa expansion [45] . An alternative explanation could be convergent evolution of lactoferrin in distinct lineages of early hominins for enhanced immune function . Recent reports indicate that the human lactoferrin K47 variant , within the N lobe lactoferricin peptide , may have a protective effect against dental cavities associated with pathogenic bacteria [46] . Moreover , saliva isolated with patients homozygous for the K47 variant possesses enhanced antibacterial activity against oral Streptococci relative to homozygous R47 individuals [47] . Future analysis of lactoferrin sequence in archaic humans could provide additional insight on the history and functional properties of this variant . Together these studies provide a direct link between variation in the lactoferrin N lobe and protection against disease-causing bacteria , consistent with adaptive evolution of lactoferrin in humans and other primates . Notably , the lactoferrin gene , LTF , is located only ~60 kilobases away from CCR5 , a chemokine receptor which is also an entry receptor for HIV [48–52] . A 32-base pair deletion in CCR5 ( CCR5-Δ32 ) confers resistance to HIV infection , and is present at a high frequency in northern Europeans while absent from African populations [53] . Although early evidence suggested that CCR5-Δ32 might itself be subject to positive selection in humans , more recent studies have concluded that these signatures are more consistent with neutral evolution [54] . It is intriguing that , like CCR5-Δ32 , the lactoferrin K47 variant exhibits increased allele frequency in European populations relative to Africans . However , the presence of the K47 variant at high frequencies in Asian and American populations points to a much earlier origin for this variant than CCR5-Δ32 . Moreover , EHH and bifurcation analyses indicate that the haplotypes associated with the lactoferrin K47 variant do not encompass CCR5 , suggesting that variation at the CCR5 locus is unlikely to contribute to signatures of selection in LTF ( Fig 2B and 2C and S9 Table ) . The proximity of the LTF and CCR5 genes combined with their high degree of polymorphism and shared roles in immunity suggest the potential for genetic interactions relating to host defense . Future studies could reveal functional or epidemiological links between these two factors in human immunity . In summary , we have discovered that lactoferrin constitutes a crucial node of host-microbe evolutionary conflict based on signatures of natural selection across primates , including humans . Our findings suggest an intriguing mechanism for molecular arms race dynamics where adaptations and counter-adaptations rapidly emerge at the level of new protein functions in addition to recurrent amino acid substitutions at a single protein interface ( Fig 4 ) . Our evolutionary analyses highlight how the process of gene duplication and subfunctionalization can drastically alter the progression of host-microbe genetic conflicts .
RNA was obtained from the following species via the Coriell Cell Repositories where sample codes are indicated: Homo sapiens ( human; primary human foreskin fibroblasts; gift from A . Geballe ) , Gorilla gorilla ( western lowland gorilla; AG05251 ) , Papio anubis ( olive baboon; PR00036 ) , Lophocebus albigena ( grey-cheeked mangabey; PR01215 ) , Cercopithecus aethiops ( African green monkey; PR01193 ) , Colobus guereza ( colobus monkey; PR00240 ) , Callithrix geoffroyi ( white-fronted marmoset; PR00789 ) , Lagothrix lagotricha ( common woolly monkey; AG05356 ) , Saimiri sciureus ( common squirrel monkey; AG05311 ) . Gene sequences from additional primate , rodent , and carnivore species were obtained from Genbank . RNA ( 50 ng ) from each primate cell line was prepared ( RNeasy kit; Qiagen ) and used as template for RT–PCR ( SuperScript III; Invitrogen ) . Primers used to amplify lactoferrin cDNA were as follows: GTGGCAGAGCCTTCGTTTGCC ( LF-forward; oMFB256 ) and GACAGCAGGGAATTGTGAGCAGATG ( LF-rev; oMFB313 ) . PCR products were TA-cloned into pCR2 . 1 ( Invitrogen ) and directly sequenced from at least three individual clones . Gene sequences have been deposited in Genbank ( KT006751 –KT006756 ) . DNA multiple sequence alignments were performed using MUSCLE and indels were manually trimmed based on amino-acid comparisons . A generally accepted primate species phylogeny [55] ( Fig 1A ) was used for evolutionary analysis . A gene tree generated from the alignment of lactoferrin corresponded to this species phylogeny ( PhyML; http://atgc . lirmm . fr/phyml/ ) . Maximum-likelihood analysis of the lactoferrin and transferrin data sets was performed with codeml of the PAML software package [21] . A free-ratio model allowing dN/dS ( omega ) variation along branches of the phylogeny was employed to calculate dN/dS values between lineages . Two-ratio tests were performed using likelihood models to compare all branches fixed at dN/dS = 1 or an average dN/dS value from the whole tree applied to each branch to varying dN/dS values according to branch . Positive selection in lactoferrin was assessed by fitting the multiple alignment to either F3X4 or F61 codon frequency models . Likelihood ratio tests ( LRTs ) were performed by comparing pairs of site-specific models ( NS sites ) : M1 ( neutral ) with M2 ( selection ) , M7 ( neutral , beta distribution of dN/dS<1 ) with M8 ( selection , beta distribution , dN/dS>1 allowed ) . Additional LRTs from the HyPhy software package that also account for synonymous rate variation and recombination ( FUBAR , REL , FEL , MEME , BUSTED ) were performed [22 , 23] . Molecular structures of lactoferrin , transferrin and associated proteins were visualized using Chimera ( http://www . cgl . ucsf . edu/chimera/ ) . For variant-based analyses we used genotype calls from the 1000 Genomes project ( release: 20130502 , shapeit2 phased ) . Weir and Cockerham’s Fst estimator [29] was used for the population comparisons , implemented in GPAT++ . EHH and the bifurcation diagrams were calculated using the [R] package REHH [56] . Genome-wide iHS scans were performed using GPAT++ and XPEHH plots were generated previously published datasets [57 , 58] . | Immunity genes can evolve rapidly in response to antagonism by microbial pathogens , but how the emergence of new protein functions impacts such evolutionary conflicts remains unclear . Here we have traced the evolutionary history of the lactoferrin gene in primates , which in addition to an ancient iron-binding function , acquired antimicrobial peptide activity in mammals . We show that , in contrast to the related gene transferrin , lactoferrin has rapidly evolved at protein domains that mediate iron-independent antimicrobial functions . We also pinpoint signatures of natural selection acting on lactoferrin in human populations , suggesting that lactoferrin genetic diversity has impacted the evolutionary success of both ancient primates and humans . Our work demonstrates how the emergence of new host immune protein functions can drastically alter evolutionary and molecular interactions with microbes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bacteriology",
"antimicrobials",
"organismal",
"evolution",
"medicine",
"and",
"health",
"sciences",
"drugs",
"population",
"genetics",
"microbiology",
"vertebrates",
"animals",
"mammals",
"primates",
"microbial",
"evolution",
"pharmacology",
"population",
"biology",
"evol... | 2016 | Antimicrobial Functions of Lactoferrin Promote Genetic Conflicts in Ancient Primates and Modern Humans |
RNA-binding proteins ( RBPs ) have roles in the regulation of many post-transcriptional steps in gene expression , but relatively few RBPs have been systematically studied . We searched for the RNA targets of 40 proteins in the yeast Saccharomyces cerevisiae: a selective sample of the approximately 600 annotated and predicted RBPs , as well as several proteins not annotated as RBPs . At least 33 of these 40 proteins , including three of the four proteins that were not previously known or predicted to be RBPs , were reproducibly associated with specific sets of a few to several hundred RNAs . Remarkably , many of the RBPs we studied bound mRNAs whose protein products share identifiable functional or cytotopic features . We identified specific sequences or predicted structures significantly enriched in target mRNAs of 16 RBPs . These potential RNA-recognition elements were diverse in sequence , structure , and location: some were found predominantly in 3′-untranslated regions , others in 5′-untranslated regions , some in coding sequences , and many in two or more of these features . Although this study only examined a small fraction of the universe of yeast RBPs , 70% of the mRNA transcriptome had significant associations with at least one of these RBPs , and on average , each distinct yeast mRNA interacted with three of the RBPs , suggesting the potential for a rich , multidimensional network of regulation . These results strongly suggest that combinatorial binding of RBPs to specific recognition elements in mRNAs is a pervasive mechanism for multi-dimensional regulation of their post-transcriptional fate .
Much of the regulation of eukaryotic gene expression programs is still unaccounted for . Although these programs are subject to regulation at many steps , most investigation has focused on regulation of transcription . There are clues , however , that a significant portion of undiscovered regulation might be post-transcriptional , acting to regulate mRNA processing , localization , translation , and decay [1–5] . For example , systematic phylogenetic comparison among yeast and mammalian genomes sequences have revealed that untranslated regions of many mRNAs are under purifying selection , and thus presumably carrying information important for fitness [6–8] . Biological regulation can be achieved by controlling any of a large number of steps in the lives of RNA molecules . Alternative splicing of transcripts can enable a single gene to encode numerous protein products , greatly expanding its molecular complexity [9] . Even in organisms with few introns , such as Saccharomyces cerevisiae , splicing is subject to regulation [10 , 11] . Notable examples of regulated RNA localization include mRNA export from the nucleus to the cytoplasm , partitioning of mRNAs to the rough endoplasmic reticulum ( ER ) membrane for cotranslational export , and the precise subcellular localization of thousands of specific mRNAs [12] . In a recent survey of mRNA localization in developing Drosophila embryos , more than 70% of the roughly 3 , 000 mRNAs examined showed distinct patterns of subcellular localization [13] . Widespread regulation of translation rates is evident in several observations . In yeast , despite extensive regulation of transcription and mRNA decay , only about 70% of the observed variance in protein abundance is accounted for by variation in mRNA abundance [14 , 15] . When cells are moved from rich media to minimal media , the abundance of hundreds of proteins change , but mRNA abundance changes parallel changes in the abundance for only about half of the cognate proteins [16 , 17] . The abundance of each RNA is determined jointly by regulated transcription and regulated degradation . Widespread , transcript-specific regulation of mRNA decay is evident from the closely matched decay rates of mRNAs encoding functionally related proteins [18–21] , particularly evident in S . cerevisiae in sets of proteins that form stoichiometric complexes [19] . Increasing evidence points to extensive involvement of specific RNA-binding proteins ( RBPs ) in regulation of these post-transcriptional events [1–5] . Pioneering studies focusing on tens of predominantly nuclear mRNA RBPs ( so-called heterogeneous ribonucleoprotein [hnRNP] proteins ) , revealed that these proteins recognize specific features in mRNAs , bind at overlapping , but distinct , times during RNA processing , and differentially associate with subsets of nascent transcripts [22] . Steps in RNA processing in the nucleus are functionally and physically coupled , providing an opportunity for coordinated control [23] . Investigations of regulation acting on RNA have usually focused on a few model RNAs , leaving unanswered the extent to which mRNAs are coordinated and differentially regulated , and this regulatory landscape is still largely unexplored . Recent studies have systematically identified the suite of mRNAs associated with some individual RBPs . Several RBPs implicated in RNA processing and nuclear export in S . cerevisiae were found to associate with distinct sets of hundreds of functionally related mRNAs [24 , 25] . Five members of the Puf family of RBPs in S . cerevisiae were each found to associate with distinct , overlapping sets of 40–250 mRNAs [26] . The specific sets of mRNAs associated with each Puf protein were significantly enriched for mRNAs encoding functionally and cytotopically related proteins . For instance , most of the approximately 220 mRNAs associated with Puf3 are transcribed from nuclear genes and encode proteins localized to the mitochondrion ( p < 10−100 ) . Puf3 , Puf4 , and Puf5 each recognize specific sequences in the 3′-untranslated regions ( UTRs ) of their targets . These results and others , from studies of a few selected RBPs , may be just a glimpse of a much larger and richer post-transcriptional regulatory network , involving dozens to hundreds of RBPs and a cognate suite of recognition elements in their RNA targets ( e . g . , [22 , 24–40] ) . But does such a multidimensional post-transcriptional regulatory network exist ? To test this hypothesis and to extend and deepen our understanding of RBP–RNA interactions , we systematically searched for the RNA targets of a select sample of 40 out of the more than 500 known and predicted RBPs in S . cerevisiae .
We first developed a list of candidate RBPs based on annotations in the Saccharomyces Genome Database ( SGD ) ( http://www . yeastgenome . org ) , the Yeast Protein Database [41] , and the Munich Information Center for Protein Sequences database [42] and on literature searches . From the assembled list of 561 genes ( Table S1 ) , we chose a set of 36 with diverse RNA-binding domains and diverse functional annotations ( Table S2 and Text S1 ) . Because many known RBPs lack recognizable RNA-binding domains , we also included two metabolic enzymes whose homologs in other species are known to associate with RNA , and two proteins that were not , a priori , expected to bind RNA , but which we suspected might have post-transcriptional regulatory functions ( Table S2 ) . To identify RNAs associated with each putative RBP , C-terminal tandem affinity purification ( TAP ) -tagged proteins , expressed under control of their native promoters , were affinity purified from whole-cell extracts of cultures grown to mid-log phase in rich medium [14 , 26 , 43] . Extracts were incubated with immunoglobulin G ( IgG ) agarose beads , washed , and ribonuclear protein complexes were eluted by tobacco etch virus ( TEV ) protease treatment ( Text S2 ) . We performed two to four independent isolations with each tagged strain . As controls , we performed 13 immunoaffinity purifications ( IPs ) of untagged strains to identify and exclude potential false-positive RNA targets . We purified total RNA from the whole-cell extracts and TEV-purified fractions , reverse transcribed with an amino-allyl-dUTP/dNTP mix , coupled the purified cDNA to Cy3 and Cy5 dyes , respectively , mixed the two differentially labeled cDNA pools , and then hybridized them to DNA microarrays ( Dataset S1 ) . We identified RNAs specifically associated with each protein using the significance analysis of microarrays ( SAM ) algorithm [44] . Although it is not possible to perfectly distinguish targets from nontargets , and the best criterion for distinguishing targets from nontargets is unlikely to be the same for all proteins , for most proteins , we chose a 1% false discovery rate ( FDR ) as a criterion for identifying targets ( Datasets S2 and S3 ) . For many RBPs , the number of RNAs called significantly enriched has an inflection point near 1% FDR , suggesting that this threshold is a good balance between sensitivity and specificity , but undoubtedly our identification of specific RBP targets is not comprehensive . For two proteins in the survey ( Ssd1 and Khd1 ) , we used a more stringent 1% local FDR criterion [45] ( details in Materials and Methods; Datasets S2 and S3 ) . We also included mRNAs specifically associated with Puf1–5 from our previous work [26] , ( defined using a 1% local FDR ) , and previously identified She2 targets [32] . The 40 proteins in the survey ( and also Puf1–5 and She2 from our previous work [26 , 32] ) displayed diverse patterns of specificity with regard to the numbers and types of RNA targets and their enrichment profiles ( Figures 1 and S1 , and Text S3 ) . The number of confidently identified RNA targets varied widely among the proteins surveyed , ranging from fewer than ten ( Nce102 , Nrp1 , Idh1 , Rib2 , Nop13 , Bud27 , Rna15 , Pbp2 , Dhh1 , Upf1 , and Mex67 ) to more than a thousand ( Pab1 , Pub1 , Scp160 , Npl3 , Nrd1 , and Bfr1 ) ( Figure 1A ) . The two “negative controls , ” Nce102 and Bud27 , were each associated with specific RNAs . Nce102 was associated with eight distinct RNAs , whereas Bud27 was associated with two putative mRNA targets; interestingly , one of these putative targets ( RPA190 ) was reproducibly enriched more than 300-fold , and both targets were lost when immunopurifications were performed in the absence of Mg2+ ( unpublished data ) . Because neither Nce102 nor Bud27 was known or expected to associate with RNA , the RNAs identified as their targets may be spurious , but we cannot exclude the possibility that the RNA interactions we found for these two proteins are real and significant . Regardless , they provide a benchmark estimate of the number of RNA targets falsely identified for other RBPs . Aconitase ( Aco1 ) and glyceraldehyde-3 phosphate dehydrogenase ( Tdh3 ) , two metabolic enzymes whose human orthologs also function as RBPs [46 , 47] , but which were not previously known to be RBPs in yeast , associated with 38 and 155 RNAs , respectively , at 1% FDR , indicating that these enzymes are also RBPs in yeast . Fourteen of the proteins we surveyed specifically associated with RNAs other than mature mRNAs encoded by nuclear genes ( Figure S2 ) . Their specific targets included intron-containing transcripts ( Cbc2 , Msl5 , Npl3 , Hrb1 , Pab1 , and Pub1 ) , H/ACA box small nucleolar RNAs ( snoRNAs ) ( Cbf5 , Nrd1 , and Pub1 ) , C/D box snoRNAs ( Nop56 , Sof1 , Nab3 , Nrd1 , Pub1 , and Pab1 ) , and mitochondrial mRNAs ( Aco1 , Tdh3 , and Nab2 ) . Several of these proteins have previously been shown to be associated with specific classes of RNA ( Cbc2 , Msl5 , Npl3 , Cbf5 , Nrd1 , Nop56 , Sof1 , and Nab3 ) , and therefore provide de facto positive controls ( Table S2 and Text S4 ) . Aco1 , a TCA cycle enzyme [48] , which has recently been implicated in maintaining mitochondrial genome integrity [49] , selectively binds transcripts encoded by the mitochondrial genome ( p < 10−38 ) . Our results also suggest unexpected associations for several noncoding-RNA–binding proteins and suggest possible regulatory links between mRNA and noncoding RNA ( ncRNA ) processing ( Text S4 ) . However , the remainder of this report will focus mostly on mRNA targets . To explore the interrelationships among RBPs and their RNA targets , we organized RNAs ( Figure 1B , columns ) and RBPs ( Figure 1B , rows ) , respectively , by hierarchical clustering based on their patterns of mutual interactions , and visualized the results as a heat map representing the confidence of an RNA–RBP interaction with a black ( >10% FDR ) to yellow ( 0% FDR ) scale . For the most part , each RBP had a unique profile of enrichment , with a few notable exceptions , including Scp160/Bfr1 and Nrd1/Nab3 , which are pairs of proteins that act together in stable stoichiometric complexes [50 , 51] and were correspondingly associated with similar sets of mRNAs . Altogether , we identified more than 12 , 000 mRNA–RBP interactions ( at a 1% FDR ) , an average of at least 2 . 8 RBPs interacting with each of 4 , 300 distinct mRNAs; 31 proteins ( including Puf1–5 and She2 ) reproducibly bound at least ten mRNAs ( at a 1% FDR ) . Most mRNAs were bound by multiple RBPs ( Figure 1C , black bars ) ; 628 mRNAs were bound by five or more of this set of 31 RBPs; intriguingly , a disproportionate fraction of the mRNAs with the greatest number of identified interactions with this set of RBPs encode proteins localized to the cell wall ( 31 , p < 10−4 ) . About 75% ( ∼9 , 000 ) of the mRNA–RBP interactions identified in this survey were accounted for by the nine proteins that targeted more than 500 mRNAs each , ( Figure 1C , grey bars ) . Our conservative approach to target identification , emphasizing specificity over sensitivity , probably underestimates the number of targets of these broad-specificity RBPs; some of these proteins , such as Scp160 and Pab1 , probably bind most or all mRNAs ( Figure S1 and Text S3 ) . The specificity and regulatory contributions of these “general” RBPs are still poorly understood . Regulatory proteins , including both transcription factors and RBPs , typically regulate sets of targets that share identifiable functional relationships ( e . g . , [26–29 , 32 , 35 , 52–60] ) . As a first step toward identifying relationships among RNAs bound by specific RBPs , we searched for gene ontology ( GO ) terms [61] that were significantly enriched among the targets of each RBP . Twenty-five of the RBPs in this survey were consistently associated with at least ten mRNAs; 13 of these sets of RNA targets specific to an RBP were significantly enriched for at least one “cellular component” GO term ( Figure 2A and Table S3 ) , representing a shared subcellular localization or in some instances a protein complex , and 13 of these RBP-specific target sets were significantly enriched for at least one “biological process” GO term ( Figure 2B and Table S3 ) . Diverse subcellular loci and biological processes were represented among the annotations enriched in the sets of RNA targets of these 15 RBPs ( as well as the five Puf proteins and She2 ) , including nearly all major subcellular compartments . Some subcellular sites and biological processes were found as shared attributes of the RNA targets associated with an unexpectedly large fraction of the RBPs in this study , perhaps highlighting processes or systems in which post-transcriptional regulation plays an especially important role . For instance , six RBPs ( Pub1 , Khd1 , Nab6 , Ssd1 , Ypl184c , and Scp160 ) were specifically associated with mRNAs encoding cell wall proteins; six ( Pub1 , Puf1 , Puf2 , Khd1 , Ypl184c , and Scp160 ) were specifically associated with mRNAs encoding plasma membrane proteins; five ( Puf3 , Nsr1 , Pab1 , Npl3 , and Nrd1 ) were significantly associated with mRNAs encoding subunits of mitochondrial ribosome; and four ( Scp160 , Bfr1 , Puf4 , and Gbp2 ) were specifically associated with mRNAs encoding proteins localized to the nucleolus and involved in RNA processing and ribosome biogenesis . For many RBPs , several distinct subcellular components or biological processes were overrepresented in the functional annotations of the associated transcripts; these subcellular loci or processes were often functionally linked . For example , RNAs associated with Ssd1 were enriched for transcripts encoding cell wall and bud proteins , whereas Gbp2-associated RNAs were enriched for transcripts encoding nuclear proteins with roles in ribosome biogenesis or chromatin remodeling . In many instances , the functional themes significantly overrepresented among the RNA targets of an RBP are congruent with previously published work on that RBP , such as phenotypes associated with mutation of altered expression ( Table S2 ) . A few examples are described in subsequent sections . Although some appear to bind to most or all mRNAs ( Figure S2 and Text S3 ) , the nine RBPs that bind large ( >500 ) sets of mRNAs display several distinct enrichment profiles ( Figure 1B ) , with correspondingly different GO annotations overrepresented among the most highly enriched mRNAs ( Figure 2 ) . In addition , for each of these nine RBPs , immunoaffinity enrichment of mRNAs with the RBP was significantly correlated with either ribosome occupancy [62] , abundance [19] , half-life [19] , 3′-UTR length [63] , 5′-UTR length [63] , mRNA length [63] , coding sequence length , or in some cases , with more than one of these features ( Figure S3 ) . Quantitative differences in the enrichment of mRNAs in association with a given RBP could result from the number or affinity of the RBP molecules bound or differences in the fraction of its lifespan that an individual mRNA spends at the specific stage during which a particular RBP plays a role ( Text S5 ) . Pab1 provides a simple and useful example of the possible functional significance of the differential enrichment; immunoaffinity enrichment of mRNAs associated with Pab1 was correlated with ribosome occupancy ( Pearson correlation = 0 . 35 ) . Pab1 is the major poly ( A ) binding protein in both the nucleus and cytoplasm [64] . In the cytoplasm , Pab1 binds to the poly ( A ) tails of mRNAs and interacts with eIF4-G to promote translation initiation [65] . Because longer poly ( A ) tails have been reported to increase translation efficiency [66] , a possible interpretation of these results is that the observed enrichment could reflect the number of Pab1 proteins bound per mRNA and thus the length of the poly ( A ) tail [39] . In contrast , immunoaffinity enrichment with Khd1 was negatively correlated with ribosome occupancy ( r = −0 . 26 ) . Khd1 is implicated in repressing translation of ASH1 mRNA during the transport of the mRNA to the bud tip [67] . The negative correlation with global ribosome occupancy and the large number of mRNAs associated with Khd1 suggest that Khd1 may similarly repress translation initiation of hundreds to thousands of mRNAs , perhaps during their transport to specific cellular loci . Many RBPs associate with mRNAs at a particular stage in their lives [2] . For the approximately 270 intron-containing genes , the relative enrichment of introns ( i . e . , unspliced pre-mRNAs and possibly uncleaved excised introns ) versus exons ( i . e . , mature mRNAs and pre-mRNAs ) should reveal whether the RBP is bound specifically to intron-containing transcripts , mature mRNAs , or both , and thus indicate when and where the RBP associates with its target RNAs . Linking these data to functional information on the RBP could then provide insights into timing and duration of specific stages in the lives of mRNAs . To test this idea , we compared the enrichment of intron and exon sequences in association with RBPs . For the approximately 120 intron/exon probe pairs for which our data were most consistently reliable , the relative enrichment profiles vary greatly among RBPs ( Figure 3 and Text S6 ) . For example , Cbc2 ( a component of the heterodimeric nuclear cap-binding protein ) and Pab1 were preferentially associated with both intron-containing transcripts and mature mRNAs derived from intron-containing transcripts ( Figure 3 ) . Cbc2 was strongly associated with intron-containing transcripts ( mean enrichment of intronic sequences = 6 . 8 ) , and also , but to a considerably lesser extent , with exon sequences from intron-derived transcripts ( mean enrichment of exonic sequences = 1 . 5 ) . These results are consistent with Cbc2 binding during transcription , prior to splicing , and being displaced shortly after the mature mRNA reaches the cytoplasm [68 , 69] . The enrichment of intron-related transcripts and the paucity of significantly enriched mature mRNAs suggest that most mRNAs spend only a very small fraction of their lives in the nucleus . That Pab1 , the major poly ( A ) binding protein , associated with intron-containing transcripts ( mean enrichment of intronic sequences = 1 . 5 ) , as well as sequences from exons ( mean enrichment of exonic sequences = 3 . 9 ) , is consistent with most splicing occurring after poly ( A ) tail addition [70] . The RBPs we analyzed bound overlapping sets of mRNAs , and many individual mRNAs were bound by more than one RBP ( Figure 1B and 1C ) . This network of interactions could support a robust and multidimensional regulatory program . To explore the relationships among the groups of RNAs bound by different RBPs , we determined the extent to which the overlaps between targets for each RBP pair differed from what would be expected by chance . The significance values from this analysis were used as a metric of similarity for hierarchical clustering to identify pairs and sets of RBPs with similar patterns of shared targets . The results are presented in Figure 4A as a heat map , in which the similarity between the target sets of each pair of RBPs is shown on a blue ( significantly fewer shared targets than expected , p = 10−25 ) to white ( p > 0 . 001 ) to red ( significantly more shared targets than expected , p = 10−25 ) scale . At a p-value threshold of 0 . 001 , 69 of 465 RBP pairs shared significantly more mRNA targets than expected by chance , whereas 11 RBP pairs shared significantly fewer mRNA targets than expected by chance . Several of the most significantly overlapping target sets belong to sets of RBPs that are known to physically interact , such as Scp160 and Bfr1 [50] , Nrd1 and Nab3 [51] , Nrd1/Nab3 and Npl3 [71] , and Nrd1/Nab3 and Pab1 [72] . To further explore the interrelationships among RBPs and their mRNA targets , we used a supervised method to identify smaller subsets of mRNAs that shared interactions with several RBPs . We did this by selecting mRNAs bound by a common set of RBPs whose targets , in turn , were enriched for common GO terms ( Figure 2 ) . The group of mRNAs , defined by interactions with at least four of a set of six RBPs ( Pub1 , Khd1 , Nab6 , Ssd1 , Ypl184c , and Scp160 ) , includes a significant excess of mRNAs encoding proteins localized to the cell wall ( Figure 4B ) ; indeed , 23 of the 78 mRNAs in this cluster encode cell-wall proteins ( p < 10−19 ) . This group also contains mRNAs that encode proteins that are secreted ( 5 ) , localized to sites of polarized growth ( 4 ) , or localized to the ER ( 14 ) . It is important to recognize that the unifying theme in this group is not narrowly restricted to simple functions in cell-wall metabolism—many mRNAs in this group encode proteins with diverse roles in regulation of cell-wall metabolism . Fifteen mRNAs encode proteins involved in post-transcriptional regulation , including SSD1 , DHH1 , and PUF5 , which are genetically implicated in cell-wall biogenesis and maintenance [73 , 74] , and NGR1 and WHI3 , which are involved in control of cell growth [75–77] . Fourteen of these mRNAs encode proteins involved in transcriptional control , including SFL1 , which is implicated in cell-wall assembly [78] , and NDD1 , YOX1 , and NRM1 , which are involved in cell-cycle control [79–81] . Seven mRNAs encode signal transduction proteins , including MFA2 , CLN2 , GIC2 , WSC2 , and MSB2 , which are implicated in cell-wall growth or cell-cycle regulation [82–88] . We identified candidates for the sequence elements that mediate regulatory interactions with specific RBPs using two related computational methods: “finding informative regulatory elements” ( FIRE ) , which searches for motifs with informative patterns of enrichment [89] , and a newly developed method , “relative filtering by nucleotide enrichment” ( REFINE ) . In brief , REFINE identifies all hexamers that are significantly enriched in putative 5′- and 3′-UTR regions of targets over nontargets , filters out regions of target sequences that are relatively devoid of such hexamers , and then applies the “multiple expectation maximization for motif elicitation” ( MEME ) motif-finding algorithm [90] . A full description of the REFINE methodology and more detailed analyses of predicted motif sequences will be published separately ( D . P . Riordan , D . Herschlag , and P . O . Brown , unpublished data ) . Herein , we combined the results from these two approaches . Using stringent statistical criteria based on randomized simulations ( details in Materials and Methods ) , we identified a total of 60 candidate RNA regulatory motifs significantly associated with 21 different RBPs; 35 motifs ( for 21 RBPs ) were predicted by REFINE , and 25 motifs ( for 13 RBPs ) were predicted by FIRE ( Table S4 ) . Since the same motifs were often predicted by both programs for the same RBP or for different RBPs with significantly overlapping target sets , we manually grouped motifs with similar consensus sequences and origins into classes ( Table S4 ) . We then included only the most significant motif from each class and for each RBP , resulting in a set of 14 nonredundant RNA motifs predicted with high confidence ( Figure 5 ) . We also evaluated the predicted RNA motifs by testing whether motif sites occurring in targets were more likely to be conserved than sites in nontargets , and whether they exhibited a forward strand bias by testing for significant enrichment of the reverse complementary motif in RBP targets ( Table S4 ) . The motifs we identified for Puf3 , Puf4 , Puf5 , Pub1 , Nab2 , Nrd1 , and Nab3 match previously described binding sites for the corresponding RBPs , validating our approach and suggesting that many of the RBP–RNA interactions we measured are likely to be directly mediated by these elements ( Text S7 ) . Interestingly , the inferred recognition element for Nrd1 , Nrd1–1 ( UUCUUGUW ) , contains both an exact match to the reported Nrd1 binding site consensus “UCUU” and a partial match to the reported Nab3 recognition site consensus “GUAR” [91 , 92] . As Nrd1 and Nab3 are known to act as a complex to control transcriptional termination of nonpolyadenylated RNAs [93] , and a nearly identical motif was identified in Nab3 targets ( Table S4 ) , it is possible that these motifs represent a favored orientation of adjacent Nrd1 and Nab3 RNA elements that facilitates specific binding of the Nrd1–Nab3 complex . The most significant novel motif we identified , Puf2–1 ( UAAUAAUUW ) , is enriched in the 3′-UTRs and coding sequences of Puf2 targets and demonstrates significant conservation and a forward strand bias ( Figure 5 ) . This motif is similar to a motif identified for the paralogous RBP Puf1 , which associates with a subset of the Puf2 target mRNAs ( Table S4 ) . The next most significant novel motif , Ssd1–1 ( AKUCAUUCCUU ) , is highly enriched in the 5′-UTRs of Ssd1 targets ( Figure 5 ) . Although its presence upstream of the coding sequences of Ssd1 target genes would also be consistent with a role as a transcription factor binding site , its tendency to occur within the annotated 5′-UTRs of targets ( 63% targets versus 19% nontargets , p < 10−6 ) [94] , its dramatic enrichment in targets , and its forward strand bias suggest that this RNA motif is recognized by Ssd1 . A selective sample of 11 mRNAs provides an unfinished , but revealing , picture of the organization of the information that specifies interactions with , and perhaps regulation by , specific RBPs examined in this study ( Figure 6 ) . For each mRNA , the location of high-confidence RNA recognition elements for RBPs that interact with the mRNA are indicated , while RBPs that interact with the mRNA , but whose binding site is uncertain , are shown to the right of the mRNA . The relative lengths of the 5′-UTR , coding sequence , and 3′-UTR are drawn to scale , and the translation start and stop codons are depicted with the corresponding “traffic signal . ” Each of these mRNAs has specific interactions with overlapping , but distinct , subsets of RBPs in the study . The putative binding patterns of specific RBPs , with respect to the number and locations of sites , vary considerably among the mRNAs , which may have important functional consequences . The first five mRNAs ( SUN4 , DSE2 , CTS1 , SCW4 , and EGT2 ) encode cell-wall enzymes ( Figure 6A–6E ) . Each of these mRNAs associated with five to nine RBPs in this study , including all five with Pub1 , Khd1 , and Ypl184c , four with Ssd1 ( SUN4 , DSE2 , CTS1 , and SCW4 ) , three with Scp160 ( CTS1 , SCW4 , and EGT2 ) , and two with Nab6 ( CTS1 and SCW4 ) and Nrd1 ( DSE2 and EGT2 ) . In addition to these overlapping interactions , most of these mRNAs associated with a unique set of additional RBPs; for instance , SUN4 contains two Puf5-binding sites in its 3′-UTR and EGT2 contains eight She2-binding sites in its coding sequence . CLN2 encodes a G1 cyclin and associated with many of the same RBPs as SUN4 , DSE2 , CTS1 , SCW4 , and EGT2 ( Figure 6F ) . PUF2 associated with several RBPs , including its cognate protein , which is common among RBPs in this study ( Text S8 ) ; there are 12 Puf2-binding sites in its coding sequence ( Figure 6G ) . PMA1 associated with a similar set of RBPs as PUF2 , including Pub1 and Puf2 , but the locations and numbers of binding sites for these RBPs are very different in the two mRNAs ( Figure 6H ) . The putative binding sites for Puf4 and Puf5 in the 3′-UTR of HHT1 partially overlap , suggesting these RBPs may compete for binding to this mRNA ( Figure 6J ) . These diagrams represent only a partial picture of the RBP interactions with these mRNAs; the mRNA targets have only been defined for a small fraction of all yeast RBPs , and the sequence elements that specify many of the interactions we have identified are not yet known . For many RBPs , our computational method did not identify any sequence motifs with statistically significant enrichment , the motifs identified significantly overlapped those associated with other RBP target sets , or the motif did not match previously reported binding preferences ( Table S4 and Text S7 ) . The large degree of motif coenrichment observed in our analysis is consistent with combinatorial regulation by a highly interconnected regulatory network and represents an important limitation of computational regulatory element identification . It is likely that some of the RBPs for which we failed to predict sequence motifs recognize RNA structural elements or features primarily present in coding sequences , which are difficult to detect with current methods for RNA motif prediction , because they are not suited to modeling structural features or handling the significant confounding sequence biases in coding sequences . Vts1 illustrates some of the limitations of current RNA motif prediction methods . Vts1 is known to bind to a structural RNA motif called the Smaug recognition element ( SRE ) , which consists of a short hairpin with the loop consensus sequence CNGGN ( 0–1 ) [95] . SRE sites are indeed significantly enriched in the coding sequences of Vts1 targets ( 65% targets versus 36% nontargets , p < 10−7 ) in agreement with previous results [96] , suggesting that SRE elements are directly responsible for these interactions in vivo . However , neither REFINE nor FIRE succeeded in identifying the SRE . Instead , both programs identified a motif , Vts1–1 ( UKWCGRGGN ) , which is indeed enriched in the 3′-UTRs of Vts1 targets but is unrelated to the SRE ( Table S4 ) . We suspect that the Vts1–1 motif may represent a binding site for an unknown factor that regulates a set of mRNAs that overlaps extensively with the targets of Vts1 . It is likely that direct high-resolution mapping of in vivo RBP binding sites and systematic in vitro characterization of binding preferences of RBPs will overcome some of the limitations in current methods for RNA motif identification [97 , 98] . The functional and cytotopic themes represented among the specific targets of each RBP have obvious implications for their possible regulatory roles , which can be integrated with previously reported information to derive further insights , and generate new hypotheses , as illustrated here for Ssd1 and Ypl184c ( see Text S9 for descriptions of Khd1 and Gbp2 ) . Ssd1 is a large ( 140 kDa ) , ribonuclease-II domain–containing , predominantly cytoplasmic protein [99] , genetically implicated in cell-wall biogenesis and function: mutant phenotypes include increased sensitivity to osmotic stress and caffeine , altered composition and structure of the cell wall , defects in germination and sporulation , premature aging , and pathogenicity [73 , 74 , 100–103] . Ssd1 physically and genetically interacts with numerous signaling proteins , many of which are genetically implicated in cell-wall function [71 , 102 , 104 , 105] . Ssd1 binds to the C-terminal domain of RNA polymerase II in vitro [106] . Of the 52 annotated mRNAs associated with Ssd1 , 16 encode proteins localized to the cell wall ( p < 10−15 ) , and 11 encode proteins localized to the bud ( p < 10−5 ) . The proteins encoded by the Ssd1-associated transcripts have diverse functional and structural roles related to cell-wall biosynthesis , or remodeling and its regulation , cell-cycle progression , and protein trafficking . Ssd1 also appears to bind its own transcript ( Text S8 ) . For both of the Ssd1 mRNA targets encoded by intron-containing genes ( PUF5 and ECM33 ) , the intron-containing primary transcripts are also enriched by Ssd1 IP , suggesting that Ssd1 binds its RNA targets in the nucleus , perhaps while they are being transcribed . A putative RNA-recognition motif is significantly enriched in the 5′-UTRs of Ssd1 targets ( Figure 5 ) . The numbers and positions of this motif in Ssd1-bound RNAs vary widely among its targets ( Figure 6A–6D and 6F ) . These data lead us to speculate that Ssd1 binds its targets cotranscriptionally by recognizing a specific RNA motif and prevents their translation initiation until these mRNAs reach specific locations in the cell , such as the ER membrane , bud , or sites of cell-wall biosynthesis . The multiple phosphorylation sites on Ssd1 could regulate the localization , binding , and release of its RNA targets . Although Ssd1 is a ribonuclease-II domain–containing protein , it has no discernable nuclease activity [99] . Given that Ssd1 does not contain any other known RNA-binding domains , we suggest that the ribonuclease-II domain may have evolved into a sequence-specific RNA-binding domain in this protein family . Ypl184c is a largely uncharacterized , predominantly cytoplasmic protein that contains three RNA recognition motifs ( RRMs ) . Of the three proteins that have been found to physically interact with Ypl184c , two are among the other RBPs included in this survey: Pab1 and Nab6 [71] . A disproportionate fraction of the 321 annotated mRNAs we found to associate with Ypl184c encode proteins localized to the cell wall ( 38 , p < 10−23 ) , ER ( 50 , p < 10−5 ) , plasma membrane ( 32 , p < 10−3 ) , or extracellular milieu ( 8 , p < 10−3 ) . Transcripts encoding components of several protein complexes were associated with Ypl184c , including three of five components of the Cdc28 complex ( CLB2 , CLN3 , and CLN2 ) for which we obtained high-quality measurements , three of three components of the plasma membrane H+ ATPase ( PMP1 , PMP2 , and PMA1 ) for which we obtained high-quality measurements , and four of nine components of the oligosaccharyltransferase complex ( OST4 , SWP1 , OST3 , and OST5 ) [107] . Components of these complexes that were not defined as targets of Ypl184c ( at a stringent 1% FDR ) were nevertheless more likely to be overrepresented in Ypl184c IPs than expected by chance , suggesting that Ypl184c may actually associate with the mRNAs encoding most or all members of these complexes . Ypl184c associated with many mRNAs that exhibit unusual modes of translation regulation . Ypl184c bound all five of the mRNAs that have experimentally confirmed short upstream open reading frames ( uORFs ) ( GCN4 , CPA1 , LEU4 , SCH9 , and SCO1 ) [108–115] in their 5′-UTRs and for which we obtained high-quality measurements; uORFs have been shown to regulate the translation of the downstream coding sequence and the stability of the mRNA [116] . Ypl184c associated with all five of the S . cerevisiae mRNAs that have been shown to have internal ribosome entry sites ( IRES ) ( HAP4 , YMR181C , GPR1 , NCE102 , and GIC1 ) in their 5′-UTRs [117 , 118] for which we obtained high-quality measurements; these IRESs enable cap-independent translation , often in response to environmental stresses [119] . Ypl184c also bound the unspliced HAC1 transcript , which associates with the cytosolic side of the ER membrane and is not efficiently translated until it is spliced by IRE1 as part of the unfolded protein response pathway [120 , 121] . Given Ypl184c's association with Pab1 and its striking association with sets of mRNAs that are known to be subject to extensive translational regulation , we speculate that Ypl184c regulates translation . The sequence motifs that we found to be significantly enriched in the mRNA targets of Ypl184c closely match the ones we found for Pub1 ( Table S4 ) . Indeed , the RNA target sets of these two proteins overlap significantly ( Figures 1B and 4A ) . Given the absence of evidence for direct interactions between Ypl184c and Pub1 , perhaps they compete for binding to overlapping groups of mRNAs . We have named YPL184C , post-transcriptional regulator of 69 kDa ( PTR69 ) .
A large body of work has given us a general picture of the relationship between the several hundred transcription factors and thousands of genes in yeast ( e . g . , [26–29 , 32 , 35 , 52–60] ) . Among the key features of transcriptional regulation are that: ( 1 ) individual transcription factors characteristically regulate sets of genes with related biological roles , ( 2 ) transcription factors are recruited to the specific genes they regulate by binding to specific sequences in the vicinity of those genes , and ( 3 ) combinatorial regulation of individual genes by two or more distinct transcription factors provides multidimensional control and precision to their regulation . Our systematic identification of RNAs associated with each of 46 proteins in yeast suggests that a system that shares these three key features , likely involving dozens to hundreds of RBPs , may regulate the post-transcriptional fate of most or all RNAs in the yeast cell . This glimpse into the landscape of RNA–protein interactions has provided tantalizing clues to its organization and role . The mRNA targets of most of the RBPs in the survey encoded sets of proteins that were significantly associated with one or several related subcellular sites or biological processes ( Figure 2 and Table S3 ) . Although the regulatory roles and molecular mechanisms of most of these interactions remain to be elucidated , it seems unlikely that they have a purely decorative function . The selective binding of RBPs to sets of mRNAs that encode functionally and cytotopically related proteins provides strong evidence for widespread regulation at the post-transcriptional level . The functional relevance of these interactions is further supported by their relationships to phenotypes associated with mutation or altered expression of the RBP ( Table S2 ) . Many RBPs , including those examined in our survey , have mutant phenotypes only in specific physiological and developmental programs , and they have diverse gene expression patterns ( http://www . yeastgenome . org ) . Thus , the regulatory program mediated by RBPs may be reorganized in response to specific physiological and developmental cues . The striking tendency of individual RBPs to bind to sets of mRNAs whose protein products are similarly localized in the cell hints at an important role for RBPs in establishing and maintaining spatial organization in the cell , perhaps through facilitating localized protein production and mRNA decay [13 , 32 , 122–131] . The cellular structures that were most often overrepresented among the mRNA targets of many RBPs were the cell wall , plasma membrane , and ER . Thus , in addition to the familiar role of the peptide signal sequence in mediating ER-localized translation [12] , RBPs may have important roles in RNA partitioning between the cytoplasm and ER , and perhaps in localization to specific sites in the periphery of the cell , such as sites of cell-wall biogenesis , bud development , and endocytosis [32 , 132–135] . Two of the RBPs whose targets disproportionably encode proteins localized to the cell periphery , She2 and Khd1 , have been shown to be involved in trafficking some of their mRNA targets to the bud tip during the G2/M phase of the cell cycle [32 , 67 , 136] . The particularly strong overrepresentation of RBPs that associate with mRNAs encoding cell-wall components may reflect the need for extensive multilayered regulation of the location and timing of assembly and remodeling of this dynamic subcellular structure . Identification of the information that specifies mRNA–RBP interactions is still in its earliest stages . The sequence motifs overrepresented in RBP targets , identified with the recently developed FIRE and novel REFINE methodologies , are diverse in design and location ( Figures 5 and 6 ) . Many of these RBPs recognized short linear sequences in the 3′-UTRs , 5′-UTRs , or coding sequences , or two or more of these regions . For about half of the RBPs , however , we were unable to find a sequence motif enriched among its RNA targets . Some of these RBPs may recognize structural elements . In support of this idea , we found the SRE hairpin loop , previously recognized as important for specific recognition of RNA by Vts1 [95] , significantly enriched in coding sequences of Vts1 targets . Another protein in this survey , She2 , is believed to recognize a three-dimensional structure in its targets [137 , 138] . We found promoter elements that likely specify transcription factor interactions enriched in the upstream regions of several RBP target sets , e . g . , Gbp2 ( Table S4 ) . It is possible these promoter elements play an indirect role in specifying RBP interactions , perhaps by cotranscriptional recruitment of an RBP to mRNA targets via interactions with specific transcription-associated factors [22 , 23 , 139] . Identification of the large amount of still-undiscovered RNA regulatory information is an essential step in uncovering the specific regulatory program of each gene . We identified over 12 , 000 mRNA–RBP interactions with high confidence . Most mRNAs in the yeast transcriptome associated with at least one of the RBPs in our survey and many associated with multiple RBPs . Some of the RBPs in the survey appear to interact with most or all mRNAs at some point in their lifecycle ( Figure S1 and Text S3 ) . Naively extrapolating from our results to the estimated 600 RBPs in Saccharomyces suggests that each mRNA might interact with a dozen or more different RBPs , on average , during its lifetime . This extrapolation is highly speculative; the sample of RBPs that we investigated is biased towards RBPs that we suspected might have a regulatory function; we do not have a good estimate of the number of regulatory RBPs that bind discrete sets of mRNAs in the manner analogous to specific transcription factors; given that three of the four proteins in this survey that were not annotated as RBPs nevertheless gave reproducible interactions with specific sets of mRNAs ( Bud27 , Aco1 , and Tdh3 ) , the number of potential noncanonical , unannotated RBPs with regulatory roles may be large , perhaps even in the hundreds [140–144] . There is no reason to believe the system we have described is peculiar to yeast . Extensive post-transcriptional regulation by combinatorial binding of a large and diverse set of specific RBPs is likely to be a general feature of regulation in eukaryotes . Indeed , several lines of evidence suggest an even greater genomic investment in post-transcriptional regulation in humans ( and other metazoans ) ; the number and diversity of RBPs encoded by the human genome seems to far exceed that of yeast [145] , untranslated regions of mRNAs are much longer in humans ( ∼1 , 300 bases on average ) than in yeast ( ∼300 bases on average ) and appear to contain much more regulatory information [6 , 146 , 147] , and the architecture of animal cells is far more diverse and complex than that of the yeast cell , with a correspondingly greater potential role for specific RNA localization [13 , 130 , 148–151] . This work has provided a glimpse of a network of RBP–mRNA interactions that is likely to play an important , but still largely undiscovered , role in biological regulation . The genes and cis-regulatory elements implicated in this process represent a substantial fraction of the genome's investment in regulation , yet the specific details and molecular mechanisms of this network of RBP–mRNA interactions are still largely terra incognita—and fertile ground for further exploration and discovery .
We carried out immunopurifications of specific proteins , together with the associated RNAs , using specific strains expressing a TAP-tagged derivative of each selected protein ( Open Biosystems Cat# YSC1177-OB ) , essentially as described in Gerber et al . [26] . After growing 1L cultures to an optical density at 600 nm ( OD600 ) of 0 . 6–0 . 9 in YPAD , we harvested cells by centrifugation , chilled the cell pellets on ice , washed them twice with 25 ml of ice cold buffer A ( 20 mM Tris–HCl [pH 8 . 0] , 140 mM KCl , 1 . 8 mM MgCl2 , 0 . 1% Nonidet P-40 , 0 . 02 mg/ml heparin ) , then froze them in LN2 and stored them at −80 °C . In a few instances , we proceeded to lyse the pelleted cells immediately without freezing . To lyse the cells , we first thawed the cell suspension at 4 °C , added 5 ml of buffer B ( buffer A plus 0 . 5 mM DTT , 1 mM PMSF , 1 μg/ml leupeptin , 1 μg/ml pepstatin , 20 U/ml DNase I [Stratagene Cat# 600032] , 50 U/ml Superasin [Ambion Cat# AM2696] , and 0 . 2 mg/ml heparin ) , and then mechanically lysed the cells by vortexing in the presence of glass beads . We removed the beads by centrifugation at 1 , 000g for 5 min , then clarified the extracts by centrifuging them twice at 7 , 000g for 5 min each . We adjusted the volume of the extract to 5 ml with buffer B , removed a 100-μl aliquot for reference RNA isolation , and then incubated the remaining 4 . 9 ml with 400 μl of 50% ( v/v ) suspension of IgG-agarose beads ( Sigma Cat# A2909 ) in Buffer A with gentle rotation for 2 h . We washed the beads once with 5 ml of buffer B for 15 min , and three times with 12 ml of buffer C ( 20 mM Tris-HCl [pH 8 . 0] , 140 mM KCl , 1 . 8 mM MgCl2 , 0 . 5 mM DTT , 0 . 01% NP-40 , 15 U/ml Superasin , 1 μg/ml pepstatin , 1 μg/ml leupeptin , 1 mM PMSF ) for 15 min with gentle rotation . We pelleted the beads by centrifugation for 5 min at 60g in a table-top centrifuge . We then transferred the beads to 1 . 2-ml micro-spin columns ( BioRad Cat# 732-6204 ) , centrifuged them briefly to pellet the beads , removed buffer C , and then added 1 volume of buffer C . We cleaved TAP-tagged proteins by incubation with 80 U acTEV protease ( Invitrogen Cat# 12575023 ) or an equivalent amount of purified TEV [152] for 2 h at 15 °C . We collected the eluent by centrifugation into 2-ml tubes . We isolated reference RNA using RNeasy Mini Kit ( Qiagen Cat# 74106 ) , while we isolated RNA from the eluate by extraction with Phenol/Chloroform/Isoamyl Alcohol , 25:24:1 ( Invitrogen Cat# 15593031 ) twice , and chloroform once , followed by ethanol precipitation with 15 μg of Glycoblue ( Ambion Cat# AM9515 ) as carrier . Starting with the Operon AROS 1 . 1 oligo set , which contains long oligonucleotides for almost all annotated S . cerevisiae nuclear and mitochondrial coding sequences , we added 3 , 072 additional probes designed to detect annotated noncoding RNAs , ribosomal RNA precursors , introns , exon-intron and exon-exon junctions , other sequences predicted to be expressed , additional probes for genes with high cross-hybridization potential , and hundreds of controls for array quality measurements and normalization . Details of oligonucleotide selection and probe sequences are available from the Operon Web site ( https://www . operon . com/; S . cerevisiae YBOX V1 . 0 ) . Detailed methods for microarray experiments are available at the Brown lab Web site ( http://rd . plos . org/pbio . 0060255 ) . For oligonucleotide microarrays , we resuspended oligonucleotides in 3× SSC ( 1× SSC = 150 mM NaCl , 15 mM sodium citrate [pH 7 . 0] ) at a final concentration of 25 μM and printed oligonucleotides on poly-lysine glass ( Erie Scientific Cat# C41–5870-M20 ) ( http://rd . plos . org/pbio . 0060255a ) . We printed each oligonucleotide twice per array . For most arrays , the second print was in reverse orientation to the first print , such that oligonucleotide pairs were printed with different pins and thus located in different sectors of the array . Prior to hybridization , the oligonucleotides were crosslinked to the poly-lysine–coated surface with 65 mJ of UV irradiation . Slides were then incubated in a 500-ml solution containing 3× SSX and 0 . 2% SDS for 5 min at 50 °C . Slides were washed for 2 min in a glass chamber containing 400 ml of water , dunked in a glass chamber containing 400 ml of 95% ethanol for 15 s , and then dried by centrifugation . Free poly-lysine groups were then succinylated by incubation with 5 . 5 g of succinic anhydride that was dissolved in 350 ml of anhydrous 1-methyl , 2-pyrolidoinone ( Sigma Cat# 328634 ) and 15 ml of 1 M sodium borate ( pH 8 . 0 ) for 20 min [53] . Slides were washed for 2 min in a glass chamber containing 400 ml of room temperature water , dunked in a glass chamber containing 400 ml of 95% ethanol for 15 s , and then dried by centrifugation . cDNA microarrays containing long double-stranded DNA ( dsDNA ) from PCR reactions were prepared as previously described [53] . A total of 3 μg of reference RNA from extract and up to 3 μg ( or 50% ) of affinity-purified RNA were reverse transcribed with Superscript II ( Invitrogen Cat# 18064–014 ) in the presence of 5- ( 3-aminoallyl ) -dUTP ( Ambion Cat# AM8439 ) and natural dNTPs ( GE Healthcare Life Sciences Cat# US77212 ) with a 1:1 mixture of N9 and dT20V primers ( Invitrogen ) . Subsequently , amino-allyl–containing cDNAs were covalently linked to Cy3 and Cy5 NHS-monoesters ( GE Healthcare Life Sciences Cat# RPN5661 ) . Dye-labeled DNA was diluted in a 20–40-μl solution containing 3× SSC , 25 mM Hepes-NaOH ( pH 7 . 0 ) , 20 μg of poly ( A ) RNA ( Sigma cat # P4303 ) , and 0 . 3% SDS . The sample was incubated at 95 °C for 2 min , spun at 14 , 000 rpm for 10 min in a microcentrifuge , and then hybridized at 65 °C for 12–16 h . For most oligonucleotide microarray experiments , we hybridized microarrays inside sealed chambers in a water bath using the M-series lifterslip to contain the probe on the microarray ( Erie Scientific Cat # 22x60I-M-5522 ) . For some oligonucleotide microarray experiments , we hybridized microarrays using the MAUI hybridization system ( BioMicro ) , which promotes active mixing during hybridization . We hybridized cDNA microarrays inside sealed chambers in a water bath using a coverslip to contain the probe on the microarray . Following hybridization , microarrays were washed in a series of four solutions containing 400 ml of 2× SSC with 0 . 05% SDS , 2× SSC , 1× SSC , and 0 . 2× SSC , respectively . The first wash was performed for 5 min at 65 °C . The subsequent washes were performed at room temperature for 2 min each . Following the last wash , the microarrays were dried by centrifugation in a low-ozone environment ( <5 ppb ) to prevent destruction of Cy dyes [153 , 154] . Once dry , the microarrays were kept in a low-ozone environment during storage and scanning ( see http://rd . plos . org/pbio . 0060255 ) . Microarrays were scanned using either AxonScanner 4200 , 4000B , or 4000A ( Molecular Devices ) . PMT levels were adjusted to achieve 0 . 1%–0 . 5% pixel saturation . Each element was located and analyzed using GenePix Pro 5 . 0 ( Molecular Devices ) . These data were submitted to the Stanford Microarray Database [155] for further analysis . Data were filtered , as described in Text S10 , to remove low-confidence measurements . Oligonucleotide pairs that both passed filtering criteria were averaged , and the data were globally normalized per array such that the mean log2 ( Cy5/Cy3 fluorescence ) ratio was zero after normalization . We analyzed a total of 123 IPs by microarray hybridization ( Dataset S1 ) . During the course of this work , we continued to improve and optimize our protocols . These changes and the manufacturing differences in reagents ( especially in the beads used in the IPs ) led to systematic differences in the background distribution of RNAs between corresponding experiments . We minimized systematic differences among sets of experiments by deriving estimates of the background separately for each set of experiments . Each group was normalized by subtracting the median log2 ratio for each molecular features across the experiments in a group from the log2 ratio of the molecular feature in each experiment . The details of the group normalization are described in Text S10 , and the groups are labeled in Table S5 . Hierarchical clustering was performed with Cluster 3 . 0 [156] , and the results were visualized as heat maps with Java TreeView 1 . 0 . 12 [157] . Clustering of FDR values ( Figures 1B and 4B ) was performed using the centered Pearson correlation as a similarity metric . FDR values that were greater than or equal to 10 and missing values were set to 10 prior to clustering . Clustering of the significance values measuring the degree of overlap between RBP target sets ( Figure 4A ) was performed using the uncentered Pearson correlation as a similarity metric . For SAM , unpaired two-class t-tests were performed with default settings . FDRs were generated from up to 1 , 000 permutations of group normalized data . Details of SAM analysis are described in Text S11 . The p-values of enrichment of specific classes of RNAs and GO terms in target sets were determined using the hypergeometric density distribution function and corrected for multiple hypothesis testing using the Bonferroni method . Enrichment of GO terms was performed with GO::TermFinder [158] . For noncoding RNAs , all RNAs for which we obtained reliable measurements on the microarray were used as background . For GO analysis , only probes that are meant to capture mature mRNAs were included in analyses . For oligonucleotide microarray experiments , this corresponds to probes that match the following regular expression: Y[A-P][RL][0–9]{3}[WC][-ABC]*_ORF ( Datasets S1–S3 ) . For cDNA microarray experiments , this corresponds to probes that match the following regular expression: Y[A-P][RL][0–9]{3}[WC][-ABC]* ( Datasets S1–S3 ) . mRNAs for which we obtained high-quality measurements were used as background . Yeast sequence files orf_genomic_1000 . fasta and orf_coding . fasta were downloaded from SGD ( ftp://ftp . yeastgenome . org ) . The 200 nucleotides upstream and downstream of coding sequences containing proper start and stop codons were extracted to create 5′-UTR and 3′-UTR databases , and the coding sequences were used for the coding sequence database . All-by-all WU-BLAST [159] ( http://blast . wustl . edu/ ) comparisons were performed for each database against itself to identify highly similar sequences ( using options -e 1e-10 -b 5000 -S 1 -F F ) . WU-BLAST output files were parsed to identify alignments of greater than or equal to 80% identity extending over half the length of the query sequence , and all such sequence pairs were grouped into redundant classes . One sequence from each redundant class was retained to create nonredundant databases for each region . The REFINE procedure was run using hexamers with significant ( p < 10−3 ) enrichment in RBP targets , as measured by the hypergeometric distribution ( using options –ss –f 3 –g 6 –ct 3 –max 15 –dust ) . MEME analysis ( version 3 . 5 . 1 ) was performed on the REFINE output sequences with options –dna –minw 6 –maxw 15 –text –maxsize 200000 –evt 10 –nmotifs 3 . Motif site sequences were extracted from MEME output and used to generate position-specific log-odds scoring matrices based on the observed frequencies and 0 . 25 pseudocounts per base , and null frequencies based on mononucleotide composition of all sequences in the corresponding ( 5′-UTR or 3′- UTR ) nonredundant database . Cutoff scores for motif classification were chosen to maximize the significance of association of motif sites with RBP target membership as measured by hypergeometric p-values for enrichment . All subsequences with scores above the cutoff threshold were classified as motif sites , and the final significance was measured as the negative log of the p-value of motif enrichment in RBP targets . FIRE analysis was run on the nonredundant 5′- and 3′-UTR databases using binary data indicating RBP target membership with options –exptype=discrete –seqlen_rna=200 –nodups=1 –dodna=0 . For both REFINE and FIRE , statistical significance of the predicted motifs was assessed by randomly generating target sets of similar size and repeating each procedure 100 times on the simulated target data . We defined a test statistic as the negative log of the p-value for motif enrichment for REFINE; the reported motif z-score was used for FIRE motifs , and we compared the observed values of these test statistics to the distributions generated by the random simulations ( Table S4 ) . Motifs were declared as significant if the observed test statistic was greater than three standard deviations above the mean , or if there was significant enrichment ( p < 10−4 ) of the motif in targets occurring in regions from which that motif was not derived .
Our microarray experiment data are publicly available from the Stanford Microarray Database and Gene Expression Omnibus . | Regulation of gene transcription has been extensively studied , but much less is known about how the fates of the resulting mRNA transcripts are regulated . We were intrigued by the fact that while most eukaryotic genomes encode hundreds of RNA-binding proteins ( RBPs ) , the targets and regulatory roles of only a small fraction of these proteins have been characterized . In this study , we systematically identified the RNAs associated with a select sample of 40 of the approximately 600 predicted RBPs in the budding yeast , Saccharomyces cerevisiae . We found that most of these RBPs bound specific sets of mRNAs whose protein products share physiological themes or similar locations within the cell . For 16 of the 40 RBPs , we identified sequence motifs significantly enriched in their RNA targets that presumably mediate recognition of the target by the RBP . The intricate , overlapping patterns of mRNAs associated with RBPs suggest an extensive combinatorial system for post-transcriptional regulation , involving dozens or even hundreds of RBPs . The organization and molecular mechanisms involved in this regulatory system , including how RBP–mRNA interactions are integrated with signal transduction systems and how they affect the fates of their RNA targets , provide abundant opportunities for investigation and discovery . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"molecular",
"biology",
"genetics",
"and",
"genomics"
] | 2008 | Diverse RNA-Binding Proteins Interact with Functionally Related Sets of RNAs, Suggesting an Extensive Regulatory System |
Oligomers of length k , or k-mers , are convenient and widely used features for modeling the properties and functions of DNA and protein sequences . However , k-mers suffer from the inherent limitation that if the parameter k is increased to resolve longer features , the probability of observing any specific k-mer becomes very small , and k-mer counts approach a binary variable , with most k-mers absent and a few present once . Thus , any statistical learning approach using k-mers as features becomes susceptible to noisy training set k-mer frequencies once k becomes large . To address this problem , we introduce alternative feature sets using gapped k-mers , a new classifier , gkm-SVM , and a general method for robust estimation of k-mer frequencies . To make the method applicable to large-scale genome wide applications , we develop an efficient tree data structure for computing the kernel matrix . We show that compared to our original kmer-SVM and alternative approaches , our gkm-SVM predicts functional genomic regulatory elements and tissue specific enhancers with significantly improved accuracy , increasing the precision by up to a factor of two . We then show that gkm-SVM consistently outperforms kmer-SVM on human ENCODE ChIP-seq datasets , and further demonstrate the general utility of our method using a Naïve-Bayes classifier . Although developed for regulatory sequence analysis , these methods can be applied to any sequence classification problem .
Predicting the function of regulatory elements from primary DNA sequence still remains a major problem in computational biology . These elements typically contain combinations of several binding sites for regulatory factors whose activity together specifies the developmental times , cell-types , or environmental signals in which the element will be active . Genetic variation in regulatory elements is increasingly thought to play a significant role in the etiology and heritability of common diseases , and surveys of Genome Wide Association Studies have highlighted the preponderance of significant variants in regulatory DNA [1] , [2] . An accurate computational model to predict regulatory elements can 1 ) help identify and link core sets of regulatory factors with specific diseases , and 2 ) predict the functional consequences of variation or mutations in specific sites within regulatory elements . We have recently introduced a successful method for regulatory DNA sequence prediction , kmer-SVM , which uses combinations of short ( 6–8 bp ) k-mer frequencies to predict the activity of larger functional genomic sequence elements , typically ranging from 500 to 2000 bp in length [3] . An advantage of k-mer based approaches relative to the alternative position weight matrix ( PWM ) approach is that PWMs can require large amounts of data to optimize and determine appropriate scoring thresholds [4] , [5] , while k-mers are simple features which are either present or absent . However , in our previous implementation of the kmer-SVM [3] , the choice to use a single k , and which k , is somewhat arbitrary and based on performance on a limited selection of datasets . A major contribution of the present work is an extension of this single k approach to include longer and much more general sequence features . The function of these DNA regulatory elements is generally thought to be specified at the molecular level by the binding of combinations of Transcription Factors ( TFs ) or other DNA binding regulatory factors , and many of these binding sites are short and fall within the range of k ( 6–8 ) where our kmer-SVM approach was successful . However , Transcription Factor Binding Sites ( TFBS ) can vary from 6–20 bp , so some are much longer ( such as ABF1 , CTCF , etc . ) , and thus cannot be completely represented by the short k-mers . Alternatively , TFBS can be defined by a set of sequences with some gaps ( non-informative positions ) as each given DNA sequence has some binding affinity for the TF . Although the kmer-SVM method can model TFBS longer than k by tiling across TFBS with overlapping k-mers , this loses some spatial information in the binding site , and overall classification accuracy can be significantly impaired when long TFBS are important predictive features [6] . Naively one could address this issue by using longer k's or combinations of k-mers spanning the expected size range of TFBS , but a major limitation of this approach is that longer k-mers generate extremely sparse feature vectors ( i . e . most k-mers simply do not appear in a training sequence and thus receive zero counts , or appear only once ) , which causes a severe overfitting problem even at quite moderate k . Therefore , the original kmer-SVM was limited in practice to k-mer lengths from 6 to 10 , with performance already degrading at k = 9 or 10 , depending on the dataset . Thus in practice , the parameter k was chosen by a tradeoff between resolving longer features and robust estimation of their frequencies . We recently introduced gapped k-mers as a way to resolve this fundamental limitation with k-mer features and showed that they can be used to more robustly estimate k-mer frequencies in real biological sequences [7] . In this paper , we present a simple and efficient method for calculation of the robust k-mer count estimates . We also expand our kmer-SVM method [3] to use gapped k-mers or robust k-mer count estimates as feature sets and present efficient methods to compute these new kernels . We show that our new method , gkm-SVM , consistently and significantly outperforms a kmer-SVM using both CTCF and EP300 genomic bound regions over a wide range of varying feature lengths . Furthermore , we show that , while kmer-SVM suffers significantly from overfitting as k is increased , gkm-SVM performance is only very modestly affected by changes in the chosen feature length parameters . Next , we systematically compare the two approaches on the complete human ENCODE ChIP-seq data sets , and show that gkm-SVM either significantly outperforms or is comparable to kmer-SVM in all cases . Of biological interest , on the ENCODE ChIP-seq data sets , we also show that gkm-SVM outperforms the best known single PWM by detecting necessary co-factors . We also systematically compare gkm-SVM to similar earlier SVM approaches [8]–[11] , and show that they perform comparably for optimal parameters in terms of accuracy , but that gkm-SVM is less sensitive to parameter choice and is computationally more efficient . To further demonstrate the more general utility of the k-mer count estimates , we apply them in a simple Naïve-Bayes classifier , and show that using k-mer count estimates instead of k-mer counts consistently improves classification accuracy . Since our proposed method is general , we anticipate that many other sequence classification problems will also benefit from using these features . For example , word based methods can also be used to detect functional motifs in protein sequences , where the length of the functional domain is unknown [12] .
To overcome the limitations associated with using k-mers as features described above , we introduce a new method called gkm-SVM , which uses as features a full set of k-mers with gaps . At the heart of most classification methods is a distance or similarity score , often called a kernel function in the SVM context , which calculates the similarity between any two elements in the chosen feature space . Therefore , in this section , we first describe the feature set and how to efficiently calculate the similarity score . This new feature set , called gapped k-mers , is characterized by two parameters; ( 1 ) l , the whole word length including gaps , and ( 2 ) k , the number of informative , or non-gapped , positions in each word . The number of gaps is thus l – k . We first define a feature vector for a given sequence S to be , where M is the number of all gapped k-mers ( i . e . for DNA sequences , ) , and 's are the counts of the corresponding gapped k-mers appeared in the sequence S . We then define a similarity score , or a kernel function , between two sequences , S1 and S2 , as the normalized inner product of the corresponding feature vectors as follows: ( 1 ) where , and . Therefore , the similarity score , K ( S1 , S2 ) , is always between 0 and 1 , and K ( S , S ) is equal to 1 . We will refer to Equation ( 1 ) as the gkm-kernel . It is similar to the wildcard kernel introduced in Ref . [9] , but our approach differs in that we do not sum over the number of wild-cards , or gaps , as formulated in Ref . [9] . Since the number of all possible gapped k-mers grows extremely rapidly as k increases , direct calculation of Equation ( 1 ) quickly becomes intractable . To implement gapped k-mers as features , it is necessary to overcome this serious issue , by deriving a new equation for K ( S1 , S2 ) that does not involve the computation of all possible gapped k-mer counts . The key idea is that only the full l-mers present in the two sequences can contribute to the similarity score via all gapped k-mers derived from them . Thus the inner product in Equation ( 1 ) , which involves a sum over all gapped k-mers , can be computed by a much more compact sum , which involves only a double sum over the sequential l-mers present in each of the two sequences: ( 2 ) where is the i'th l-mer appearing in S1 , is the j'th l-mer appearing in S2 , and n1 and n2 are the numbers of full l-mers in S1 and S2 respectively , i . e . n1 = length ( S1 ) −l+1 and n2 = length ( S2 ) −l+1 . Evaluation of Equation ( 2 ) is much more efficient than Equation ( 1 ) because almost always , . As will be shown below , hlk ( u1 , u2 ) only depends on the number of mismatches , m , between the two full l-mers , u1 and u2 , i . e . hlk ( u1 , u2 ) = hlk ( m ) . Therefore , we can rewrite Equation ( 2 ) by grouping all the l-mer pairs of the same number of mismatches together as follows: ( 3 ) where Nm ( S1 , S2 ) is the number of pairs of l-mers with m mismatches , and hlk ( m ) is the corresponding coefficient . We refer to Nm ( S1 , S2 ) as the mismatch profile of S1 and S2 . Since each l-mer pair with m mismatches contributes to common gapped k-mers , the coefficient hlk ( m ) , denoted in short by hm , is given by: ( 4 ) Determining a mismatch profile in Equation ( 3 ) is still computationally challenging since the numbers of mismatches between all possible l-mer pairs has yet to be determined . To address this issue , we have developed two different algorithms . We first considered direct evaluation of the mismatch profiles between all pairs of training sequences . To minimize the cost of counting mismatches between two words , we develop an efficient mismatch counting algorithm that practically runs in constant time , independent of k and l parameters ( see Methods ) . We then use Equation ( 3 ) to obtain the inner products for every pair of sequences . The direct and sequential evaluation of the kernel function between all training sequences becomes less practical as the number of training sequences gets larger , since it requires O ( N2L2 ) operations of mismatch counting between l-mer pairs , where N is the number of training sequences and L is the average sequence length . Because of this unfavorable scaling , we implemented an alternative method using a k-mer tree data structure , similar to one previously introduced in Ref . [8] , but with some modifications ( see Methods ) . This method simultaneously calculates the mismatch profile for all the sequence pairs , and , therefore , can significantly reduce the computation time especially when the number of gaps is relatively small , typically when l – k< = 4 . We can further improve the efficiency if we truncate the sum in Equation ( 3 ) to only consider up to a maximum number of mismatches , mmax ( see Methods ) . This approximate method is especially favorable when the number of gaps is large , but the efficiency comes at the cost of exact evaluation of the kernel and classification accuracy , which we will discuss in greater detail below . Therefore , we used one of the two algorithms depending on the size of data sets and the number of gaps we choose for analysis . Because of the difficulty of reliably estimating long k-mer counts , we hypothesized that gkm-SVM would perform better than kmer-SVM , and that gapped k-mers would be most advantageous as features , when long TFBSs are important sequence elements in a given data set . To directly test this idea , we compared the classification performance of gkm-SVM to kmer-SVM in predicting the binding sites of CTCF [13] in the human genome , a TF whose binding specificity has been well-characterized [14] . As shown in Figure S1 , CTCF recognizes very long DNA sequences ( the full PWM is 19 bp ) , and the genomic CTCF bound regions are almost perfectly predicted by matches to the CTCF PWM ( Figure S2 ) : in the PWM analysis , we used as a predictor the best matching log-odd score to the PWM model in the region , and achieved area under the ROC curve ( AUC ) of 0 . 983 . It is very rare for a single PWM to perform this well , and in our experience CTCF is unique in this regard . The CTCF dataset therefore provides an excellent opportunity to test our gapped k-mer classifier . We used the top 2 , 500 CTCF ChIP-seq signal enriched regions in the GM12878 cell line available at Gene Expression Omnibus ( GSE19622 ) [13] as a positive dataset , and equal numbers of random genomic sequences ( 1× ) as a negative dataset . We generated these negative sequences by matching length , GC and repeat fraction of the positive set [6] . We compared the performance of gkm-SVM and kmer-SVM on the CTCF data set for a range of oligomer lengths by varying either k ( for kmer-SVM ) or l ( for gkm-SVM ) from 6 to 20 . We fixed the parameter k = 6 for gkm-SVM . We then quantified the classification performance of each by calculating test-set AUC with standard five-fold cross validation ( CV ) ( see Methods ) . Figure 1A shows a summary of the comparisons . As anticipated , gkm-SVM performs consistently better than kmer-SVM for all lengths . More significantly , while kmer-SVM suffers severely from overfitting when k is greater than 10 , gkm-SVM is virtually unaffected by l . In fact , gkm-SVM achieves the best result ( AUC = 0 . 967 ) when l = 14 and k = 6 , which is significantly better than the kmer-SVM ( AUC = 0 . 912 when k = 10 ) ; the best ROC curve is shown in Figure 1C . It should be noted , however , that the PWM classification result ( Figure S2 ) is still the best ( AUC = 0 . 983 ) among the three methods we tested in this analysis . A complicating factor is that while both kmer-SVM and gkm-SVM use entire sequences ( average length is 316 bp ) to calculate the prediction scores , the PWM scores are from the best matching 19 bp sub-sequence in the region . It may be that the extra ∼300 bp sequences contribute noise in the SVM prediction scores , which slightly impairs the overall classification accuracy . In any event , the gkm-SVM is a significant improvement in accuracy over the kmer-SVM , and both gkm-SVM and the PWM are excellent predictors on this dataset . Interestingly , gkm-SVM shows consistently better performance than kmer-SVM even if l is relatively small ( l<10 ) ( Figure 1A ) . This suggests that gkm-SVM may also be better at modeling diverse combinations of TFBSs than kmer-SVM . To test this hypothesis , we analyzed a mouse enhancer dataset of more varied sequence composition: genomic EP300 bound regions in embryonic mouse forebrain [15] . We have previously shown that our original kmer-SVM classifiers can accurately predict EP300 binding when mediated by sets of active TFBSs [3] . This EP300 data set thus provides a direct test of the effectiveness of using gapped k-mer features to detect more complex regulatory features . For this analysis , we defined a new set of the 1 , 693 400 bp sites that maximize the EP300 ChIP-seq signal within each of the peaks determined by MACS [16] after removing any regions which were more than 70% repeats . We repeated the k and l scaling with the EP300 data set and a 1× negative set , and again found that gkm-SVM consistently outperforms kmer-SVM for all feature lengths ( Figure 1B ) . Analogous to the observations modeling CTCF binding , gkm-SVM AUC is high and does not degrade with large l . In contrast , the kmer-SVM accuracy drops rapidly as k increases . Moreover , although the difference in performance is smaller than found for the CTCF data set , the gkm-SVM achieves the best AUC ( 0 . 947 ) with l = 9 and k = 6 , while the kmer-SVM achieves 0 . 932 with k = 7 , suggesting that longer k-mers with some flexibility do contain more complete information about TF binding ( Figure 1D ) . At the same time , the gapped k-mer features are more robustly estimated ( having more counts ) and for this reason make more reliable predictors . The consequences of these improvements in AUC are significant when considering the genome-scale precision of the improved gkm-SVM classifiers . The rate of false positive predictions is dominated by the large neutral fraction of the genome , so the precision of a genome-scale classifier is best assessed by a Precision-Recall curve in combination with a much larger negative set , as discussed in Ref . [3] . The Precision-Recall curves for the gkm-SVM and kmer-SVM classifiers on a 100× negative set are shown in Figure S3 . For CTCF , at a recall of 50% , the precision increases from 36% to 59% . These ranges of precision and recall are in the relevant range of experiments aiming to discover and test novel enhancers , and we therefore expect that predictions based on gkm-SVM will have up to a two-fold higher successful validation rate . One further modification can substantially reduce the computational cost of using gapped k-mers with little degradation in performance . The algorithm using the k-mer tree data structure produces identical results to the direct evaluation of Equation ( 3 ) , but typically is much faster when the number of mismatches , l – k , is smaller than four , and the number of training sequences is large . The k-mer tree algorithm can be made even more computationally efficient , if we prune the traversal of the tree , by ignoring any k-mer pairs that have more mismatches than a predetermined parameter , mmax . This provides an approximation to the exact kernel calculation , but the approximation error is usually negligible given that the coefficient hm for large numbers of mismatches are generally much smaller compared to those with small m . This approximation significantly reduces the total number of calculations and allows the user to control the running time of the algorithm by setting the parameter mmax , and makes the use of longer word lengths l feasible for any given k . To systematically investigate the classification performance of this approximation , we applied the same analysis above using both CTCF and EP300 data sets ( Figure 1A , B ) , and found that AUCs from the approximate method are virtually identical to the exact method when the difference between mmax and l – k are small . Interestingly , the approximation method achieved even higher AUC with CTCF data set in some cases . Encouraged by the analyses of CTCF and EP300 data sets above , we systematically compared gkm-SVM to kmer-SVM using a very broad range of human data sets generated by the ENCODE project [17] , [18] . We used 467 sets of ChIP-seq peaks produced by the ENCODE uniform processing pipeline containing at least 500 regions ( see Methods ) . We truncated any data set with greater than 5 , 000 regions by random sampling . We then trained both kmer-SVM and gkm-SVM on each set against an equal size ( 1× ) negative set of random genomic regions and calculated AUCs with five-fold cross validation . We used k = 6 for kmer-SVM , and l = 10 and k = 6 for gkm-SVM , but as shown in Figure 1 the improvements are generally insensitive to these parameter choices . Strikingly , we find that gkm-SVM almost always outperforms kmer-SVM ( Figure 2A ) . We also find that variances of AUCs from test CV sets are generally reduced , suggesting that gkm-SVM is more robust than kmer-SVM ( Figure S4 ) . More significantly , gkm-SVM performs much better especially for TFs with long binding sites . In this dataset , most of these long binding sites arise in ChIP-seq data sets for CTCF and members of the cohesin complex ( RAD21 , SMC3 ) known to be physically associated with CTCF [19] . On these CTCF associated factors gkm-SVM exhibits much higher AUC than kmer-SVM , as highlighted by the cluster of purple circles in Figure 2A . We have also compared gkm-SVM to the best single PWM AUC as shown in Ref . [6] ( Figure 2B ) . As expected , gkm-SVM outperforms all datasets except CTCF , for which gkm-SVM performance is only marginally reduced . For a consistent analysis of this dataset , we used l = 10 and k = 6 , although for CTCF the gkm-SVM performance is optimal at larger l , as seen in Figure 1A . The predictive sequence features that allow gkm-SVM to outperform the single best PWM imply that cooperative binding is the underlying molecular mechanisms that targets TFs to these regulatory regions . Previously we have typically focused on a handful of the highest SVM weight k-mers ( say top ten positive and top ten negative weight k-mers ) to interpret the classification results [3] , [6] , [20] . This simple method becomes unwieldy when applied to the gkm-SVM results because of the large number of very similar significant features ( when l and/or k are large ) . Although the k-mers at the extreme top and bottom tails of the k-mer weight distribution are still important and biologically meaningful , those k-mers usually cover only a fraction of the significant feature set , and many more important features are included in the larger tails of the k-mer weight distribution . Therefore , more sophisticated algorithms are needed to extract the biologically relevant features from the classification results . To directly address this issue , we developed a new method to combine multiple similar k-mers into more compact and interpretable PWMs and analyzed the 467 ENCODE data sets [18] . In this approach , we used a larger number of predictive k-mers to build de novo PWMs ( see Methods ) . We used the top 1% of 10-mers from each of the gkm-SVMs trained on the ENCODE data sets and identified up to three distinct PWMs ( Figure S5 ) from k-mers in this set . We then compared our results with the previous PWMs found in the same data sets using a conventional tool ( MEME-ChIP ) [18][21] . Similar to our approach , Wang et al . analyzed 457 ENCODE ChIP-seq data sets ( 440 sets are in common with those we analyzed above ) and identified five PWMs from each data set . Collectively , Wang et al . found 79 distinct PWMs enriched , of which our method recovered 74 . Comparing each ChIP-seq data set individually , we recovered most of the PWMs reported by Wang et al . using our method ( Figure 2C ) . Interestingly , while Wang et al . largely failed to identify biologically meaningful PWMs from most of the POL2 ChIP-seq data sets ( 47 out of 58 sets returned no meaningful PWMs ) , our methods frequently identified cell-specific TFs as well as promoter specific TFs ( Figure S5 ) . For example , the GATA1 TF identified from POL2 ChIP-seq in the erythroleukemic cell line K562 is known to play central roles in erythroid differentiation [22] . The ETS1 TF from HUVEC is another extensively studied TF , known to be important for angiogenesis [23] . A major difference between the two methods is the number of training sequences . While the previous study was limited to the top 500 of ChIP-seq peaks ( ranked by ChIP-seq signal ) , we were able to use 10× larger numbers of ChIP-seq peaks ( 5 , 000 regions ) , and the large training sizes enabled us to robustly identify diverse combinatorial sequence features . Since the early development of k-mer based supervised machine learning techniques [24] , there have been a number of improvements . Some of these extend the feature set to include imperfect matches , similar in spirit to our gkm-SVM . The mismatch string kernel [8] is one such method , originally motivated by the fact that homologous protein sequences are not usually identical and have many frequently mutated positions . The mismatch kernel also uses k-mers as features , but allows some mismatches when counting k-mers and building feature vectors . The wildcard kernel [9] is another variant of the original string kernel , which introduces a wildcard character that matches any single letter in the given alphabet . More recently , an alternative di-mismatch kernel [10] has been proposed to directly model TFBSs , and has been successfully applied to protein binding microarray ( PBM ) data sets [25] and several other ChIP-seq data sets [10] , [11] . The di-mismatch method tries to overcome the limitation of the mismatch kernel by favoring k-mers with consecutive mismatches . However , in a recent comparison of methods for modeling transcription factor sequence specificity , full k-mer methods outperformed the di-nucleotide approaches when applied to PBM data [26] . To further evaluate our proposed method , we directly compared the gkm-kernel with the aforementioned three alternative methods , Mismatch kernel [8] , Wildcard kernel [9] , and Di-mismatch kernel [10] , [11] , using the mouse forebrain EP300 data set . As shown in Figure 3 , gkm-kernel outperforms the other three existing methods both in terms of the classification accuracy and running time . The best AUC we achieved for gkm-kernel is 0 . 947 as compared to 0 . 937 , 0 . 935 , and 0 . 944 for the wildcard kernel , mismatch kernel , and di-mismatch kernel , respectively ( Figure 3A ) . Although the wildcard kernel and gkm-kernel are quite similar , the systematic improvement in gkm-kernel AUCs is primarily due to the incorporation of reverse complement sequences . We directly tested this by adding reverse complement sequences to the feature set for the previously published methods , and indeed found that with this modification , these methods were also able to achieve comparable AUCs ( Figure S6 ) . To further evaluate our proposed method , we directly compared the gkm-kernel with the aforementioned three alternative methods , Mismatch kernel [8] , Wildcard kernel [9] , and Di-mismatch kernel [10] , [11] , using the mouse forebrain EP300 data set . As shown in Figure 3 , gkm-kernel outperforms the other three existing methods both in terms of the classification accuracy and running time . The best AUC we achieved for gkm-kernel is 0 . 947 as compared to 0 . 937 , 0 . 935 , and 0 . 944 for the wildcard kernel , mismatch kernel , and di-mismatch kernel , respectively ( Figure 3A ) . Although the wildcard kernel and gkm-kernel are quite similar , the systematic improvement in gkm-kernel AUCs is primarily due to the incorporation of reverse complement sequences . We directly tested this by adding reverse complement sequences to the feature set for the previously published methods , and indeed found that with this modification , these methods were also able to achieve comparable AUCs ( Figure S6 ) . More significantly , when we compare running times at parameters which maximize AUC for each method , our gkm-SVM implementation ( l = 9 , l-k = 3 ) is roughly two orders of magnitude faster than di-mismatch ( 10 , 3 ) , and slightly more than one order of magnitude faster than mismatch ( l = 10 , m = 2 ) and wildcard ( l = 8 , m = 3 ) on the EP300 data set ( Figure 3B and Figure S7 ) . Also , by fixing k = 6 and the parameter mmax in our algorithm , the AUC becomes less sensitive to the feature length l , compared to a scan at fixed m , varying k ( Figure 3A ) . Direct running time comparisons using our tree structure in the mismatch and wildcard kernels ( described below ) are shown in Figure S7B and S7C . We should note that we were only able to test the di-mismatch kernel up to l = 10 , because it required more than 128 GB of memory and did not finish within 2000 minutes when using l = 11 . Interestingly , we also note that both Mismatch kernel and Wildcard kernel are special cases of the more general class of kernels , defined by Equation ( 3 ) . This unification allows direct application of the methods we developed for mismatch profile computation and therefore gives more efficient methods for computation of these existing methods ( see Methods ) . As an alternative to the gapped k-mer feature set , we also developed an alternative kernel by replacing the k-mer counts with robust l-mer count estimates [7] in our original kmer-SVM framework . We have developed efficient methods to compute this new kernel ( see Methods ) . In Ref . [7] , we considered the mapping from l-mers to gapped k-mers . Among all possible sets of l-mer frequencies that could produce the same gapped k-mer frequency distribution , we developed a method to estimate the “most likely” l-mer frequency set . Full details of this method are described in the Ref . [7] . In brief , we first define a gapped k-mer count vector similar to the definition of the gapped k-mer feature vector for gkm-SVM as shown above . Then , the count estimate , , for l-mer u is given by ( 5 ) The weight in Equation ( 5 ) was shown to only depend on the number of mismatches , m , between the gapped k-mer corresponding to and u , and takes the following form: ( 6 ) where b is the alphabet size and is equal to four in case of DNA sequences ( A , C , G and T ) . Since the above equation is applied to every l-mer , it would provide a non-zero frequency even for an l-mer that does not have any exact match appearing in any training set sequence . Direct calculation of Equation ( 5 ) , however , requires actual counting of all of the M gapped k-mers , which becomes computationally intractable for large l and k in a way similar to Equation ( 1 ) . Besides , summing up a large set of floating point numbers may result in poor numerical precision . To overcome these issues , we developed a simple method , referred to as the gkm-filter , to more efficiently calculate the robust l-mer count estimates , , without calculating the intermediate gapped k-mer counts ( see Methods ) . In summary , in the calculation of the robust l-mer count estimates , we give a non-zero weight to l-mers with few numbers of mismatches . The k-mer frequency estimation method is not constrained to produce non-negative frequencies and may occasionally generate negative count estimates . To obtain strictly positive frequencies , we used a revised version of the gkm-filter method , which we call the truncated gkm-filter . Finally , we developed a method to directly calculate the kernels using these feature sets ( see Methods ) . An important result here is that the evaluation of the gkm-kernel ( the inner product of the l-mer count estimates vectors ) is still given by Equation ( 3 ) , but with a new set of weights clk ( m ) given by Equation ( 14 ) , below , replacing hlk ( m ) . Therefore , efficient algorithms for pairwise mismatch profiles that we developed for the gkm-kernel can be directly used for this new feature set without any modification . Because of this symmetry , we also refer to this method as gkm-kernel with ( full or truncated ) filter . A numerical example using count estimates on two short sequences is provided in Text S1 . To systematically compare the classification performance of these new methods with the original gapped k-mers , we repeated the previous analysis with the ENCODE ChIP-seq data sets . Using the truncated gkm-filter yields results highly comparable to the original gkm-SVM for most datasets with modestly but consistently better relative performance when AUC is greater than 0 . 9 ( shown as purple circles in Figure S8A ) . Any improvement in the range of high AUC ( >0 . 9 ) typically strongly reduces the classifier's False Prediction Rate [27] , therefore , we generally recommend the truncated filter method as the method of choice for most analyses . Compared to the original gkm-SVM , using the gkm-SVM with full filter yields lower AUCs ( Figure S8B ) although it is still significantly higher compared to the kmer-SVM method . So far , we have focused on using gapped k-mer based methods for improving sequence kernel methods . We have shown that , by direct use of gapped k-mers as features or by using the robust l-mer count estimates , we can significantly overcome the long k-mers' sparse count problem for these methods . We further demonstrate the general utility of the robust l-mer count estimates in sequence classification problems by applying it to a simple Naïve-Bayes ( NB ) classifier similar to the one previously introduced in Ref . [28] and show that by using robust count estimates instead of conventional k-mer counts we can significantly boost the performance of the Naïve-Bayes classifier for long k-mers . Here , we used the log-likelihood ratio of the estimated l-mer frequencies in the positive and negative sets as a predictor , using the NB assumption of feature independence . The prediction score of any given sequence of length n , denoted by S = s0s1…sn–1 , is then given by: ( 7 ) where NP and NN are the robust count estimates of the corresponding l-mers , sisi+1…si+l−1 , in the positive and negative training set , and are given by Equation ( 11 ) below . We used the truncated gkm-filter method adding pseudo-count ( half of the smallest positive coefficient of the truncated gkm-filter ) to each of the estimated frequencies to obtain strictly positive frequencies for log-likelihood ratio . As a comparison , we also implemented the NB classifier without the gkm-filter , using actual l-mer counts with a pseudo-count ( 0 . 5 ) for NP and NN . We predicted the CTCF and EP300 genomic bound regions with both NB classifiers ( i . e . with and without using robust count estimates ) . As shown earlier , genomic CTCF bound regions are almost perfectly predicted by the single CTCF PWM ( Figure S2 ) , and the local sequence features around the CTCF binding motif do not seem to significantly contribute to the prediction . Thus , to precisely detect the CTCF binding motif and achieve the best classification performance , we scored every substring of length n = 15+l−1 for each sequence and assigned the maximum as the final score for the sequence . The window size of 15 was chosen to optimize the detection of the CTCF site within a small window of flanking sequence , which maximizes the performance of the NB classifier without the gkm-filter . For the EP300 genomic bound regions , in contrast , we used the full sequence in both classifiers . We compare the performance of these NB classifiers on both data sets in Figure 4 for a range of feature length ( 6–20 bp ) . Similar to the previous analysis using gkm-SVM and kmer-SVM ( Figure 1 ) , using robust count estimates ( gkm-filter ) significantly improves the classification accuracy especially for longer k-mers ( Figure 4 ) . On the CTCF data set , the NB classifier using the gkm-filter achieves best performance with l = 20 ( AUC = 0 . 99 ) , which is even better than that of the CTCF PWM ( red dotted line , AUC = 0 . 983 ) ( Figure 4A ) . Also on the EP300 dataset , the gkm-filter significantly improves the overall performance of NB classifier ( Figure 4B ) . The superior classification performance using gapped k-mer based features is thus consistent for both SVM and NB classifiers , and strongly suggests that the robust l-mer count estimates provide a more complete and robust set of sequence features than simple k-mers in most sequence classification problems , as hinted at in our preliminary analysis of k-mer frequency spectra in Ref . [7] .
In this paper , we presented a significantly improved method for sequence prediction using gapped k-mers as features , gkm-SVM . We introduced a new set of algorithms to efficiently calculate the kernel matrix , and demonstrated that by using these new methods we can significantly overcome the sparse k-mer count problem for long k-mers and hence significantly improve the classification accuracy especially for long TFBSs . Detailed comparisons of our proposed method with some existing methods show that our gkm-SVM outperforms existing methods in terms of classification accuracy on benchmark data and is also typically orders of magnitude faster . We also introduced the concept of gkm-filters for efficient calculation of the robust k-mer count estimates and derived optimal weights for penalizing different number of mismatches . We showed that one could successfully replace k-mers with robust k-mer count estimates to avoid long k-mer sparse count problem , and demonstrated the effectiveness of this method by showing examples in SVM and Naïve-Bayes classifiers . We thus expect that most k-mer based methods can be significantly improved by simply using this generalized k-mer count . The main biological relevance of the computational method we present in this paper is that gkm-SVM is capable of accurately predicting a wide range of specific classes of functional regulatory elements based on DNA sequence features in those elements alone . This in itself is interesting and implies that the epigenomic state of a DNA regulatory element primarily is specified by its sequence . In addition , our predictions facilitate direct investigation of how these elements function , either by targeted mutation of the predictive elements within the larger regulatory region , or by modulating the activity of the TFs which bind the predictive sequence elements . We are currently using changes in the gkm-SVM score to systematically evaluate the predicted impact of human regulatory variation ( single nucleotide polymorphisms ( SNPs ) or indels ) to interpret significant SNPs identified in genome wide association studies . We demonstrated that gkm-SVM is better at predicting all ENCODE ChIP-seq data than the best single PWM found from the ChIP-seq regions , or previously known PWMs . The gkm-SVM is able to do so by integrating cofactor sequences which may not be directly bound by the ChIP-ed TF but facilitate its occupancy . To predict this ChIP-seq set accurately required the improved accuracy of the gkm-SVM and its ability to describe longer binding sites such as CTCF , which were very difficult for our earlier kmer-SVM approach . We recovered most of the cofactors found by traditional PWM discovery methods , but we further show that these combinations of cofactors are predictive in the sense that they are sufficient to define the experimentally bound regions . There are some further issues that need to be considered in the application of these methods . First , one will typically be interested in finding an optimal set of the parameters ( l and k ) to achieve the best classification performance . A significant advantage of gapped k-mer methods over k-mer methods is that they are more robust and are less sensitive to the particular choices of l or k compared to kmer-SVM or NB classifiers , as shown in Figure 1 and Figure 4 . Nevertheless , these parameters can still be optimized to maximize cross validation AUC . As a general rule , we have found that when choosing the parameter k , which determines how different numbers of mismatches are weighted , given a whole word length l , smaller values of k ( typically less than 8 ) are usually better when important sequence elements are believed to be more degenerate or when only small amount of training data is available . Although the choice of k directly affects the feature set , our analysis of several datasets shows that the overall performance of the classifier is not very sensitive to changes in k . The parameter l is directly related to TFBS lengths and should be comparable to or slightly larger than the longest important feature , as demonstrated by our analysis of the CTCF and EP300 data sets in Figure 1 and Figure 4 . Our approach also avoids an issue that would arise if one chose instead to directly use Equation ( 5 ) for computing count estimates . This would involve a large number of floating point operations , and accumulated round-off error could become significant in the large summations . There are some algorithms , such as Kahan compensated summation [29] , which can significantly reduce this error , however , we explicitly avoided evaluating this sum by first computing the mismatch profiles between sequences , which involves only integer calculations . Then , we calculate the weighted sum of the number of mismatches using Equation ( 11 ) , which involves a much smaller number of floating point operations . Two issues which are left for future investigation are different treatment of end vs . internal gaps , and allowing imperfect mismatches . We currently do not make special consideration for gaps which occur at the end of a k-mer instead of internal gaps . Also , our implementation of a mismatch treats all nucleotides equally , but often TF binding sites can prefer an A or T in a given position , or a purine vs . pyrimidine pair . Our approach recovers these preferences by assigning different weights to k-mers which do not have gaps at these positions , but including a wider alphabet including ( W , S , Y , R ) for ( AT , GC , AG , CT ) may have some advantages . Throughout this paper , we have focused on using DNA sequences as features for classifying the molecular or biological function of a genomic region . However , in principle , our method can be applied to any classification or prediction problem involving a large feature set . In general , when the number of features used by a classifier increases , the number of samples in the training set for each point in the feature space becomes smaller , and small sample count issues occur ( which we have resolved using gapped k-mers ) . One approach to the large feature space is feature selection , which selects a subset of features and builds a classifier only using those features , ignoring all the other features . However , usually a limited subset of features cannot explain all the variation in the predicted quantity . While hypothetical at this point , our analysis suggests that an alternative approach might be of general value . Analogous to the way we have used gapped k-mers to more robustly estimate k-mer feature frequencies , we speculate that there may be a general approach which uses subsets of a larger feature set to combine observed feature counts with weights reflecting the similarity to some generalized feature . These estimated feature frequencies will be less susceptible to statistical noise by construction , and thus may provide consistently better classification performance , as we have shown for gapped k-mers .
The Support Vector Machine ( SVM ) [301] , [31] is one of the most successful binary classifiers and has been widely used in many classification problems . We have previously developed an SVM based framework , or “kmer-SVM” , for enhancer prediction and have successfully applied to embryonic mouse enhancers [3] and many other regulatory datasets [6] , [20] . Briefly , our kmer-SVM method finds a decision boundary that maximally discriminates a set of regulatory sequences from random genomic non-regulatory sequences in the k-mer frequency feature vector space . Here , we developed new kernel functions using gapped k-mers and l-mer count estimates as features , and software that calculates the kernel matrix . For SVM training , we developed a custom Python script that takes the kernel matrix as input and learns support vectors . We used Shogun Machine Learning Toolbox [32] and SVM-light [33] for the SVM training script . As an alternative method , we also implemented an SVM classifier based on the iterative algorithm described in Ref . [34] . For direct computation of the gkm-SVM kernel matrix , we represent each training sequence with a list of l-mers and corresponding count for each l-mer . Then for each pair of sequences , we compute the number of mismatches for all pairs of l-mers and use the corresponding coefficient hm to obtain the inner product of Equation ( 3 ) . As the number of unique l-mers in each sequence is L and the number of sequences is N , this algorithm would require O ( N2L2 ) comparisons . In addition , a naive algorithm for counting the number of mismatches between two l-mers ( i . e . the hamming distance ) would be O ( l ) . Our implementation employs bitwise operators , providing a constant-factor speedup . Briefly , using two bits to represent each base ( A , C , G and T ) , we used an integer variable to represent non-overlapping substrings of t base pairs of the l-mer , therefore using total integers to represent each l-mer , where is the ceiling function . For counting the number of mismatches , we take the bitwise XOR ( exclusive OR ) of the integer representations of the two l-mers and use a precomputed look-up table to obtain the total number of mismatches using the XOR result . This method requires a look-up table of size 22t . The optimal value of t depends on the processor architecture and amount of cache memory . We used t = 6 for our analysis . As depicted in Figure 5 , we use a k-mer tree to hold all the l-mers in the collection of all of the sequences . We construct the tree by adding a path for every l-mer observed in a training sequence . Each node ti at depth d represents a sub-sequence of length d , denoted by s ( ti ) , which is determined by the path from the root of the tree to the node ti . Each terminal leaf node of the tree represents an l-mer , and holds the list of training sequence labels in which that l-mer appeared and the number of times that l-mer appeared in each sequence . As an example , Figure 5 shows the tree that stores all the substrings of length l = 3 in three sequences S1 = AAACCC , S2 = ACC , and S3 = AAAAA . Then , to evaluate the mismatch profile we traverse the tree in a depth-first search ( DFS ) [35] order . In contrast to the mismatch tree used in Ref . [8] , here for each node ti , at depth d , we store the list of pointers to all the nodes tj at depth d for which s ( ti ) and s ( tj ) have at most l – k number of mismatches . We also store the number of mismatches between s ( ti ) and s ( tj ) . Similar to the mismatch tree [8] , we do not need to store these values for all the nodes in the tree , but we compute them recursively as we traverse the tree . When reaching a leaf node , we increment the corresponding mismatch profile Nm ( Si , Sj ) for each pair of sequences Si in that leaf node's sequence list , and all the Sj's in the list of sequences in the pointer list for that leaf node . At the end of one DFS traversal of the tree , the mismatch profiles for all pairs of sequences are completely determined . To increase the speed further , we also introduce an optional parameter mmax , which limits the maximum number of mismatches . By setting mmax smaller than l – k , we only consider l-mer pairs that have at most mmax number of mismatches . This can reduce calculation significantly by ignoring l-mer pairs which potentially contribute less to the overall similarity scores . This method provides fast and efficient approximations of the exact solution . In addition , we only compute the lower triangle of the matrix because of the symmetry in the kernel matrix . Hence , at each node ti , we exclude the nodes tj in the list that have maxID ( ti ) <minID ( tj ) , where minID ( ti ) and maxID ( tj ) are the maximum and minimum sequence ID in the subtrees of ti and tj respectively and are computed and stored for each node at the time we build the tree . We developed a new method for building de novo PWMs by systematically merging the most predictive k-mers from a trained gkm-SVM . We first determined a set of predictive k-mers by scoring all possible 10-mers and selecting the top 1% of the high-scoring 10-mers . We then found a set of distinct PWM models from these predictive 10-mers using a heuristic iterated greedy algorithm . Specifically , we first built an initial PWM model from the highest scoring 10-mer . Then , for each of the remaining predictive 10-mers , we calculated the log-odd ratios of all possible alignments of the 10-mer to the PWM model , and identified the best alignment ( i . e . the position and the orientation that give rise to the highest log-odd ratio value ) . Since multiple distinct classes of TFBSs are expected to be identified in most cases , we only considered 10-mers with good alignments ( i . e . we used threshold of 5 . 0 for log-odd ratio scores relative to a genomic GC = 0 . 42 background ) . After each of the 10-mers was aligned , we updated the PWM model only with successfully aligned 10-mers . To further refine the PWM , we repeated this by iterating through all of the top 1% 10-mers until no changes were made . When updating the PWM model , we assumed that the contribution of each k-mer is exponentially weighted proportional to its SVM score , using exp ( α wi ) , with α = 3 . 0 . The 10-mers used for creating the 1st PWM were then removed from the list , and the process was repeated on the remaining predictive k-mers , to find up to three PWMs . Lastly , we matched our PWMs to the previously identified PWMs [18] using TOMTOM [36] software . Each of the PWMs identified by our method were associated with Ref . [18] PWMs if the q-value ( false discovery rate ) <0 . 05 . In the gkm-kernel , we define the feature vector to consist of the frequency of all the l-mers with exactly k known bases and l – k gaps . In contrast , the wildcard kernel [9] also includes all the l-mers with l – k wildcards , where l – k ranges from 0 to the maximum number of wildcards allowed , M . Thus in the wildcard kernel , the parameter M replaces k in our gkm-kernel . In the sum , these are weighted by λl - k to penalize sequences with more wildcards [9] . We derived an equation to directly compute the inner products from the mismatch profiles without the need to calculate the actual gapped k-mer counts . Here we show that a similar approach can be used to calculate the wildcard kernel . We derive a new set of coefficients that can substitute hm , in Equation ( 3 ) . To evaluate we only need to consider the contribution of each pair of l-mers with m mismatches in the inner product of the corresponding feature vectors of the two sequences . Equation ( 8 ) gives those weights: ( 8 ) Using the above form allows us to directly use the fast algorithms we have developed for calculation of the mismatch profiles to calculate the wildcard kernels . Although there are similarities between our tree algorithm and the tree algorithm described in Ref . [9] , there are some key differences . In the Ref . [9] , the algorithm literally transverses all the possible gapped l-mers ( with maximum M number of gaps ) while our algorithm takes advantage of the fact that the final inner product will only depend on the number of pairwise mismatches and hence only traverses all the l-mers that are present in the input data . Another difference is that Ref . [9] uses a list of all partially matching l-mers at each node of the tree , while we use a list of pointers to tree nodes instead . So , for example , at the beginning of the algorithm ( at depth d = 0 ) they start with a large list consisting of all the possible l-mers in the input data , while in our algorithm the list at depth d = 0 consists of only one node ( the root of the tree ) . Using this representation of all the partially matching l-mers , we can more efficiently perform the comparisons at each step of the algorithm when the tree is dense . In the mismatch string kernel described in Ref . [8] and [9] , the feature vectors consist of the counts for all the l-mers with maximum distance M from the l-mers in the sequence . Here we show that the approach above can be used to implement the mismatch kernel . Again , the only difference is in the set of weights used in Equation ( 3 ) . To calculate the mismatch string kernel value for two sequences we replace hlk ( m ) in Equation ( 3 ) by : ( 9 ) where b is the alphabet size ( b = 4 for DNA sequences ) and r = m1+m2−m−2t . Given two l-mers x1 and x2 where x1 and x2 differ in exactly m places , the term inside the summations counts the number of all possible l-mers that exactly differ x1 in m1 places and x2 in m2 places t of which fall in the common l-m bases of x1 and x2 . ( See Figure S9 ) . So the result of the summation is the number of all l-mers that differ x1 and x2 in at most M places . This form for the mismatch string kernel has the advantage that we can directly use equation ( 3 ) to compute the kernels by only having the mismatch profiles that can be computed more efficiently . To compute the l-mer count estimates by using Equation ( 5 ) , one should first calculate the gapped k-mer counts , yi , and then use Equation ( 5 ) to combine the yi with a weight corresponding to the number of mismatches , given by Equation ( 6 ) . This is shown schematically in Figure S10 . The mapping from observed l-mer counts to gapped k-mer counts is performed by the matrix A , whose elements are aij . If the gapped k-mer vi matches l-mer uj , then aij = 1 , otherwise aij = 0 . There is a second matrix W , which performs the mapping from gapped k-mer counts to estimated l-mer counts , and whose elements are wij . In a previous work we showed that matrix W is the Penrose-Moore pseudo-inverse of A [7] . The element wij only depends on the number of mismatches between the l-mer ui and the gapped k-mer vj , and is given by Equation ( 6 ) . Here we show that , for efficient computation , we can combine the two mapping matrices , A and W , and directly calculate the minimum norm l-mer count estimates from actual l-mer counts in a sequence . We refer to this combined mapping as the gkm-filter . The combined mapping matrix G = WA , has elements gij , shown on the bottom of Figure S10 . As shown below , gij also only depends on the number of mismatches , m , between the l-mers ui and uj . We denote these values by glk ( m ) and refer to this as the gkm-filter since the domain and range of this mapping is the same . To obtain the element glk ( m ) , that gives the weight for the contribution of an observed l-mer ui in the training set to the minimum norm l-mer count estimate uj that has exactly m mismatches with ui , we sum over the contribution of all the gapped k-mers vτ that match ui . Note that aij = 0 for all other gapped k-mers . There exist different gapped k-mers that match ui and have exactly m mismatches with uj . Figure S11 shows how we enumerate all these gapped k-mers . The black solid circles denote the m mismatch positions of ui and uj , the gray circles denote the l – m match positions and the empty dotted circles denote the l – k gap positions . For a gapped k-mer to have exactly t mismatches with uj , there are ways to select the t mismatch positions and ways to select the k – t match positions . Now considering the weight w ( t ) for the gapped k-mers with t mismatches , the gapped k-mer filter elements , glk ( m ) can be obtained as follows: ( 10 ) In other words , there are different ways we can construct a gapped k-mer that matches ui , and has exactly t mismatches with uj , by selecting t positions from the m mismatch positions and k – t positions from the l – m match positions as explained above ( Figure S11 ) . It can be easily shown that glk ( m ) is a polynomial of degree k in m . Now using the weights given in Equation ( 10 ) , for any given l-mer , u we finally obtain the minimum norm l-mer count estimate as follows: ( 11 ) where Ntr ( u , m ) is the number of l-mers with exactly m mismatches with u in the training set . For large values of l and k , the number of all possible gapped k-mers gets exponentially large and since this method avoids evaluating the gapped k-mer counts , it significantly reduces the cost of calculating the l-mer count estimates compared to the original method we developed in Ref . [7] . In summary , we defined a generalized k-mer count ( referred to as the robust l-mer count estimates ) by giving a non-zero weight to l-mers with few number of mismatches ( In the conventional k-mer count only perfectly matching k-mers are counted ) . These weights are given by glk ( m ) . Figure S12A shows the plots for glk ( m ) for l = 20 and various values of k . Each plot is normalized so that weight corresponding to zero mismatches is equal to one . The case with l = k is equivalent to the conventional k-mer count . Also Figure S12B shows glk ( m ) for and various values of l . With a fixed length l , higher values of k result in smaller coefficients for larger mismatches , and therefore less smoothing of the estimated counts ( Figure S12 ) . Moreover , glk ( m ) can become slightly negative for large numbers of mismatches . This is because in our estimation of the frequencies we did not restrict the frequencies to be positive , and doing so would yield a more complicated expression . The assumed Gaussian distribution allows non-physical negative frequencies to have non-zero probability . A beta-distribution would not have this problem but would introduce offsetting complications . In cases where the estimated counts are required to be strictly positive , such as when we need to calculate the logarithm or ratios of the estimated frequencies , we truncate the gkm-filter glk ( m ) by setting glk ( m ) = 0 for every m≥m0 , where m0 is the smallest number of mismatches for which glk ( m0 ) <0 . This will give an approximation to the value of in Equation ( 5 ) , so it will no longer strictly be the minimum norm estimate , but it will guarantee that all the count estimates are non-negative . Given a sequence S , we define an l-mer count estimate vector where N is the number of all l-mers ( 4l in case of DNA sequences ) , and is the estimated count of the ith l-mer appearing in sequence S using Equation ( 11 ) . Then , we can calculate a standard linear kernel simply by using this vector in Equation ( 1 ) . Similar to the gkm-kernel method , we can further simplify this equation using the same technique introduced in Equation ( 2 ) which does not involve the computation of individual l-mer estimates . We show that the inner product of the two l-mer count estimate vectors can be obtained as follows: ( 12 ) where n1 and n2 are the number of l-mers in S1 and S2 , and is the i'th l-mer in S1 and is the j'th l-mer in S2 . If u1 and u2 have exactly m mismatches then c ( u1 , u2 ) = cm . Grouping all the l-mer pairs with m mismatches , we can rewrite Equation ( 12 ) as follows: ( 13 ) where Nm ( S1 , S2 ) is the mismatch profile of S1 and S2 as previously defined in Equation ( 3 ) . We show that the weight clk ( m ) , denoted in short by cm , can be obtained as: ( 14 ) where r = m1+m2−2t−m , b is the alphabet size . The summations are taken over the range 0 to l . Figure S13 shows how we obtained the equation for cm , similar to the previous development shown in Figure S11 . Given two l-mers u1 and u2 , with mismatches and l – m matched positions , we want to enumerate the number of all possible l-mers , u , that have m1 mismatches with u1 and m2 mismatches with u2 . For this , we assume that t of the m1 mismatches are among the l – m match positions and m1 – t of them are among the m mismatch positions . There are ways to choose these m1 positions and ( b – 1 ) t choices for the values of the t mismatches . These t mismatches plus the m− ( m1−t ) unselected mismatch positions also do not match u2 . Then , for the remaining r = m2− ( t+m− ( m1−t ) ) mismatches for u2 there are ways to select the positions and ( b – 2 ) r ways to select the values . Hence the total number of l-mers , u with m1 mismatches with u1 and m2 mismatches with u2 , where t of the mismatches of u1 and u are among the l – m match positions of u1 and u2 is given by . Using matrix notation , we can further show that cm = gm if we use the full filter glk ( m ) . To see this , note that where and are the l-mer count vectors for S1 and S2 . Given G = WA , we have . Hence , . Here A is the binary incidence matrix that maps l-mer counts to gapped k-mer counts as defined in Ref . [7] and W is the Moore-Penrose pseudo-inverse of A . Note that this result does not hold for the truncated filter gm . In that case , we directly use Equation ( 14 ) to obtain cm coefficients . To compare the performance of different classification methods , we calculated the area under the receiver operating characteristic ( ROC ) curve for each classifier . To plot the ROC curves and calculate area under the curves ( AUCs ) we used the ROCR package [37] in R . Following standard five-fold cross validation procedures , we divided the positive and negative sets into five segments , left one segment out as the test set and used the other four segments for training . We repeated for all of the five segments and calculated the mean and standard error of the prediction accuracy on the test set elements . The ENCODE ChIP-seq datasets were downloaded from ftp://ftp . ebi . ac . uk/pub/databases/ensembl/encode/integration_data_jan2011/byDataType/peaks/jan2011/spp/optimal/hub/ . We have implemented these algorithms in C++ , and the source code and executable files are available on our website at http://www . beerlab . org/gkmsvm/ . | Genomic regulatory elements ( enhancers , promoters , and insulators ) control the expression of their target genes and are widely believed to play a key role in human development and disease by altering protein concentrations . A fundamental step in understanding enhancers is the development of DNA sequence-based models to predict the tissue specific activity of regulatory elements . Such models facilitate both the identification of the molecular pathways which impinge on enhancer activity through direct transcription factor binding , and the direct evaluation of the impact of specific common or rare genetic variants on enhancer function . We have previously developed a successful sequence-based model for enhancer prediction using a k-mer support vector machine ( kmer-SVM ) . Here , we address a significant limitation of the kmer-SVM approach and present an alternative method using gapped k-mers ( gkm-SVM ) which exhibits dramatically improved accuracy in all test cases . While we focus on enhancers and transcription factor binding , our method can be applied to improve a much broader class of sequence analysis problems , including proteins and RNA . In addition , we expect that most k-mer based methods can be significantly improved by simply using the generalized k-mer count method that we present in this paper . We believe this improved model will enable significant contributions to our understanding of the human regulatory system . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"sequencing",
"techniques",
"sequence",
"analysis",
"computer",
"and",
"information",
"sciences",
"functional",
"genomics",
"molecular",
"biology",
"network",
"analysis",
"genetics",
"regulatory",
"networks",
"biology",
"and",
"life",
"sciences",
"molecular",
"biology",
... | 2014 | Enhanced Regulatory Sequence Prediction Using Gapped k-mer Features |
The exclusive localization of the histone H3 variant CENP-A to centromeres is essential for accurate chromosome segregation . Ubiquitin-mediated proteolysis helps to ensure that CENP-A does not mislocalize to euchromatin , which can lead to genomic instability . Consistent with this , overexpression of the budding yeast CENP-ACse4 is lethal in cells lacking Psh1 , the E3 ubiquitin ligase that targets CENP-ACse4 for degradation . To identify additional mechanisms that prevent CENP-ACse4 misincorporation and lethality , we analyzed the genome-wide mislocalization pattern of overexpressed CENP-ACse4 in the presence and absence of Psh1 by chromatin immunoprecipitation followed by high throughput sequencing . We found that ectopic CENP-ACse4 is enriched at promoters that contain histone H2A . ZHtz1 nucleosomes , but that H2A . ZHtz1 is not required for CENP-ACse4 mislocalization . Instead , the INO80 complex , which removes H2A . ZHtz1 from nucleosomes , promotes the ectopic deposition of CENP-ACse4 . Transcriptional profiling revealed gene expression changes in the psh1Δ cells overexpressing CENP-ACse4 . The down-regulated genes are enriched for CENP-ACse4 mislocalization to promoters , while the up-regulated genes correlate with those that are also transcriptionally up-regulated in an htz1Δ strain . Together , these data show that regulating centromeric nucleosome localization is not only critical for maintaining centromere function , but also for ensuring accurate promoter function and transcriptional regulation .
The eukaryotic genome is packaged into chromatin , which consists of 147 bp repeating units of DNA wrapped around histone proteins to form nucleosomes [1] . Chromatin is important not only for packaging and protecting DNA , but also for regulating access of genes and other DNA elements to nuclear proteins involved in processes such as transcription , replication , and chromosome segregation . Most nucleosomes are composed of the canonical histone proteins , H2A , H2B , H3 , and H4 [2] . However , the behavior and functions of nucleosomes can be altered both by chemically modifying canonical histones through post-translational modifications and by exchanging canonical histones for histone variants that alter nucleosome composition [2] . For example , H2A . Z is a variant of histone H2A and is found at promoter nucleosomes genome-wide where it regulates transcription [2–4] . In contrast , the conserved CENP-A variant ( also called CenH3 ) replaces H3 in nucleosomes exclusively at the centromere where it regulates chromosome segregation [5–7] . Because changes in nucleosome composition can have a major impact on the underlying functions of the genome , it is critical to understand the mechanisms that control the localization of histone modifications and variants . The genomic incorporation of the budding yeast H2A . ZHtz1 ( SGD ID: S000005372 ) histone variant is regulated by the SWR1 ( SWR-C ) and INO80 ( INO80-C ) chromatin remodeling complexes [8] . H2A . ZHtz1 localizes to intergenic regions , specifically near transcription start sites ( TSS ) at the +1 and -1 nucleosomes surrounding nucleosome-depleted regions ( NDRs ) [3 , 4 , 9–12] . In budding yeast , H2A . ZHtz1 nucleosomes are correlated with high nucleosome turnover [13] , which is proposed to assist transcriptional initiation or rapid changes between transcriptional states [14–16] . SWR-C incorporates H2A . ZHtz1 into nucleosomes by exchanging H2A/H2B dimers for H2A . ZHtz1/H2B dimers [17–19] . In contrast , the mechanism of H2A . ZHtz1 removal from nucleosomes by INO80-C is less well understood because it has two reported activities that both lead to H2A . ZHtz1 exchange , either by swapping H2A . ZHtz1/H2B dimers for H2A/H2B dimers [20] or by promoting turnover of the entire nucleosome [8 , 19] . The localization of the CENP-A variant is regulated by the histone chaperone HJURP ( Scm3 in budding yeast ) , which is targeted specifically to centromeres [21–25] . Centromeric sequence and size are highly variable throughout eukaryotes and can be specified by either an underlying sequence or through epigenetic inheritance [26 , 27] . Despite the diversity of centromeres , CENP-A is a conserved hallmark of all centromeres . The presence of CENP-A directs the formation of the kinetochore , a large protein complex that mediates attachments between the microtubules of the mitotic spindle and the chromosome during cell division [26 , 28 , 29] . CENP-A mislocalization to euchromatin through overexpression or tethering can lead to ectopic kinetochore formation and genomic instability [30–32] . However , CENP-A mislocalization has not been reported to disrupt other genomic processes [33 , 34] . Multiple mechanisms ensure that CENP-A does not mislocalize to euchromatin . A number of chromatin remodelers and histone chaperones are reported to help maintain centromeric chromatin or prevent CENP-A mislocalization , including Fun30 , RSC , INO80-C , CAF-1 , HIR , FACT , and RbAp48 and SWI/SNF [35–41] . In addition , ubiquitin-mediated proteolysis prevents ectopic CENP-A localization by controlling total CENP-A protein levels [42–45] . In budding yeast , proteolysis of CENP-ACse4 ( SGD ID: S000001532 ) is mediated by an E3 ubiquitin ligase called Psh1 ( SGD ID: S000005415 ) [46 , 47] . When CENP-ACse4 is overexpressed in the absence of Psh1-mediated proteolysis [42 , 46–48] , cells accumulate high levels of CENP-ACse4 in euchromatin . This also results in lethality , although the underlying cause has not been determined [42 , 46 , 47] . Similar to CENP-A , H2A . Z also contributes to chromosome segregation . In human cells , H2A . Z is found at pericentromeric regions , where it is incorporated at the inner centromere between the CENP-A nucleosome domains , and helps to establish centromeric heterochromatin [49 , 50] . Similarly , H2A . ZHtz1 is also a component of pericentromeric chromatin in budding yeast , where it localizes to nucleosomes flanking the CENP-ACse4 nucleosome and is important for chromosome segregation through unknown mechanisms [4 , 51 , 52] . However , it is unclear whether there is a connection between the localization of the histone variants . In human cells , overexpressed CENP-A was found to mislocalize to regions enriched for H2A . Z , although no physical interaction was detected between these two histone variants [33] . In contrast , studies in S . pombe have shown that CENP-ACnp1 tends to mislocalize to ectopic regions that are depleted of H2A . ZHtz1 [53] . We set out to determine whether there are features of euchromatin that normally prevent budding yeast CENP-A misincorporation as well as to identify the functional consequences of CENP-A mislocalization to euchromatin . The identification of euchromatic sites that strongly misincorporate CENP-A may also shed light on the underlying cause of the lethality . To address these questions , we performed the first genome-wide analysis of CENP-A overexpression in the absence of ubiquitin-mediated degradation . We found that overexpressed CENP-ACse4 mislocalizes to promoters that are enriched for NDRs flanked by H2A . ZHtz1 , and this mislocalization is dramatically enhanced in cells that cannot degrade CENP-ACse4 . This localization pattern appears to be due in part to co-opting of INO80-C to incorporate excess CENP-ACse4 into promoter nucleosomes that normally contain H2A . ZHtz1 . Consistent with this , there was a significant correlation between transcripts that were misregulated in cells lacking H2A . ZHtz1 and those with high levels of CENP-ACse4 mislocalization . We also found that a subset of promoters that misincorporate CENP-ACse4 have decreased transcription , which may be the underlying cause of lethality . Together , these data suggest that it is essential that cells regulate CENP-ACse4 localization not only to ensure proper chromosome segregation , but also to protect cells from promoter nucleosome disruption and transcriptional misregulation .
To identify the precise genomic sites of CENP-ACse4 mislocalization in budding yeast , we performed ChIP-seq on endogenous and overexpressed CENP-ACse4 in the presence and absence of Psh1-mediated proteolysis . All strains contained a fully functional ectopic 3Flag-CSE4 gene integrated at the URA3 locus under the endogenous promoter and were deleted for the endogenous CSE4 gene . Cells overexpressing CENP-ACse4 contained an additional copy under the control of the GAL promoter ( pGAL-3Flag-CSE4 ) . As seen previously , CENP-ACse4 overexpression inhibited the growth of WT cells and resulted in lethality in psh1Δ cells ( S1A Fig ) [46 , 47] . The growth inhibition correlated with the total amount of chromatin-bound CENP-ACse4 protein ( Fig 1A , S1B Fig ) . To analyze CENP-ACse4 localization , cells were crosslinked with formaldehyde and the chromatin was isolated and subsequently digested with Micrococcal nuclease ( MNase ) , which cuts linker DNA between nucleosomes . The CENP-ACse4 nucleosomes were purified from the MNase-treated chromatin by immunoprecipitation of 3Flag-Cse4 . The amount of CENP-ACse4 recovered in the ChIP samples reflected the starting levels in the chromatin ( S1C Fig ) . The input samples ( MNase-digested chromatin ) and ChIP samples ( 3Flag-Cse4-bound chromatin after immunoprecipitation ) were made into paired-end sequencing libraries using a modified Solexa library preparation protocol that captures DNA particles down to ~25 bp ( S1D Fig ) [54 , 55] . Paired-end sequencing resulted in greater than 1 . 5 million reads/sample , with an average read length ranging from 147–164 bp ( S1 Table ) . The mononucleosome-sized sequencing reads from the input and ChIP samples for each strain were mapped to the S . cerevisiae reference genome version SacCer3 [56] . The peaks of CENP-ACse4 enrichment genome-wide correlated with the levels of chromatin-bound CENP-ACse4 ( Fig 1B , Table 1 ) . Seventeen peaks were identified for the 3Flag-CSE4 strain , representing the sixteen centromeres as well as a peak 150 bp from CEN9 . A small amount of CENP-ACse4 mislocalization was seen starting in the psh1Δ strain with 66 peaks , and was further increased in cells overexpressing CENP-ACse4 with 4043 peaks . The greatest enrichment in the euchromatin was detected in the psh1Δ cells overexpressing CENP-ACse4 with 14 , 199 peaks . An example of the coverage data and corresponding peaks for a representative region around Centromere 4 shows a single centromere peak for the WT strain and additional peaks around the centromere in the other strains ( Fig 1C ) . The increased CENP-ACse4 mislocalization in surrounding euchromatin is especially apparent in the pGAL-3Flag-CSE4 and psh1Δ pGAL-3Flag-CSE4 strains that have the highest levels of CENP-ACse4 . We independently confirmed the CENP-ACse4 enrichment at CEN4 and at other representative peaks by ChIP-qPCR ( S2A Fig ) . Our initial analysis also identified a CENP-ACse4 peak at the rDNA locus in all strains . This did not show significant enrichment in the 3Flag-CSE4 strain by ChIP-qPCR but did in the cells with overexpressed CENP-ACse4 , similar to previously reported data [46 , 57] ( S2A Fig ) . However , due to the difficulty in analyzing this repetitive region by standard mapping algorithms , ChIP coverage of this region was excluded from further computational analyses . To determine if mislocalized CENP-ACse4 favors certain genomic regions , we analyzed the percentage of CENP-ACse4 peaks in various functional regions of the genome , including centromeres , pericentromeres , telomeres , replication origins , genes , and intergenic regions ( Fig 1D ) . We defined pericentromeres as 20 Kilobases ( Kb ) flanking each centromere , consistent with the 20–50 Kb size of cohesin enrichment around each centromere in budding yeast [58 , 59] . As expected , the majority of CENP-ACse4 peaks in WT cells were at centromeres , with an increase in pericentromeric peaks in the psh1Δ mutant . However , the majority of peaks in the strains overexpressing CENP-ACse4 were in the intergenic regions , with a smaller percentage within genes . As intergenic regions make up less than 30% of the entire genome , these data indicate a strong enrichment of CENP-ACse4 in intergenic regions in cells overexpressing CENP-ACse4 . We next asked whether the intergenic enrichment correlates with features known to be associated with centromeres . ChIP-seq of mildly overexpressed CENP-ACse4 previously identified 23 centromere-like regions ( CLRs ) on chromosome arms that are enriched for mislocalized CENP-ACse4 and other kinetochore proteins [60] . These CLRs share characteristics with centromeric sequences such as having a high AT% and conferring stability to plasmid DNA . As expected , most of the CLRs have CENP-ACse4 peaks in the psh1Δ pGAL-3Flag-CSE4 strain ( S2B Fig ) . However , CENP-ACse4 was overexpressed to much higher levels in our study ( 150-fold compared to 3-fold ) , so the CLRs are a small fraction of the total peaks . Consistent with this , there was also enrichment in low confidence negative control regions ( LCNCRs ) , indicating there is no preference for CLR localization . We also analyzed the AT content of the DNA bound by mislocalized CENP-ACse4 , as this is a defining characteristic of centromeric DNA in budding yeast . As expected , CENP-ACse4 peaks were highly enriched for AT nucleotides in the WT strain . However there was only a moderate increase in AT% in the psh1Δ strain compared to the input nucleosomes , and almost no AT bias in the strains with overexpressed CENP-ACse4 ( S2C–S2F Fig ) . Together , these data indicate that the mislocalization of CENP-ACse4 is due to a more widespread effect than just centromere-like characteristics . We next asked whether the intergenic enrichment of overexpressed CENP-ACse4 was specific to either promoters ( defined as 500 bp upstream of the transcription start site ( TSS ) ) or transcription terminators ( defined as 500 bp downstream of the transcription termination site ( TTS ) ) by calculating the number of peaks in these regions ( Table 1 ) . CENP-ACse4 was enriched in both regions when overexpressed , so we more precisely analyzed the pattern by plotting the average coverage in 10 bp windows for regions 500 bp upstream and downstream of all TSS or TTS ( the TSS or TTS is plotted at position 0 based on previously reported RNA-seq transcription start positions [61] ) . In the psh1Δ cells overexpressing CENP-ACse4 , there was enrichment -200 bp from the TSS and directly over the TSS , which correspond to the -1 and +1 nucleosomes respectively ( Fig 2A ) . At the TTS , CENP-ACse4 was enriched in the nucleosome just after the termination site , and was shifted slightly into the NDR compared to the WT nucleosomes . Although the level of CENP-ACse4 enrichment in the other three strains was much lower overall , the trend is similar in the cells with increased CENP-ACse4 . This pattern is reminiscent of the pattern of CENP-ACse4 mislocalization upon deletion of CAC1 and HIR1 , which leads to ectopic CENP-ACse4 enrichment at promoters in the presence of Psh1 [38] . We next asked whether the accumulation of CENP-ACse4 in promoters and terminators is associated with the basal level of transcription in WT cells . We plotted CENP-ACse4 enrichment at the TSS and TTS of genes binned into quartiles by the published transcription levels in a WT strain , ranked from lowest transcription to highest transcription [62] . However , there was no correlation between CENP-ACse4 enrichment and the different transcription levels ( Fig 2B , S3 Fig ) . Therefore , the CENP-ACse4 localization to promoters in the psh1Δ pGAL-3Flag-CSE4 strain was not an artifact of increased chromatin accessibility in areas of high transcription , such as was found in the previously reported CENP-ACse4 ChIP-seq for slightly overexpressed or hypomorphic CENP-ACse4 [63] . We also analyzed whether CENP-ACse4 mislocalization correlated with the direction of transcription of the surrounding genes , since this has been shown for cohesin localization , which is specifically enriched in convergent intergenic regions outside of the pericentromere [58 , 64] . We classified the intergenic regions as tandem ( between two genes transcribed in the same direction ) , convergent ( between two genes transcribed towards each other ) , or divergent ( between two genes transcribed away from each other ) ( Fig 2C ) . In promoters , CENP-ACse4 was enriched at the tandem and divergent genes ( Fig 2D , S4 Fig ) . At the terminators , CENP-ACse4 was enriched at the tandem TTS and depleted at the convergent TTS . Because convergent regions lack promoters , these data are consistent with the enrichment of CENP-ACse4 to promoter regions . Since CENP-ACse4 mislocalization to promoters was not correlated with transcription levels , we looked for another chromatin feature specific to promoters that might enhance CENP-ACse4 incorporation . One characteristic of promoters that is less commonly found at the 3’ ends of genes is the NDR between the -1 and +1 nucleosomes at the TSS [65] . We therefore compared CENP-ACse4 profiles centered on all NDRs and found a strong CENP-ACse4 enrichment in the nucleosomes flanking the NDRs in the psh1Δ pGAL-3Flag-CSE4 strain ( Fig 3A ) . Because NDRs vary in length up to 557 bp , we asked whether there was a specific NDR length that correlated with CENP-ACse4 mislocalization and found the highest enrichment in NDRs longer than 65 bp ( Fig 3B ) . We obtained similar results when the analysis was centered on the TSS instead of the NDR ( S5A–S5D Fig ) , consistent with the enrichment of CENP-ACse4 in NDR containing promoters . The localization of CENP-ACse4 to the nucleosomes flanking the NDRs is similar to H2A . ZHtz1 , the only other histone variant in budding yeast [4] . In addition , the SWR-C chromatin-remodeling complex that incorporates H2A . ZHtz1 preferentially binds to NDRs greater than 50 bp [66] , similar to the length of NDRs that have the highest CENP-ACse4 enrichment ( greater than 65 bp ) ( Fig 3B ) . We therefore investigated the relationship between previously reported H2A . ZHtz1 localization [4] and the mislocalization of overexpressed CENP-ACse4 in psh1Δ cells . There was a striking similarity in their enrichment at NDRs ( Fig 3C ) , as well as a similar trend of co-enrichment in the nucleosomes flanking replication origins ( S5E and S5F Fig ) and centromeres ( Fig 3D , S5G Fig ) . The CENP-ACse4 coverage at the TSS was also similar to H2A . ZHtz1 coverage , while at the TTS CENP-ACse4 was shifted more into the 3’ NDR than H2A . ZHtz1 ( S6A and S6B Fig ) . The histone variants exhibited a genome-wide trend to co-localize , as seen in a representative region of the arm of Chromosome 4 ( Fig 3E and S6C Fig ) . There was a high coincidence of overlap between CENP-ACse4 peaks in the experimental strains with H2A . ZHtz1 peaks in WT cells , although they were not specifically enriched in any of the genomic features correlated with CENP-ACse4 mislocalization ( Fig 3F , S6D and S6E Fig ) Together , these data indicate a significant enrichment of misincorporated CENP-ACse4 at sites where H2A . ZHtz1 nucleosomes are normally located genome-wide in psh1Δ cells overexpressing CENP-ACse4 . The co-localization of the histone variants led us to further analyze their relationship . First , we tested whether H2A . ZHtz1 promotes CENP-ACse4 localization by performing ChIP on WT , psh1Δ , and psh1Δ htz1Δ cells overexpressing CENP-ACse4 . htz1Δ cells are defective in induction from the GAL promoter [67] , so we used a tetracycline promoter to control CSE4 levels . Overexpressed CENP-ACse4 bound to promoter regions in the psh1Δ htz1Δ double mutant , at levels similar to or even higher than the psh1Δ strain ( Fig 4A ) . These data indicate that H2A . ZHtz1 is not required for CENP-ACse4 mislocalization , so we next asked whether the H2A . ZHtz1 incorporation machinery is involved . Swr1 ( SGD ID: S000002742 ) is the Swi/Snf related ATPase in SWR-C that deposits H2A . ZHtz1 into nucleosomes [11 , 19 , 68] , so we measured the levels of chromatin-bound CENP-ACse4 in swr1Δ cells . We confirmed that H2A . ZHtz1 was reduced at a previously reported promoter nucleosome locus by ChIP-PCR ( Fig 4B ) [17 , 68 , 69] . Similar to our findings with the htz1Δ mutant , bulk H2A . ZHtz1 was not depleted in the chromatin fraction in swr1Δ , but CENP-ACse4 chromatin levels were somewhat higher in the swr1Δ psh1Δ cells compared to psh1Δ ( Fig 4C , S7A and S7B Fig ) . In addition , there was no change in CENP-ACse4 stability in swr1Δ cells ( S7C Fig ) . We also tested whether CENP-ACse4 overexpression in the psh1Δ mutant affects H2A . ZHtz1 promoter occupancy , but did not detect an effect at the loci analyzed ( S7D Fig ) . However , given that H2A . ZHtz1 is estimated to occupy only a small proportion of nucleosomes at any given locus in the population , it may be difficult to detect a significant difference [11 , 19] . Together , our data suggest that although ectopic CENP-ACse4 and WT H2A . ZHtz1 localize to similar sites , the H2A . ZHtz1 incorporation machinery does not promote CENP-ACse4 mislocalization and may instead help to prevent CENP-ACse4 promoter incorporation . Since the ectopic localization of CENP-ACse4 does not depend on H2A . ZHtz1 incorporation , we asked whether chromatin remodelers that remove H2A . ZHtz1 are involved . INO80-C has been reported to act preferentially on H2A . ZHtz1-containing +1 nucleosomes and to promote full nucleosome turnover [19 , 20] . We therefore hypothesized that CENP-ACse4 might be incorporated into chromatin when canonical H3 is removed by INO80-C-mediated nucleosome turnover . Previous work showed that deletion of the ATPase Ino80 ( SGD ID: S000003118 ) leads to a global alteration of H2A . ZHtz1 localization patterns genome-wide without affecting the overall levels of H2A . ZHtz1 incorporation in the genome [20 , 70] . However , this deletion mutant is not viable in the strain background we used in this study [71] . We therefore used a deletion of NHP10 ( SGD ID: S000002160 ) , a non-essential INO80-C subunit that facilitates binding to nucleosomes and DNA , but that does not affect catalytic activity in vitro [72–74] . To analyze CENP-ACse4 levels , we performed chromatin fractionation in WT and nhp10Δ cells overexpressing CENP-ACse4 . Similar to previously reported work , we did not detect a change in total H2A . ZHtz1 levels in the chromatin in the nhp10Δ strain ( S8A and S8B Fig ) [20 , 70] . However , CENP-ACse4 chromatin levels were somewhat reduced when NHP10 was deleted ( Fig 5A and S8B Fig ) , suggesting that INO80-C histone exchange activity contributes to CENP-ACse4 misincorporation . To more directly test this possibility , we asked whether Ino80 associates with CENP-ACse4 in vivo . CENP-ACse4 co-immunoprecipitated with Ino80 ( Fig 5B ) , and this interaction increased in the absence of Psh1 . To determine how this affects cell viability , we also analyzed the growth of nhp10Δ mutant cells overexpressing CENP-ACse4 . Although strong CENP-ACse4 overexpression is lethal to psh1Δ cells regardless of the presence of NHP10 ( S8C Fig ) , a deletion of NHP10 improved the growth of psh1Δ mutant cells that were moderately overexpressing CENP-ACse4 ( Fig 5C ) . We confirmed these effects were not due to altered levels or stability of CENP-ACse4 in nhp10Δ mutant cells ( S8D and S8E Fig ) . Together , these data suggest that at least some of the ectopic CENP-ACse4 deposition is likely coupled to the chromatin remodeling activity of INO80-C . The mislocalization of CENP-ACse4 to promoters suggested that it could lead to transcriptional changes in the downstream genes . In addition , the relationship between CENP-ACse4 incorporation and H2A . ZHtz1 removal by INO80-C suggested that any transcriptional changes might correlate with those in htz1Δ cells . We therefore performed RNA-seq on WT , psh1Δ , pGAL-3Flag-CSE4 , psh1Δ pGAL-3Flag-CSE4 and htz1Δ strains that were treated with galactose for two hours . As a control , we also included a pGAL-H3 strain to ensure any effects were specific to CENP-ACse4 overexpression and not just an effect of increased histone turnover . Cells containing just a PSH1 deletion or overexpressing CENP-ACse4 or H3 had very little change in transcription ( Fig 6A and 6B ) . However , a large number of genes were misregulated in psh1Δ cells overexpressing CENP-ACse4 , as well as in htz1Δ cells as previously described [75 , 76] . We confirmed that these gene expression changes were not due to an indirect effect of CENP-ACse4 mislocalization to the rDNA by measuring the rDNA copy number and rRNA transcript levels , which were not significantly different between the strains ( S9A and S9B Fig ) . We also confirmed that the differentially transcribed genes in the psh1Δ pGAL-3Flag-CSE4 strain are not a consequence of altered cell cycle progression [47 , 77] ( S9C Fig ) . To determine whether CENP-ACse4 mislocalization to promoters correlates with transcriptional misregulation of downstream genes , we compared the promoters with CENP-ACse4 peaks to the genes showing altered transcription in the psh1Δ strain overexpressing CENP-ACse4 . While there was a significant overlap ( p = 0 . 0009 , hypergeometric distribution ) between the down-regulated genes and those with promoter CENP-ACse4 peaks ( S9D Fig ) , the vast majority of genes with CENP-ACse4 promoter peaks do not have changes in transcription . This is similar to the relationship between H2A . ZHtz1 peaks and the genes that are differentially regulated in htz1Δ [9] , confirming that changes in the histone composition of promoters does not always lead to direct transcriptional effects . However , the downregulated genes have much higher CENP-ACse4 coverage at the +1 nucleosome compared to other promoters , suggesting that both the amount and position of CENP-ACse4 misincorporation may determine which downstream genes become misregulated ( Fig 6C ) . Analysis of transcription factor binding sites enriched at promoters of the downregulated genes with CENP-ACse4 promoter peaks identified Cse2 ( SGDID: S000005293 ) as the most significantly enriched transcription factor ( S2 File ) . Cse2 is a subunit of the RNA Polymerase II Mediator complex , and has also been shown to be required for chromosome segregation [79 , 80] , leading to the possibility that the transcriptional defects are correlated with altered Cse2 function . Given the relationship between CENP-ACse4 and H2A . ZHtz1 localization , we also asked whether there was a correlation between the transcriptional changes in psh1Δ pGAL-3Flag-CSE4 and htz1Δ mutant cells . Interestingly , there was a significant overlap between the genes that increased transcription in both strains ( Fig 6D ) , and these were also enriched for CENP-ACse4 in the NDR ( S9E Fig ) . We analyzed the promoters of the affected genes for common transcription factors and found 24 that are enriched at the promoters of these genes ( S2 File ) , so the underlying mechanism for the misregulation is not clearly associated with one factor . However , these data are consistent with the relationship between CENP-ACse4 mislocalization and the INO80-C chromatin remodeling machinery that controls H2A . ZHtz1 .
We identified a strong similarity between H2A . ZHtz1 localization and CENP-ACse4 mislocalization in nucleosomes flanking NDRs , such as replication origins , centromeres , and +1 nucleosomes at promoters . We also found that INO80-C contributes to CENP-ACse4 mislocalization . CENP-ACse4 co-immunoprecipitates with INO80-C , and this interaction is increased in the psh1Δ mutant where there are higher levels of CENP-ACse4 . Consistent with this , an nhp10Δ mutant reduced the ectopic localization and partially rescued the growth defect of the psh1Δ mutant when CENP-ACse4 was overexpressed . However , nhp10Δ does not fully rescue the lethality or ectopic deposition , so additional chromatin remodelers or histone chaperones must also contribute to ectopic CENP-ACse4 incorporation . In humans , the chaperone activity of DAXX is involved in CENP-A deposition in euchromatin [33] , but there is no ortholog of this protein in budding yeast . H2A . ZHtz1 localization to nucleosomes flanking NDRs requires SWR-C binding , and SWR-C enrichment is increased with longer NDRs in vivo [66] . Similarly , we found that CENP-ACse4 is enriched at longer NDRs . However , we determined that H2A . ZHtz1 and SWR-C are not required for CENP-ACse4 deposition . Our work is instead consistent with the possibility that the two yeast histone variants could have an antagonistic relationship , such that they are found at the same places in the genome , but never at the same time . This is reminiscent of the relationship between CENP-ACnp1 and H2A . ZHtz1 in fission yeast , where CENP-ACnp1 forms neocentromeres in regions with low H2A . ZHtz1 when the endogenous centromere is deleted [53] . However , we detect CENP-ACse4 mislocalization at nucleosomes that normally have high H2A . ZHtz1 enrichment . We speculate that this is due to different mechanisms leading to ectopic deposition . In fission yeast , the ectopic CENP-ACnp1 localization to neocentromeres depended on the centromeric chaperone [53] , while our data suggests a role for INO80-C in the ectopic deposition of highly expressed CENP-ACse4 . Given that INO80-C acts in opposition to SWR-C to remove H2A . ZHtz1 from nucleosomes , we propose that the full nucleosome turnover activity of INO80-C leads to the removal of H3 and the incorporation of CENP-ACse4 into promoter nucleosomes ( Fig 6E ) . This model explains both the co-localization of the histone variants and the potentially antagonistic relationship between H2A . ZHtz1 and CENP-ACse4 in the chromatin . Although there is a significantly higher level of euchromatic CENP-ACse4 in the absence of Psh1 , the locations of the ectopic nucleosome positions are similar regardless of Psh1 activity . In both cases , overexpressed CENP-ACse4 is enriched intergenically , suggesting that Psh1 does not have preferential sites of action genome-wide . However , CENP-ACse4 was not significantly incorporated into genes even in the absence of Psh1 , suggesting that additional mechanisms control its localization . We previously showed that the FACT complex , which was recently demonstrated to remove H2A . ZHtz1 from genes , interacts with Psh1 to facilitate CENP-ACse4 degradation [47 , 70 , 81] . However , FACT does not interact with CENP-ACse4 in the absence of Psh1 [47] . One possibility is that FACT could indirectly antagonize CENP-ACse4 mislocalization into genes by ensuring that H3 is quickly reincorporated into nucleosomes following transcription , similar to its role in fission yeast [40] . In the future , it will be important to understand how intragenic regions are protected from CENP-ACse4 deposition . For the first time in any organism , we detected large-scale changes in transcription when CENP-A mislocalized to euchromatin . This only occurred in cells lacking Psh1 , and the downregulated genes had very high levels of CENP-ACse4 in their promoters . This suggests that strong misincorporation of CENP-ACse4 at a promoter may be required to cause transcriptional defects , and may explain why this has not been previously observed . The levels of CENP-ACse4 overexpression achieved in the absence of proteolysis are much higher than previous studies that have analyzed CENP-ACse4 mislocalization . It is not clear whether mislocalization of CENP-ACse4 at a given promoter is sufficient to directly decrease transcription . We found a significant enrichment of the Cse2 transcription factor in the promoters of the downregulated genes , leading to the intriguing possibility that CENP-ACse4 incorporation alters Cse2 function at a subset of genes to inhibit transcription . It is interesting to note that Cse2 and Cse4 were identified in the same genetic screen for mutants in chromosome segregation [5 , 79] , and it will be important to further explore their relationship in the future . We also identified genes that increased transcription when CENP-ACse4 was mislocalized , and these significantly overlap with those altered in htz1Δ mutant cells . This further confirms the potential antagonistic relationship between the yeast histone variants , and suggests that high levels of CENP-ACse4 may lead to similar chromatin changes at a subset of promoters as cells lacking H2A . ZHtz1 . The underlying mechanism for why only a fraction of promoters that contain H2A . ZHtz1 are transcriptionally up-regulated in its absence is not known . We speculate that a change in nucleosome positioning or stability occurs at these promoters that facilitates the access of transcriptional machinery . Consistent with this , we found that the up-regulated gene promoters have CENP-ACse4 enrichment within rather than flanking the NDR and lack strong +1 enrichment . We found that regulating the levels and localization of the centromeric histone variant is critical to prevent transcriptional misregulation in budding yeast . Although CENP-A mislocalization leads to the formation of ectopic kinetochores in other organisms , we have not been able to determine whether this occurs in budding yeast due to the difficulty of detecting ectopic kinetochores [47] . Our work suggests the possibility that transcriptional defects due to the mislocalization of CENP-ACse4 in the absence of proteolysis may be the underlying cause of lethality in these cells . These data highlight the need to accurately regulate the localization of the centromeric histone variant CENP-ACse4 to both ensure genomic stability through its centromeric functions , as well as to prevent the disruption of euchromatic functions .
Microbial techniques and media were as described [82 , 83] . For all experiments involving induction of pGAL-3Flag-CSE4 or pGAL-H3 , budding yeast cells of indicated strains were grown to log phase ( OD 0 . 55–0 . 8 , Bio-Rad SmartSpec 3000 ) in lactic acid media at 23°C and induced for 2 hours with 2% galactose . Yeast strains were constructed using standard genetic techniques . Epitope-tagged proteins were constructed using either a PCR integration technique [84] or by the integration of plasmids after restriction digestion . Specific plasmids and yeast strains used in this study are described in the S2 and S3 Tables . Protein extracts to check total CENP-ACse4 levels were prepared as described [85] . Immunoblots using chemiluminescence were performed as previously described [85] . For all immunoblots , the antibody dilutions were as follows: Mouse anti-Pgk1 monoclonal antibodies ( Invitrogen Catalog # 459250 ) at a 1:10 , 000 dilution were used as a loading control . Mouse anti-Flag M2 monoclonal antibodies ( Sigma-Aldrich Catalog # F3165 ) were used at a 1:3000 dilution , Mouse anti-HA 12CA5 monoclonal antibodies ( Roche Catalog # 1-583-816 ) were used at a 1:10 , 000 dilution , and rabbit anti-H2B polyclonal antibodies ( Active Motif Catalog # 39237 ) were used at a 1:3 , 000 dilution . Mouse anti-Myc 9E10 monoclonal antibodies were used at a 1:10 , 000 dilution ( Covance Catalog # MMS-150R ) . Co-IP experiments were performed as previously described [81] for Psh1-Myc and Ino80-Myc strains using 5ul Protein G Dynabeads conjugated with 1 . 5ul anti-Myc ( A-14 , SC-789 ) and run on a gradient SDS-PAGE gel . Quantitative immunoblots were carried out according to [86] with the modification of using 4% non-fat milk in PBS as the blocking agent for the anti-Flag immunoblot . Briefly , IRDye anti-mouse and anti-rabbit secondary antibodies from LI-COR were used at a 1:15 , 000 dilution . The immunoblots were imaged on a LI-COR imaging system , and the protein levels were quantified using Image Studio Lite . Chromatin fractionation assays were performed as described [81] , followed by quantitative immunoblots . The mean and SEM of three independent experiments is reported . anti-PGK1 was used as a marker and loading control for the soluble fraction , and anti-H2B was used as a marker and loading control for the chromatin fraction . The Cse4:H2B and H2A . Z:H2B ratios were normalized to the pGAL-3Flag-CSE4 strain . Note that the levels of H2A . ZHtz1 and H2B are somewhat variable between strains . This may be due to differential susceptibility of the cell wall to zymolyase digestion during the chromatin fractionation procedure , which seems to vary between strains . To control for this , we used H2B to determine the level of total chromatin in each condition . 3Flag-Cse4-containing nucleosomes were isolated by ChIP of 3Flag-Cse4 using monoclonal anti-Flag M2 antibodies ( Sigma-Aldrich Catalog # F3165 ) . ChIPs were performed with Micrococcal nuclease ( MNase , Worthington Biochemical Corporation Catalog # LS004798 ) -treated chromatin as described [55] with the following addition . Before nuclei isolation , proteins were crosslinked to DNA with 1% formaldehyde for 15 minutes . Crosslinks were then reversed before DNA extraction by the addition of 1% SDS and an overnight incubation at 65°C [87] . DNA was extracted using phenol:chloroform extraction and ethanol precipitation , and was treated with RNAse and purified using a Qiagen Reaction Clean-up kit before library construction . Paired-end sequencing libraries of both input DNA from MNase-digested chromatin and 3Flag-Cse4 ChIP DNA were prepared using a modified Solexa library preparation protocol that captures DNA particles down to ~25 bp [55] . Cluster generation , followed by 25 cycles of paired-end sequencing on an Illumina HiSeq 2000 , was performed by the Fred Hutchinson Cancer Research Center Genomics Shared Resource facility , resulting in 24 bp paired end reads . Base calling was performed using Illumina's Real Time Analysis software v1 . 13 . 48 . 0 . Raw FASTQ sequence files were deposited in the NCBI GEO Series GSE69696 . Raw reads ( passing Solexa quality test ) were mapped to the S . cerevisiae reference genome version SacCer3 ( Saccharomyces Genome Database ( SGD ) /UCSC ) using the Burrows-Wheeler Aligner ( BWA ) [88] . The resulting Binary Sequence Alignment/Map ( BAM ) files were filtered for proper pairs with a mapping score > = 30 using samtools [89] . Mononucleosomes were identified as paired-end reads with insert sizes between 50 bp and 240 bp using R Bioconductor packages GenomicRanges , rtracklayer , Rsamtools , nucleR , and the UCSC SacCer3 reference genome [56 , 90–92] . ChIP reads were compared to the input reads for each strain using the Dynamic Analysis of Nucleosome and Protein Occupancy by Sequencing , version 2 ( DANPOS2 ) function Dpos with background subtraction [93] , and the background-subtracted ChIP signal was normalized to the coverage at centromeric regions for each strain , which contains a CENP-ACse4 nucleosome throughout the cell cycle [87] , and smoothed using the default DANPOS2 Dpos smoothing parameters [93] . The resulting normalized coverage data was visualized using the Integrated Genomics Viewer ( IGV ) [94 , 95] . Wiggle track format ( WIG ) files of the normalized coverage for each sample in 10 bp steps are available under NCBI GEO Series GSE69696 . To identify genomic loci enriched for CENP-ACse4 , we analyzed the coverage relative to the centromere . Although CENP-ACse4 is constitutively localized to the centromere [96] , its coverage at the centromere is under-represented relative to other genomic regions . This effect is likely due to the decreased solubility of the centromere to MNase digestion due to kinetochore protein binding , which makes it possible for other genomic regions to appear enriched above its occupancy at the centromere [54] . We called peaks of CENP-ACse4 occupancy in each strain as any region where the CENP-ACse4 enrichment was above the threshold of the minimum average coverage at any centromere in the 3Flag-CSE4 strain using R Bioconductor packages Genomic Ranges , rtracklayer , and the UCSC SacCer3 reference genome [56 , 90–92] and the DANPOS2 function Dtriple to call peaks without any further normalization or smoothing [93] . rDNA ChIP coverage was set to 0 before peak calling due to the high copy number of this region , and this locus was excluded from subsequent computational analyses . Input nucleosome peaks were also called using DANPOS2 [93] . Browser Extensible Data ( BED ) files of the called peaks for each sample are available at NCBI GEO Series GSE69696 . Genomic regions were annotated using the following strategy: Saccharomyces Genome Database ( SGD ) annotations of the SacCer3 genome were used to call regions of centromeres , pericentromeres , telomeres , origins of replication , genes , and intergenic regions in that order , such that each base was assigned to only the first overlapping region type . To analyze the percentage of peaks from each strain in each genomic region , 1 bp regions at the center of each CENP-ACse4 peak were overlapped with each region so that each peak was counted only once using R Bioconductor packages Genomic Ranges , rtracklayer , and UCSC SacCer3 [56 , 90–92] . The same analysis was performed with CENP-ACse4 peaks that either did or did not overlap WT H2A . ZHtz1 peaks . We analyzed mean CENP-ACse4 and H2A . ZHtz1 enrichment at the starts and ends of genes as well as centered on NDRs , origins of replication , or centromeres using the DANPOS2 profile function [65 , 93] . H2A . ZHtz1 ChIP data is from [4] . H2A . ZHtz1 coverage was calculated from the mapped reads with greater than 90% identity using the DANPOS2 function dpos with the default parameters [93] , after lifting over the coordinates to the SacCer3 genome using R Bioconductor packages Genomic Ranges , rtracklayer , and UCSC SacCer3 [56 , 90–92] . For the analysis of the transcription start sites ( TSS ) and transcription termination sites ( TTS ) , the mean CENP-ACse4 or H2A . ZHtz1 coverage in 10 bp windows was calculated for 500 bp upstream and downstream of 3987 transcripts using custom gene files modified to use experimentally derived TSS data instead of open reading frame ( ORF ) start sites from Nagalakshmi et al , 2008 ( GSE11209 ) [61] . For the analysis of specific groups of genes , the gene file was divided into the specified bins using R Bioconductor packages before using the DANPOS2 function . For NDRs , origins , and centromeres , DANPOS2 profile was run centered on the genomic features using bed files containing either each NDR [65] , origin ( from SacCer3 annotation ) or centromere ( from SacCer3 annotation ) . All plots were made using GraphPad Prism version 6 . 0 for OSX , GraphPad Software , La Jolla California USA , www . graphpad . com . Coverage data was visualized using IGV [94 , 95] . The fraction of overlap between CENP-ACse4 peaks for each strain and reported H2A . ZHtz1 peaks ( coarse grain nucleosome positions ) from wild-type ( WT ) cells [4] was calculated using R Bioconductor packages GenomicRanges and rtracklayer [90–92] . ChIP was performed from 50 ml formaldehyde-crosslinked cultures as described [97] . Chromatin was fragmented by sonication to approximately 500 bp fragments . For HA-Htz1 ChIP , 3HA-Htz1 was immunoprecipitated using anti-HA ( 12CA5 ) antibodies ( Roche ) . 3-fold serial dilutions of the Input ( 1:100 , 1:300 , 1:900 ) and ChIP ( 1:3 , 1:9 , 1:27 ) DNA were used for PCR reactions to detect the amount of DNA pulled down with 3HA-Htz1 in each strain [97] and were analyzed on 1 . 4% agarose gels . Primers for the RDS1 promoter are from [69] and are listed in S4 Table . For CENP-ACse4 ChIP , CENP-ACse4 was immunoprecipitated using anti-Cse4 ( 235N ) antibodies [6] from strains with CENP-ACse4 expression induced from an inducible tetracycline repressed promoter [98] after a 6-hour washout of doxycycline ( 5ug/ml ) in YC-URA media . 3-fold serial dilutions of the Input ( 1:100 , 1:300 , 1:900 ) and ChIP ( 1x , 1:3 , 1:9 ) DNA were used for PCR reactions at CEN3 [87] , the SAP4 promoter and the SLP1 promoter . Total RNA was extracted from each sample using a hot acid phenol extraction protocol [99] , followed by DNAse I treatment ( Invitrogen Amplification Grade ) phenol:chloroform extraction , and ethanol precipitation . Two or three independent biological replicates of each genotype were used . Total RNA integrity was checked using an Agilent 2200 TapeStation ( Agilent Technologies , Inc . , Santa Clara , CA ) and quantified using a Trinean DropSense96 spectrophotometer ( Caliper Life Sciences , Hopkinton , MA ) . RNA-seq libraries were prepared from total RNA using the TruSeq RNA Sample Prep v2 Kit ( Illumina , Inc . , San Diego , CA , USA ) and a Sciclone NGSx Workstation ( PerkinElmer , Waltham , MA , USA ) . Library size distributions were validated using an Agilent 2200 TapeStation ( Agilent Technologies , Santa Clara , CA , USA ) . Additional library QC , blending of pooled indexed libraries , and cluster optimization were performed using Life Technologies’ Invitrogen Qubit® 2 . 0 Fluorometer ( Life Technologies-Invitrogen , Carlsbad , CA , USA ) . RNA-seq libraries were pooled ( 18-plex ) and clustered onto a flow cell lane . Sequencing was performed using an Illumina HiSeq 2500 in “rapid run” mode employing a single-read , 50 base read length ( SR50 ) sequencing strategy . Image analysis and base calling was performed using Illumina's Real Time Analysis v1 . 18 software , followed by 'demultiplexing' of indexed reads and generation of FASTQ files , using Illumina's bcl2fastq Conversion Software v1 . 8 . 4 ( http://support . illumina . com/downloads/bcl2fastq_conversion_software_184 . html ) . Reads of low quality were filtered prior to alignment to the reference genome ( UCSC SacCer3 assembly ) using TopHat v2 . 1 . 0[100] . Counts were generated from TopHat alignments for each gene using the Python package HTSeq v0 . 6 . 1[101] . Genes with low counts across all samples were removed , prior to identification of differentially expressed genes using the Bioconductor package edgeR v3 . 12 . 0[78] . A false discovery rate ( FDR ) method was employed to correct for multiple testing[102] . Differential expression was defined as |log2 ( ratio ) | ≥ 0 . 585 ( ± 1 . 5-fold ) with the FDR set to 5% . Normalized differential expression data are available as excel files ( S3 File ) , and raw data is available under NCBI GEO Series GSE69696 . The overlap between lists of genes with significantly changed transcription compared to WT yeast in htz1Δ at t = 2 hours , and psh1Δ pGAL-3Flag-CSE4 at t = 2 hours—t = 0 vs . WT t = 2 hours—t = 0 was found using the Whitehead Institute Compare Two Lists tool ( http://jura . wi . mit . edu/bioc/tools/compare . php ) . The number of significantly up or down regulated transcripts overlapped between the genotypes was compared using the hypergeometric distribution ( p-value is probability of getting more than the observed number of successes ) using the total number of genes in the edgeR result files as the total population , using the GeneProf hypergeometric distribution calculator [103] . | Chromosomes carry the genetic material in cells . When cells divide , each daughter cell must inherit a single copy of each chromosome . The centromere is the locus on each chromosome that ensures the equal distribution of chromosomes during cell division . One essential protein involved in this task is CENP-ACse4 , which normally localizes exclusively to centromeres . Here , we investigated where CENP-ACse4 spreads in the genome when parts of its regulatory machinery are removed . We found that CENP-ACse4 becomes mislocalized to promoters , the region upstream of each gene that controls the activity of the gene . Consistent with this , the mislocalization of CENP-ACse4 to promoters leads to problems with gene activity . Our work shows that mislocalization of centromeric proteins can have effects beyond chromosome segregation defects , such as interfering with gene expression on chromosome arms . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"chromosome",
"structure",
"and",
"function",
"molecular",
"probe",
"techniques",
"centromeres",
"dna-binding",
"proteins",
"immunoblotting",
"dna",
"transcription",
"fungi",
"model",
"organisms",
"genome",
"analysis",
"epigenetics",
"molecular",
"biology",
"techniques",
... | 2016 | Regulation of Budding Yeast CENP-A levels Prevents Misincorporation at Promoter Nucleosomes and Transcriptional Defects |
Ovulation is critical for successful reproduction and correlates with ovarian cancer risk , yet genetic studies of ovulation have been limited . It has long been thought that the mechanism controlling ovulation is highly divergent due to speciation and fast evolution . Using genetic tools available in Drosophila , we now report that ovulation in Drosophila strongly resembles mammalian ovulation at both the cellular and molecular levels . Just one of up to 32 mature follicles per ovary pair loses posterior follicle cells ( “trimming” ) and protrudes into the oviduct , showing that a selection process prefigures ovulation . Follicle cells that remain after egg release form a “corpus luteum ( CL ) ” at the end of the ovariole , develop yellowish pigmentation , and express genes encoding steroid hormone biosynthetic enzymes that are required for full fertility . Finally , matrix metalloproteinase 2 ( Mmp2 ) , a type of protease thought to facilitate mammalian ovulation , is expressed in mature follicle and CL cells . Mmp2 activity is genetically required for trimming , ovulation and CL formation . Our studies provide new insights into the regulation of Drosophila ovulation and establish Drosophila as a model for genetically investigating ovulation in diverse organisms , including mammals .
Ovulation , the liberation of a mature oocyte from the ovary , is one of the critical events of metazoan reproduction . In mammals , where ovulation has been studied most thoroughly , several important steps have been identified [1–4] . First , among a cohort of mature ovarian follicles , a dominant follicle arises . Eventually , proteolytic enzymes are locally activated that digest a small part of the dominant follicle’s wall and extracellular matrix , releasing the oocyte into the oviduct [5] . Finally , residual follicular cells remodel the ruptured follicle into the yellowish corpus luteum , an endocrine body that secretes the steroid hormones progesterone , estrogen , and other factors . While much has been learned , genetically testing the roles proposed for specific genes and pathways has been difficult . For example , the importance of matrix metalloproteinases ( Mmps ) in ovulation has not been demonstrated using knockout mice , possibly due to redundancy [6–9] . A genetically tractable system containing fewer redundant genes such as Drosophila would greatly facilitate ovulation studies . However , ovulation in Drosophila has not been well characterized and is not known to involve the same processes as mammalian ovulation . The Drosophila female reproductive system is anatomically similar to mammals , having two ovaries connected by lateral and common oviducts to the uterus , where fertilization occurs and one egg is retained prior to laying ( Fig . 1A ) [10] . Ovulation does not follow a simple cycle , however . Multiple eggs are laid when suitable food resources are available [11] , and ovulation follows each oviposition to replenish the uterus . Egg laying and ovulation are extensively regulated by octopaminergic neural inputs [12–14] and can be elicited by peptides transferred in semen from the male [15–18] . Ovulation requires reproductive tract secretions controlled by the NR5a class nuclear hormone receptor Hr39 [19] . A mammalian ortholog , LRH-1 , is required in mouse granulosa cells for ovulation , to maintain progesterone production in the corpus luteum and for decidual cell function in the uterus [20 , 21] . These similarities highlight the potential value of Drosophila as a genetically tractable model of ovulation . Here , we show that a similar follicle rupturing process leads to Drosophila ovulation . Posterior follicle cells of a mature egg chamber are first degraded and the residual follicle cells are squeezed toward the anterior while the oocyte moves posteriorly into the lateral oviduct . Membrane-tethered Mmp2 , but not Mmp1 , functions in follicle cells to control follicle rupture and oocyte release . Residual follicle cells remain in the ovary , accumulate yellowish pigmentation , express ecdysone biosynthetic genes , and persist for an extended period before degradation , events reminiscent of the corpus luteum in mammals . Our data indicate that the cellular and molecular regulation of ovulation has been more conserved than previously thought .
The fate of Drosophila follicle cells after ovulation has not been clearly described . If ovulation involves a programmed rupture of the follicular layer as in mammals , then most follicle cells would remain behind in the ovary after the egg is released into the oviduct . Alternatively , follicle cells might degenerate randomly or some might accompany the oocyte into the oviduct and the uterus . Using a mature ( stage 14 ) follicle cell marker ( Fig . 1B , Methods ) , we observed a cluster of GFP-labeled cells at the posterior end of each ovariole in the basal ( calyx ) region of ovary ( Fig . 1C ) . In contrast , few , if any , follicle cells leave the ovary upon ovulation because GFP-labeled cells were not seen associated with eggs in the oviduct or uterus . Post-ovulation mammalian follicles transform into the corpus luteum and similar behavior was reported previously in several other insects [22] . Consequently , we termed each cluster of residual Drosophila ovarian follicle cells as a corpus luteum ( CL ) , and observed that CL cells continue a program of gene expression . The zinc-finger transcription factor Hindsight ( Hnt ) , a major follicle cell regulator [23] , is upregulated in stage 14 follicle cells and is expressed in CL cells ( Fig . 1D-E ) . Expression of the adherens junction protein Arm ( Fig . 1F-F’ ) , suggests that CL cells maintain apical-basal epithelial polarity . Most dramatically , CL cells acquire a yellowish pigmentation ( Fig . 1G-I ) and express the ecdysone biosynthetic enzymes [24 , 25] Shade ( Shd ) in mitochondria ( Fig . 1J-L ) and Phantom ( Phm ) in the endoplasmic reticulum ( ER; S1A Fig . ) . To examine the functional significance of steroid biosynthetic gene expression , we knocked down expression of shd and phm , as well as dib ( disembodied , encoding another enzyme in the ecdysone synthesis pathway ) , in CL and mature follicle cells . These females all laid significantly fewer eggs than control females ( S1B Fig . ) . This functional requirement for steroid biosynthetic enzymes supports the view that the Drosophila CL and mature follicle cells produces a steroid hormone such as ecdysone or 20-hydroxyecdysone , consistent with a recent finding that follicle cells in middle stages of oogenesis produce ecdysone to regulate border cell migration [26] and also reminiscent of steroid hormone production by the mammalian corpus luteum . We observed that only one CL is present in each ovariole , hence each CL must either degrade in situ over time following ovulation , or be extruded from the ovariole when the next egg is ovulated . The organization of the CL reflects its origin in the follicle . All CL cells are labeled by a mature follicle cell driver , suggesting that cells from other sources are absent ( Fig . 2A ) . Little cellular rearrangement occurs , since only anterior or middle cells of the corpus luteum were labeled by lines specifically expressed in anterior or middle stage 14 follicle cells ( Fig . 2B-C ) . Lines specifically expressed in the posterior follicle cells did not label the CL , suggesting that these cells were degraded during ovulation ( Fig . 2D ) . Drosophila ovaries each usually contain at least 15 mature follicles , one per ovariole , oriented with their posterior ends facing the oviduct , raising the question of how one particular follicle is selected for ovulation . We examined ovary pairs from females cultured under conditions favorable for egg laying and found that at most one mature follicle protrudes significantly into a lateral oviduct ( Fig . 2E ) . The protruding follicle always lacked posterior follicle cells covering the part of egg inside the lateral oviduct ( Fig . 2F ) . We termed this process of losing posterior follicle cells as “trimming” . The trimmed follicle’s location indicates that trimming and protrusion represent preludes to ovulation . Frequently , another stage 14 follicle was present that had lost a smaller area of posterior follicle cells , but did not protrude ( Fig . 2F-G and S2A-C Fig . ) , which likely represents the next follicle to ovulate . These observations show that a follicle is preselected in Drosophila well before ovulation , undergoes trimming , and awaits the next ovulation event while protruding into the lateral oviduct . In flies that were laying few eggs , for example in unmated females , up to 6 trimmed follicles could be present per female ( Fig . 2G ) , but the follicle with the greatest level of trimming continues to protrude into the oviduct and remains in a position poised for ovulation . The study of explanted mammalian follicles strongly implicates matrix metalloproteinases ( Mmps ) as important contributors to oocyte release [5] . In these follicles , Mmp activity is localized to the apex region where rupture will later occur [6] . We carried out gelatinase assays in situ to measure localized Mmp activity within Drosophila follicles before and during ovulation . High localized activity was found at the posterior end of one mature follicle in each ovary pair while a second follicle sometimes had lesser activity ( Fig . 3A ) ; the location of the activity correlated with the site of follicle cell trimming at the posterior ( Fig . 3B ) . As eggs begin to enter the oviduct , the fraction of the follicular surface with gelatinase activity ( Fig . 3C ) increased from posterior to anterior , and matched where follicle cells no longer reside ( Fig . 3C’ ) . In eggs that have nearly separated from their follicle cells , gelatinase activity covered the entire surface ( Fig . 3D ) , however , the anterior and middle follicle cells remained in a mass at the base of the ovary ( Fig . 3D’ ) . These data tightly associate Mmp activity with posterior follicle cell trimming , and suggest that more anterior Mmp activity subsequently degrades just the extracellular matrix separating the oocyte from intact middle and anterior follicle cells . Drosophila has two genes encoding matrix metalloproteinases , mmp1 and mmp2 , and one Timp ( Tissue inhibitor of matrix metalloproteinase ) [27] . Genetically reducing Mmp2 but not Mmp1 expression dramatically lowered egg laying ( Fig . 3E; S3 Fig . ) . Mature egg chambers accumulated in Mmp2 knockdown females ( Fig . 3F ) , indicating a block in ovulation , and the average time required to ovulate ( see Methods ) increased fivefold ( Fig . 3G and Table 1 ) . Similarly , overexpressing Timp , a protein that inhibits both Mmp1 and Mmp2 activity , also decreased egg laying and increased egg retention and ovulation time ( Fig . 3E-G , Table 1 ) . These data show that Mmp2 enzymatic activity is required for normal ovulation in Drosophila . Mmp2 was also required for follicle trimming ( Fig . 3H-I and Table 2 ) . The rate of trimming was reduced at least three fold in Mmp2 knock down animals both before mating and at 6-hour post mating ( Fig . 3J and Table 2 ) . In addition , Mmp2 knock down ovaries lacked corpus luteum structures ( Fig . 3K-L ) , and instead accumulated mature egg chambers ( Fig . 3L ) . Both Mmp2 knock down females and females expressing Timp , displayed severe egg laying defects even within 6 hours of mating ( Fig . 3M ) . Thus , Mmp2 activity is required in adult females for follicle cell trimming , ovulation , corpus luteum formation , and egg laying . We generated an in vivo Mmp2::GFP fusion allele at its normal genomic location by swapping an in-frame GFP exon into a MiMIC transposon inserted within an Mmp2 coding intron ( S4 Fig . ) . We also employed a Gal4 enhancer trap line ( see Methods ) , which mimics Mmp2 expression during pupal imaginal disc eversion , to monitor Mmp2 transcription . Mmp2 fusion protein and RNA are specifically expressed in posterior follicle cells in all mature stage 14 follicles but not in earlier follicles ( Fig . 4A-C , S4C Fig . and S5A-D Fig . ) . Mmp2 is also expressed in some anterior follicle cells that help form dorsal eggshell structures . The reporters show expression at the posterior edge of surviving follicle cells during trimming , and in anterior and posterior corpus luteum cells ( Fig . 4D-E and S5D Fig . ) . We interfered with Mmp2 expression specifically in mature follicle cells by using a mature follicle cell driver ( R47A04 ) to express Mmp2 RNAi or to overexpress Timp , and observed that ovulation and egg laying were defective ( Fig . 4F-G and Table 1 ) . The defect is not likely due to disruption of Mmp2 in neurons as R47A04 is not expressed in sensory neurons innervating the female reproductive tract ( S5E-F Fig . ) . Action outside of neurons is also supported by the observation that knocking down Mmp2 with a more restricted mature follicle cell driver ( R42A05: expressed in posterior and anterior mature follicle cells; Fig . 2D ) showed a similar egg laying defect ( Table 1 ) , although one of lower severity . When Mmp2 was overexpressed in mature follicle cells , mature eggs ruptured and were released into the female abdominal cavity ( Fig . 4H-I ) . Most such eggs lacked covering follicle cells ( Fig . 4J-K ) . When Mmp1 was ectopically expressed in mature follicle cells with the same Gal4 driver , fewer follicles ruptured into the abdominal cavity and most released eggs retained some follicle cells ( Fig . 4K ) . Consequently , Mmp2 is required in mature follicle cells to trim the follicular layer leading to ovulation , and its level of expression must be controlled or normal activity regulation may be overwhelmed .
Despite different biological strategies of ovulation in mice and Drosophila , our studies reveal strong similarities in the underlying mechanisms . In both species , a follicle is selected before ovulation , and its oocyte is released at an appropriate time by inducing Mmp proteolytic activity , either in the apex region ( mouse ) or at the follicle posterior ( Drosophila ) . Mmp2 activation is likely controlled by pro-domain processing , and may also be modulated at the level of protein secretion and/or by the presence of the endogenous inhibitor Timp . Pharmacological inhibition of Mmp activity prevents ovulation in vitro in mice [28 , 29] , as well as in other vertebrate and primate species [30 , 31] . However , knockouts of individual Mmp genes have not been reported to affect ovulation , presumably due to redundancy , although individual Mmp gene knockouts frequently have specific phenotypic effects ( reviewed in [9] ) . In contrast , our studies clearly show a genetic requirement of Mmp2 but not Mmp1 for trimming , ovulation and CL formation . However , follicles from MMP2 RNAi females still retained some gelatinase activity based on the in situ assay , and our experiments cannot rule out that MMP1 or other proteases also contribute to follicle trimming . The value of Drosophila for studies of ovulation is further illustrated by the discovery that after ovulation , residual follicle cells form a corpus luteum . The corpus luteum ( Latin for “yellow body” ) was first described by Volcherus Coiter in 1573 , but its relationship to ovulation rather than pregnancy was not understood until the early 19th century [32] . The existence of a pigmented structure in insect ovaries was also recognized in the 19th century , at least in a few species [22] . However , it has remained unclear whether a CL exists in Drosophila , whether it is a universal feature of insect oogenesis , whether the CL functions in reproduction , and whether any such functions have been conserved during evolution . Our studies indicate that a CL is formed in Drosophila and that Mmp2 activity is required for its production . The mammalian CL contains at least two cell types , small CL cells which are thought to be derived from thecal cells , and large CL cells that produce progesterone . Our gene expression studies suggest that at least two cell types are also likely in the Drosophila CL [33] . Some anterior CL cells may derive from stretch cells , which never secrete eggshell proteins [34] . The Drosophila CL may function at least in part by producing the steroid hormones , ecdysone or 20-hydroxyecdysone , as suggested by continuous expression of genes encoding P450 enzymes Phantom and Shade in CL cells . Mated females are known to have a higher ecdysone titer than unmated females [35] , consistent with the idea that the CL contributes substantially to ecdysone production . A common role in steroid hormone production might explain the conserved pigmentation of the CL . In mammals , carotenoid accumulation is beneficial to gametogenesis and is associated with steroid hormone production . These molecules may influence free radical balance , which might otherwise interfere with steroid production [36] . Alternatively , carotenoids may simply accumulate because they are found within the circulating lipoprotein particles that must be taken up to support steroid production [37] . The easy of genetic manipulation in Drosophila may allow the biochemical nature and function of the yellow pigmentation in the CL to be further characterized . Finally , we propose that a major function of the CL in Drosophila is to control the maturation of younger follicles in the ovariole , and to select mature follicles for the next ovulation . In their location at the posterior end of each ovariole , CL cells are well positioned to govern the orderly usage of mature follicles . If CL cell secretory activity decreases with age , the corpus luteum in each ovariole might communicate the elapsed time since an ovariole was last used , for example by local inhibition , promoting the relatively uniform usage of all ovarioles . Although the large scale organization of follicles within the mammalian ovary is less obvious , signals from its corpora lutea might likewise spatially control follicle maturation . Knowledge that some fundamental aspects of ovulation are similar in Drosophila and mammals will accelerate the study of these and many other questions .
Flies were reared on standard cornmeal-molasses food at 25°C unless otherwise indicated . The following Gal4 lines from the Janelia Farm collection [38] were used to label follicle cells and corpus luteum cells: R47A04 ( Oamb ) , R49E12 ( 5-HT2A ) , R10E05 ( AstC-R2 ) , and R42A05 ( kay ) . To knockdown mmp1 or mmp2 or overexpress Timp in adult flies , actGal4/Cyo; tubGal80ts virgin females were crossed to the following lines at 18°C and shifted to 29°C immediately after adult eclosion: UAS-mmp1RNAi ( Bloomington Drosophila stock center , B31489 ) , UAS-mmp1RNAi2 [39] , UAS-mmp1RNAi3 ( Vienna Drosophila RNAi Center , V108894 ) , UAS-mmp1E225A ( a dominant negative form of Mmp1 ) [40] , UAS-mmp2RNAi [39] , UAS-mmp2RNAi2 ( VDRC , V107888 ) , UAS-mmp2RNAi3 ( BDSC , B31371 ) , UAS-Timp [41] . To knock down mmp1 or mmp2 or overexpress Timp in follicle cells of mature egg chambers , UAS-dcr2; R47A04 virgin females were crossed to the RNAi lines described above at 29°C . To knock down ecdysone synthesis genes , UAS-dcr2; R47A04 virgin females were crossed shdRNAi ( VDRC , V17203 ) , dibRNAi ( VDRC , V101117 ) , or phmRNAi ( VDRC , V108359 ) . To overexpress mmp1 or mmp2 in mature follicle cells , R47A04 virgin females were crossed to UAS-mmp1 or UAS-mmp2 [41] at 21°C . Control flies were derived from specific Gal4 driver crossed to wild-type Oregon-R . Mmp2::GFP fusion genes were generated through recombinase mediated cassette exchange of MiMIC insertion ( MI02914 ) in the third coding intron of mmp2 ( S4 Fig . ) [42] . Mmp2-Gal4 line is from an Gal4 enhancer trap [43] , and UAS-RedStinger ( BDSC , B8547 ) and UASpGFP-act79B; UAS-mCD8-GFP were used as reporters . sqh-EYFP-Mito ( BDSC , B7194 ) and sqh-EYFP-ER ( BDSC , B7195 ) were used for tracking mitochondria and endoplasmic reticulum , respectively . Egg laying and ovulation was determined essentially as described [19] . Virgin females were aged for four to five days , and fed with wet yeast one day before experiments . To measure egg laying time ( the average time between successive ovipositions ) , five females were mated to 10 Oregon-R males in each bottle covered with a molasses plate at 29°C and five or more bottles were set up for each genotype . The molasses plates were replaced every 22 hours and the number of eggs laid was counted for 44 hours and used to determine the average time required per egg . Egg laying time ( min ) was then calculated as 22 hr x 60 min / eggs laid per 22hr . To determine ovulation time , about 30 single-pair matings with one virgin female and one Oregon-R male were carried out for each genotype at 29°C for 6 hours , an interval sufficient for all females to reach a steady state level of ovulation and egg laying . Females were then frozen in -80°C for four minutes , their reproductive tracts were dissected to identify eggs inside the reproductive tract , and the percentage of females with an egg in the uterus or actively ejecting out of the uterus ( U% ) was calculated . Free eggs were never observed in the common oviducts in control or mutant flies ( N = 196 ) , indicating that eggs spend a negligible amount of time moving through the oviducts . In addition , we never observe more than one egg in the female reproductive tract ( N = 196 ) . Therefore , the egg laying time is partitioned into the uterus time ( the average time eggs reside in the uterus or actively ejecting ) and the ovulation time ( the average time eggs prepare to be released from the ovary ) , which includes the time when the follicle protrudes into the lateral oviduct , because our study indicates that these eggs are in the process of ovulation and have not yet been released from the ovary ( Fig . 2E-F , 3D , and 4E ) . Uterus time was then calculated as egg laying time x U% , and ovulation time = egg laying time x ( 1—U% ) . The ovulation time is a proxy for the average time needed for an egg to be released from the ovary while flies are laying eggs rapidly at steady state . The ovulation process is complex and is known to be affected by many factors , such as glandular secretions , male peptides , moisture , and egg laying substrates , whose influences are aggregated using this approach . The 95% confidence intervals were calculated correspondingly . Females were frozen in -80°C for four minutes before dissection . Ovaries were dissected out immediately afterwards , fixed with 4% EM Grade paraformadehyde , and stained with DAPI . Care was taken to make sure that two ovaries from single female were intact after staining and mounted in the same well on slides by carefully separating the ovarioles . The number of trimming follicles was scored according to the criteria that a quarter of the egg chamber at the posterior end has no follicle cells covering , and the number of mature follicles was scored according to their fully elongated dorsal appendage . The normalized trimming follicles were then calculated by the number of trimming follicles divided by the number of mature follicles in each female . For follicle cell trimming analysis in Fig . 2G , three to four days old w1118 mated or unmated females were used . For analyzing trimming in females with Mmp knock down , three to four days old virgin females were mated with Oregon-R males for six hours before dissection . For trimming with misexpressing mmp1 and mmp2 , follicles were directly collected from female abdominal cavity . Antibody staining followed a standard procedure [44] . Briefly , tissues were fixed in 4% EM-Grade paraformadehyde for 15 minutes and blocked in PBTG ( PBS+ 0 . 2% Triton+ 0 . 5% BSA+ 2% normal goat serum ) . Primary antibodies were incubated overnight at 4°C and secondary antibodies were incubated for two hours at room temperature , followed by DAPI staining . The following primary antibodies were used: mouse anti-Hnt ( 1: 75 ) and anti-Arm ( 1:40 ) from Developmental Study Hybridoma Bank , rabbit anti-Shd ( 1:250; a gift from Michael O’Connor ) and anti-Phm [25] ( 1:250 ) , rabbit anti-GFP ( 1:4000 , Invitrogen ) , and rabbit anti-RFP ( 1:2000 , MBL international ) . Secondary antibodies were Alexa 488 goat anti-rabbit and 546 goat anti-mouse and anti-rabbit ( 1:1000 , Invitrogen ) . Images were acquired using the Leica TCS SP5 confocal microscope , and assembled using Photoshop software ( Adobe , Inc . ) . Images for yellow pigmentation of the corpus luteum were taken with the Macropod with Canon 6D camera and Olympus SZX16 stereomicroscope with Olympus DP72 color camera . The in situ zymography technique for gelatinase activity was performed as previously described with minor modifications [45] . Ovaries were dissected in pre-warmed Grace’s media and incubated immediately in 100 μg/ml DQ-gelatin conjugated with fluorescein ( Invitrogen ) for an hour . To increase substrate penetration , the peritoneal muscle sheath was broken at the ovarian anterior . Ovaries were then fixed in 4% EM-Grade Paraformadehyde for 15 minutes and mounted for visualization . | Sexual reproduction is thought to be a highly divergent process due to fast evolution and speciation . For example , sperm from one species can seldom fertilize eggs from another species , indicating that different molecular machinery for fertilization is applied in different species . In contrast to this divergent view , ovulation , the process of liberating mature eggs from the ovary , is a general phenomenon throughout the Metazoa . We provide evidence that basic mechanisms of ovulation are conserved . Like mammalian follicles , Drosophila follicles consist of single oocytes surrounded by a layer of follicle cells . Drosophila follicles degrade their posterior follicle cells to allow the oocyte to rupture into the oviduct during ovulation . The residual postovulatory follicles reside in the ovary , accumulate yellowish pigmentation , and produce the steroid hormone ecdysone , features which resemble the mammalian corpus luteum . We also showed that matrix metalloproteinase , a type of proteinase proposed to degrade the mammalian follicle wall during ovulation , is required in Drosophila for posterior follicle cell degradation and ovulation . These findings are particularly important because this simple genetic model system will speed up the identification of many conserved regulators required for regulating matrix metalloproteinase activity and ovulation in human , processes that influence ovarian cancer formation and cancer metastasis . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Matrix Metalloproteinase 2 Is Required for Ovulation and Corpus Luteum Formation in Drosophila |
Activation-induced cytidine deaminase ( AID ) is required for initiation of Ig class switch recombination ( CSR ) and somatic hypermutation ( SHM ) of antibody genes during immune responses . AID has also been shown to induce chromosomal translocations , mutations , and DNA double-strand breaks ( DSBs ) involving non-Ig genes in activated B cells . To determine what makes a DNA site a target for AID-induced DSBs , we identify off-target DSBs induced by AID by performing chromatin immunoprecipitation ( ChIP ) for Nbs1 , a protein that binds DSBs , followed by deep sequencing ( ChIP-Seq ) . We detect and characterize hundreds of off-target AID-dependent DSBs . Two types of tandem repeats are highly enriched within the Nbs1-binding sites: long CA repeats , which can form Z-DNA , and tandem pentamers containing the AID target hotspot WGCW . These tandem repeats are not nearly as enriched at AID-independent DSBs , which we also identified . Msh2 , a component of the mismatch repair pathway and important for genome stability , increases off-target DSBs , similar to its effect on Ig switch region DSBs , which are required intermediates during CSR . Most of the off-target DSBs are two-ended , consistent with generation during G1 phase , similar to DSBs in Ig switch regions . However , a minority are one-ended , presumably due to conversion of single-strand breaks to DSBs during replication . One-ended DSBs are repaired by processes involving homologous recombination , including break-induced replication repair , which can lead to genome instability . Off-target DSBs , especially those present during S phase , can lead to chromosomal translocations , deletions and gene amplifications , resulting in the high frequency of B cell lymphomas derived from cells that express or have expressed AID .
Activation-induced cytidine deaminase ( AID ) is required for initiation of somatic hypermutation ( SHM ) of Ig variable region genes and class switch recombination ( CSR ) of IgH genes in B cells during an immune response [1 , 2] . Both SHM and CSR are required for effective humoral immune responses , and thus humans ( and mice ) lacking AID are severely immunocompromised . AID deaminates cytosines ( dC ) in expressed Ig variable region genes and in IgH switch ( S ) regions , converting dC to uracil ( dU ) , which can then be replicated by DNA polymerase ( Pol ) to form dC>dT mutations . Alternatively , the dU base is excised by uracil DNA glycosylase ( primarily Ung ) , which leaves an abasic , or apyrimidinic/apurinic ( AP ) site [3 , 4] . AP sites cannot be copied by high-fidelity DNA Pol , but can serve as templates for error-prone translesion DNA Pols , which insert any base across from the AP site . Alternatively , AP sites are incised by AP-endonucleases ( Ape1/Ape2 , also termed Apex1/Apex2 ) to create single-strand DNA breaks ( SSBs ) . If SSBs on opposite strands are sufficiently near each other , they form a double-strand break ( DSB ) . If they are farther apart , they can still generate DSBs with the help of the mismatch repair ( MMR ) system , after recognition of a dU:dG mismatch by Msh2-Msh6 , followed by excision of one strand from a nick created by Ape1/2 [5] . During CSR , AID-dependent DSBs are induced within IgH S regions , which are highly enriched in the AID target hotspot , WGCW , in which W is A or T , and the C on both strands is a hotspot target , thus increasing the probability of AID-induced SSBs leading to DSBs . For unknown reasons , AID acts predominantly on Ig genes in activated B cells , although it can act at other sites in the genome with reduced frequency . This was first demonstrated by the finding of AID-dependent mutations in several actively transcribed non-Ig genes in germinal center B cells , where AID is highly expressed and SHM of Ig genes occurs [6–11] . In addition , AID has been demonstrated to instigate off-target DSBs and chromosomal translocations in B cells induced to undergo CSR in culture [12–22] . Chromosomal deletions , duplications , and translocations are found in human B cell lymphomas and gastric and prostate cancers , many of which might be instigated by AID [23–25]; thus , it is important to understand what causes non-Ig chromosomal sites to become susceptible to AID-dependent DSBs . Furthermore , what causes some off-target sites mutated by AID to progress to DSBs is unknown . Genome-wide AID-dependent DSBs have been detected in mouse splenic B cells undergoing CSR by using Nbs1-ChIP followed by hybridization to tiling arrays of the entire genome ( ChIP-chip ) [15] . Nbs1 has been shown to bind AID-dependent DSBs , most strongly at the IgH Sμ region , which is the upstream/donor S region for most CSR events [15 , 26] . CSR occurs by non-homologous end-joining ( NHEJ ) in the G1 phase of the cell cycle [5] . Consistent with this , Ku70-Ku80 and DNA-PKcs bind to S region DSBs , and cells deficient in these NHEJ proteins show reduced CSR [27 , 28] . Recent results suggest that during CSR , blunt or nearly blunt DSBs are recombined by NHEJ , but those with longer 3’ ss tails recombine using micro-homology-mediated end-joining , also termed alternative-end joining ( A-EJ ) [29] . The Mre11-Rad50-Nbs1 ( MRN ) complex and CtIP are important for end-resection during A-EJ , which also occurs during G1 phase [29–34] . Ku binding at DSBs is transient , as Ku slides away from DSB ends [35] , and Ku80 is rapidly ubiquitinated by RNF8 [36] . MRN could subsequently bind DNA ends that are not rapidly recombined by NHEJ , perhaps because they do not have the correct blunt structure . A-EJ , rather than NHEJ , has been shown to be involved in AID-dependent chromosomal translocations in mouse cells [37–39] . Homologous recombination in G2 phase cells also involves MRN , with more extensive end-resection by CtIP [40 , 41] . By using Nbs1 ChIP , our screen could be biased towards detecting off-target DSBs that are not immediately repaired/recombined , and are therefore capable of causing genomic instability . In this study , we identify off-target AID-dependent DSBs in mouse splenic B cells induced to switch in culture using Nbs1 ChIP-Seq , as this allows a more precise determination of the Nbs1-binding sites than does ChIP-chip . The Nbs1-binding sites separate into different classes , 66–70% are within genes/regions transcribed by RNA polymerase II ( Pol II ) , many contain tandem repeats of the AID hotspot target motif , WGCW , and others have tandem CA repeats but very few AID hotspots , and most are two-ended DSBs but a minority are one-ended , indicating they were generated by replication . Our data suggest that whether an AID-induced deamination progresses to a SSB , and then on to a DSB , is highly dependent upon its sequence context , and we have identified sites where AID-induced mutations are prone to generate DSBs .
To verify that the off-target AID-dependent Nbs1-binding sites are located at AID-dependent DSBs , we performed LM-PCR for several of the sites , using activated B cell DNA from two or more biologically independent experiments . We examined 6 of the 37 reproducible sites and 7 that were detected only in Exp 1 . Eleven of these 13 Nbs1-binding sites showed AID-dependent DSBs in at least two independent experiments ( Figs 2 and S1–S4; S1–S3 Tables ) . The cultures used for the LM-PCR experiments were independent of those used for the Nbs1 ChIP-seq experiments , suggesting that most of the AID-dependent DSBs are reproducible , despite the fact that they were not detected by Nbs1-ChIP in both experiments . Although Ig Sμ DSBs are detected reproducibly by LM-PCR in populations of B cells undergoing CSR , 50–150 cell-equivalents of genomic DNA are required to detect one Sμ DSB , suggesting they are present in only a small proportion of the cells at any one time [45 , 46] . Sμ DSBs are reproducibly detected in our ChIP-chip and ChIP-Seq experiments , including a few experiments that we do not include in this report . The weaker Nbs1 signals and fewer DSBs detected in LM-PCR assays of the off-target sites , relative to Sμ indicate that off-target DSBs are much less frequent . To detect one Sγ3 DSB in switching cells in our LM-PCR requires approximately 350–1100 cell-equivalents of genomic DNA . As Sγ3 DSBs are at the borderline of detection by Nbs1 ChIP-Seq , this suggests that the reproducible off-target DSBs are present in a somewhat greater proportion of cells than Sγ3 DSBs at any one moment . This low frequency could explain why two of the 13 Nbs1-binding sites tested by LM-PCR assay did not show AID-dependent DSBs . Examining strand specificity of the aligned tags provided further evidence that Nbs1 binding sites correspond to DSBs . Note that in the browser tracks of off-target sites shown in Fig 2A and 2B , the minus strand tags are located to the left of the plus strand tags . This is different from what is observed in ChIP-Seq data for transcription factors , where the plus strand tags are located to the left of the minus strand tags , as diagrammed in Fig 3A . In contrast , ChIP for proteins that bind at either side of a DSB should lead to the pattern observed in Figs 2A , 2B , S1 and S2 , as diagrammed in Fig 3B and further explained in the figure legend . This pattern is reproducibly found at nearly all AID-dependent binding sites , unless there is a broad peak of Nbs1-binding , indicating numerous DSBs , which obscures this pattern ( S3 and S4 Figs ) . This asymmetric pattern was also seen in most of the reproducible AID-independent sites , indicating these are also true DSBs ( browser views available in the GEO database accession #GSE66424 ) . The LM-PCR results and the strand-specific positions of the aligned tags relative to the called Nbs1 peaks indicate that most of the Nbs1-binding sites are indeed DSBs . AID-dependent Sμ DSBs are generated and repaired/recombined during G1 phase [26 , 45 , 46] . Interestingly , ~6% of the AID-dependent DSBs ( Table 1; example shown in Fig 2C ) have tags that align on only one of the two strands , consistent with the pattern expected if the DSB is one-ended , as would be generated when DNA Pol encounters a SSB during replication . As a comparison , we performed the same analysis for Pol II binding sites and found less than 1 in 104 sites have similarly skewed tags ( S5 Fig ) . The one-ended DSBs are probably generated during S phase , suggesting that a small portion of off-target AID-dependent DSBs form when a SSB enters S phase . AID-dependent SSBs should rarely be introduced during S phase as Ung activity is restricted to G1 phase in activated B cells [53] . Two of the 4 one-ended reproducible AID-dependent DSBs are one-ended in only one of the two experiments . This suggests that some AID-dependent lesions can become DSBs within G1 phase in some cells , or be converted to DSBs by replicative Pol in other cells . DSBs generated by DNA Pol encountering a SSB would cause the replication fork to arrest . One-ended DSBs are usually repaired by homology-directed repair , explaining why B cells treated with an inhibitor of RAD51 or deficient in XRCC2 , a protein important for homologous recombination , show unrepaired off-target AID-dependent DSBs [14 , 54 , 55] . Break-induced replication , a type of homologous recombination , is often used to repair one-ended DSBs , and this can lead to duplications , deletions , and inversions [56] . When homologous recombination is impaired , NHEJ might attempt to repair the one-ended DSB , and this can also result in gross chromosomal rearrangements [57] . Canonical MMR is important for correcting mutations introduced during DNA replication in S phase . However , MMR is also important for formation of Ig Sμ DSBs in G1 phase , as Sμ DSBs are decreased by 50–80% in MMR-deficient B cells [45 , 58–60] . MMR is especially important for generating DSBs in Ig switch regions where the AID hotspot target sequence is not abundant , such as when the Sμ tandem repeat region has been deleted [58] . We asked if off-target AID-dependent DSBs are also dependent upon MMR in LM-PCR experiments using genomic DNA from msh2-/- cells , and found that all of the AID-dependent DSBs analyzed are reduced in frequency in Msh2-deficient cells ( Figs 2 and S2–S4 ) . Although Msh2 primarily protects against human B cell lymphoma [60–62] , our data suggest that , in some cases , Msh2 might contribute to DSBs that could be associated with lymphomas initiated by AID activity . Msh2-deficient mice have been reported to have increased T cell but not B cell lymphomas , although Msh6-deficient mice develop both B and T cell lymphomas [63 , 64] . Table 1 summarizes additional characteristics of the 37 reproducible AID-dependent Nbs1-binding sites , the AID-dependent DSBs detected in Exps 1 and 2 , and reproducible AID-independent sites . For these analyses , the Nbs1 site called was extended by 1 kb on both sides of the peak center . This was done because Nbs1 has been shown by ChIP to bind within 1 kb of a defined DSB [65] . AID only targets Ig genes that are transcriptionally active , and in AID ChIP-Seq experiments performed in B cells induced to switch , the off-target AID-binding sites were mostly in transcribed genes [16] . As shown in Table 1 , 70% of the reproducible AID-dependent Nbs1 binding sites and almost as many of the AID-dependent sites in the individual experiments are transcribed , as evidenced by the binding of Pol II at the site or within the gene in which the site is located . This result is similar to that obtained in the Nbs1 ChIP-chip study [15] . Note that some of the sites that bind Pol II are not in annotated genes ( for example , Fig 2A ) . Interestingly , all of the reproducible AID-independent Nbs1 binding sites have Pol II binding ( Table 1 ) , indicating that transcriptionally active regions are prone to DSBs . It is possible that the 30% of AID-dependent sites that do not have detectable Pol II binding have very low levels of transcription or are transcribed by RNA Pols I or III , although we cannot rule out the possibility that ssDNA , the substrate for AID can be generated by means other than transcription , as discussed below . The reproducible AID-dependent Nbs1-binding sites are highly enriched in tandem repeats of WGCW , the AID hotspot target , relative to reproducible AID-independent sites and random sequences of the same lengths and chromosome distributions ( Table 2; Fig 4 ) . In fact , 46% of the AID-dependent off-target reproducible sites contain WGCW repeats that are at least 400 bp in length ( Fig 4A ) . Although this motif is found at some of the reproducible AID-independent Nbs1 sites , they are fewer and the lengths of the repeats much shorter ( median values: 1000 bp vs 100 bp , for reproducible AID-dependent and–independent sites , respectively ) . Also remarkable is that the density of the WGCW repeats ( WGCW motifs per 100 bp ) is much greater in AID-dependent sites than in the AID-independent sites ( Fig 4B ) . As a comparison , in Sμ there are 19 WGCW motifs per 100 bp , and this same density is present in 43% of the reproducible off-target AID-dependent sites . In the off-target sites , the motif is a 5 bp motif , just as in Sμ , although the most common sequence of the motif is CAGCA , slightly different from Sμ , where it is GAGCT . As these motifs create AID target hotspots on both strands , this provides an attractive explanation for why reproducible AID-dependent DSBs are found at these tandem repeats . About one-third of the reproducible AID-dependent DSBs contain a different tandem repeat , CA repeats at least 100 bp in length . The frequency of CA repeats at these sites is highly increased relative to that in random sequences ( 30% vs 1% ) ( Fig 4C ) ( Table 2 ) . The median length of the repeats in reproducible AID-dependent sites is ~315 bp . CA repeats ( ≥100 bp in length ) are also found at AID-independent Nbs1-binding sites , although much less frequently ( 7% of the sites ) . CA repeats greater than 30 bp in length can form unstable Z-DNA , a left-handed helix [66] . Due to the instability of this Z-DNA , it transitions between Z and B DNA; during the transition ss DNA might be accessible to AID . In addition , two bases are extruded from the helix at the junctions of Z and B DNA [67 , 68] . It is possible that CA repeats form ss DNA targets for AID , leading to SSBs , which are converted to DSBs by nuclease specific for structurally aberrant DNA , or perhaps during attempts to repair AID-induced lesions . Although CA repeats can lead to replication errors , this does not seem likely to explain their role in creating off-target AID-dependent DSBs since Ung activity , which is essential for nearly all AID-dependent SSBs and DSBs , is limited to G1 phase in activated B cells [53] . Other types of repeats , besides WGCW and CA , are not significantly enriched in the AID-dependent sites relative to AID-independent sites ( Fig 4D ) . Also , at the reproducible AID-dependent sites there is no enrichment of inverted repeats , although they have been shown to cause genomic instability [69] . Although only a few ( 4 ) of the AID-dependent Nbs1 ChIP-Seq sites correspond with the reproducible AID-dependent Nbs1-binding sites previously detected by ChIP-chip [15] , a high proportion of the AID-dependent ChIP-Seq sites were identified as AID-dependent sites in one of the two ChIP-chip experiments ( Table 1 ) . To make this comparison we chose the ChIP-chip experiment with the higher signal-to-noise ratio and a total of 54 , 976 AID-dependent peaks called by NimbleScan Find-Peaks ( Roche ) . The NimbleScan peak calls showed better correspondence with the AID-dependent ChIP-Seq sites than those produced by the Tamalpais peak caller used in ref [15] . Of the reproducible AID-dependent ChIP-Seq sites , 32% coincided with AID-dependent sites in the ChIP-chip experiment ( Table 1 ) . Two examples of intersecting sites are shown in S6 and S7 Figs . The AID-dependent ChIP-chip sites originally reported were also highly enriched in CA repeats and WGCW motifs [15] . Although the correspondence between the ChIP-Seq and ChIP-chip results is high , it is clear that our Nbs1-ChIP libraries are not saturated . As shown in Table 1 , a significant portion of the AID-independent sites identified by ChIP-Seq also intersected with the AID-dependent ChIP-chip sites , suggesting that some of the AID-independent sites identified by ChIP-Seq might actually be weak AID targets . However , as a group the AID-independent sites have different properties from the AID-dependent sites , as discussed above . Approximately 25% of the AID-dependent DSBs correspond to previously-identified AID-binding sites in cells induced to switch with LPS+IL-4 [16] , and the correspondence is highly significant compared with random sequences ( Table 1 ) . Surprisingly , the reproducible AID-independent sites show an even higher correlation with AID-binding than AID-dependent sites , perhaps because the AID-independent breaks are all found at Pol II binding sites or in genes with Pol II-binding sites , and because ChIP favors transcriptionally active accessible chromatin regions . AID interacts with Spt5 , a factor associated with paused RNA Pol II , and Spt5 is thought to be important for recruiting AID to the genome [10] . Thousands of Spt5 binding sites have been identified by ChIP-Seq in B cells induced to switch with LPS+IL-4 , and we compared the Nbs1-binding sites with these . About 29% of the AID-dependent DSBs occur at Spt5-binding sites , a highly significant correspondence ( Table 1 ) . However , 50% of AID-independent DSBs also occur at Spt5-binding sites . Off-target AID-dependent DSBs can lead to chromosomal deletions , duplications , or translocations . Thus , we compared the Nbs1-binding sites with 234 AID-dependent translocation hotspots as defined by DNA regions that translocate to introduced I-Sce1 sites near IgH Sμ or within the c-myc locus in cells activated with LPS+IL-4 and over-expressing AID [17] . A small proportion ( 8 ) of the AID-dependent DSBs we identified occur at these AID-dependent translocation hotspots , but this is highly significant ( Table 1 ) . In a different study [20] , 51 hotspots of AID-dependent translocation events with an I-Sce1 site introduced into c-myc were identified in anti-CD40+IL-4 activated B cells , but none of these sites are present among our AID-dependent Nbs1-binding sites . Possible explanations for why our AID-dependent DSB sites do not overlap at a higher frequency with translocation sites are: our Nbs1-ChIP library is not saturated; differences in activation methods ( +/- IL-4 ) , their use of over-expressed AID [17] , and the DSBs we identify might be involved in translocations with sites other than IgH or c-myc . Also , it is possible that Nbs1-ChIP preferentially detects off-target DSBs that are slowly repaired or recombined . It is likely that the AID-dependent translocation hotspots identified in these studies [17 , 20] are within regions sufficiently near the IgH or c-myc loci to be able to recombine with them at a high frequency [70] . This possibility is consistent with the very low Nbs1 signals detected at Sγ3 in cells undergoing active IgG3 CSR . We hypothesize that Sγ3 DSBs are induced only when Sγ3 is synapsed with Sμ , and that Sγ3 DSBs are then rapidly recombined with Sμ DSBs [46 , 47] . Despite the facts that AID-dependent c-myc-IgH translocations have been detected in human and mouse germinal center B cells , lymphomas , and plasmacytomas [71–73] , and also in cultured activated mouse B cells with mutated DNA damage response genes [74] , we did not detect AID-dependent Nbs1-binding sites in the c-myc locus . We were also unable to detect AID-dependent DSBs in the c-myc locus by LM-PCR [15] . This is consistent with the report that AID-dependent mutations per se are extremely rare ( 4x10-5 per bp ) in the c-myc locus in germinal center B cells , except in cells lacking Ung and Msh2 where they increased by 16 . 8-fold [9] . These apparently conflicting results indicate that AID-induced mutations in c-myc are usually corrected by DNA repair [9] , and only lead to detectable translocations when under selection pressure or in cells lacking DNA repair or damage response genes . AID-dependent DSBs and translocations with I-Sce1 sites occur preferentially in super-enhancers [20 , 21]; super-enhancers are longer than general enhancers , are transcribed , and consist of clusters of transcription factor binding sites that regulate genes involved in cell-type specific functions [75 , 76] . Thus , we asked if the AID-dependent Nbs1-binding sites are located in super-enhancers , and found that although a minority of the AID-dependent and AID-independent sites are within super-enhancers , the association is highly significant ( Table 1 ) . The RNA exosome , which degrades nascent RNA from the 3’ end when transcription is arrested , is important for allowing AID to access the transcribed DNA strand , in addition to the non-transcribed strand [77] . This would be important for forming DSBs . Recently , by the use of RNA-Seq , Pefanis et al [78] showed that transcripts initiated in the antisense direction from numerous promoters are degraded by the RNA exosome , by demonstrating that these antisense transcripts are increased in splenic B cells deficient in exosomes . They termed these exosome-dependent RNA loci xTSS , and found that they often correspond with regions identified by translocation capture to be AID-dependent translocation hotspots [17] . Interestingly , several of the AID-dependent DSBs detected in either of the two experiments occur at xTSS , and the association is highly significant ( Table 1 ) . The reproducible AID-independent Nbs1-binding sites are all in transcriptionally active regions ( Table 1 ) , and most within annotated genes ( S4 Table ) . As discussed above , they correspond to two-ended DSBs , according to the observed positions of the strand-specific tags . Several mechanisms can generate DSBs in transcribed regions . ( 1 ) 10% of the AID-independent Nbs1 sites occur at CA repeats long enough to form Z DNA ( ≥50 bp ) . Z DNA has been shown to cause DSBs and deletions in an AID-independent manner , independent of replication , and involving NHEJ [79–81] . ( 2 ) If R loops within the genome are not removed by RNA-DNA helicase , RNaseH1 , or exosome activity [78 , 82–84] they can lead to DSBs , perhaps due to activities of the transcription-coupled nucleotide excision repair enzymes XPF and XPG [85] . ( 3 ) Early replicating fragile sites ( ERFS ) ( differing from common fragile sites ) have recently been identified as sites where DSBs are induced early during S phase in cells undergoing replication stress in an AID-independent manner [86] . 14% of the reproducible AID-independent sites correspond to ERFS , whereas their frequency among the reproducible AID-dependent Nbs1-binding sites is not higher than random intervals ( 4% ) . ( 4 ) Topoisomerase I is known to nick transcribed regions , and recently its ability to nick DNA has been shown to be important for allowing transcription from enhancers [87] . Interestingly , SSBs introduced by Topoisomerase I can be converted to DSBs , and have been shown to bind the MRN complex . In summary , by the use of Nbs1 ChIP-Seq , we have identified hundreds of off-target AID-dependent DSBs in the genome of activated splenic B cells . More than two-thirds occur at transcriptionally active sites , as determined by RNA Pol II binding . The notable observations about these sites are ( 1 ) that ~10% of the DSBs in each experiment and 46% of the reproducible AID-dependent DSBs occur within tandem pentamer repeats ≥400 bp in length that contain WGCW motifs , the AID target hotspot . This motif creates AID hotspot targets on both strands , thus readily generating DSBs . ( 2 ) Also notable , CA repeats ( ≥100 bp in length ) are found within ~20% of the AID-dependent DSB sites , and in 30% of reproducible sites . CA repeats form unstable Z-DNA , which could generate transient ss targets for AID; and CA repeats also increase AID-independent genome instability , perhaps due to recognition by structure specific nuclease . ( 3 ) Interestingly , Msh2 appears to contribute to DSBs at off-target sites , just as it does in the IgH S region , where it increases the conversion of SSBs induced by AID-Ung-Ape to DSBs [5] . ( 4 ) A small fraction of the DSBs appear to be generated during S phase , as they are one-ended DSBs , consistent with the finding that deficiencies in homologous recombination can increase AID-dependent genomic damage . It is also possible that some of the off-target DSBs generated during G1 phase escape into S phase , as the G1-S phase checkpoint appears to be quite weak in B cells undergoing CSR in culture [46 , 88] . DSBs in S phase are dangerous as they can lead to genome instability .
Mouse strains were extensively ( ≥8 generations ) backcrossed to C75BL/6 . AID-deficient mice were obtained from T . Honjo ( Kyoto University , Kyoto , Japan ) [1] . Msh2-deficient mice [89 , 90] were obtained from T . Mak ( University Health Network , Toronto CA ) . Knock-out mice were always derived by breeding heterozygotes . This study was approved by , and performed in according with the guidelines provided by , the University of Massachusetts Medical School Animal Care and Use Committee . Mice were housed in a pathogen-free facility . Mouse splenic B cells were isolated and induced to switch for two days to IgG3 as previously described [46] . pMX-PIE-AID-FLAG-ER-IRES-GFP-puro [44] was received from Drs V . Barreto and M . Nussenzweig ( The Rockefeller University , NY ) . The control retrovirus pMX-PIE-ER-IRES-GFP was previously described [91] . Production of viruses and infection of B cells was previously described [91] . Genomic DNA preparation , LM-PCR , and quantitative ChIP were performed as described [46] . Antibodies for ChIP were: Nbs1 ( Abcam , ab32074 ) , RNA Pol II ( Millipore , 04–1572 ) , and ER ( Santa Cruz Biotechnology sc-8002X ) . Primers used for LM-PCR are listed in S5 Table . Three-fold more template DNA was used in each lane of the LM-PCR gel to examine off-target DSBs compared with that used for Sμ DSBs . A modified version of the Illumina protocol was followed to prepare ChIP DNA samples for the deep sequencing pipeline . Briefly , blunting of the fragments was performed using the END-IT DNA repair kit ( Epicentre ) followed by the addition of a dA overhang using exo-minus Klenow ( Epicentre ) . Paired-end adapters ( Illumina ) were ligated using the fast link kit ( Epicentre ) . The fragments were amplified twice using the Illumina PE primers and PfuUltra II Fusion HS DNA polymerase ( Stratagene ) , and each round of PCR was followed by gel purification and sizing of the fragments . Samples were cloned using the Topo cloning system ( Invitrogen ) and several clones were sequenced to assess sample quality prior to submission for sequencing on the Illumina GAII ( Exp 1 ) or HiSeq 2000 ( Exp 2 ) platforms at the UMASS Deep Sequencing Core facility , obtaining either 36 bp single-end ( Exp 1 ) or 50 bp paired-end reads ( Exp 2 and Pol2 ) . | Activation-induced cytidine deaminase ( AID ) is required for diversifying antibodies during immune responses , and it does this by introducing mutations and DNA breaks into antibody genes . How AID is targeted is not understood , and it induces chromosomal translocations , mutations , and double-strand breaks ( DSBs ) at sites other than antibody genes in activated B cells . To determine what makes an off-target DNA site a target for AID-induced DSBs , we identify and characterize hundreds of genome-wide DSBs induced by AID during B cell activation . Interestingly , many of the DSBs are within or adjacent to two types of tandemly repeated simple sequences , which have characteristics that might explain why they are targeted . We find that most of the DSBs are two-ended , consistent with their generation during G1 phase of the cell cycle , which is when AID induces DNA breaks in antibody genes . However , a minority is one-ended , consistent with replication encountering an AID-induced single-strand break , thereby creating a DSB . Both types of off-target DSBs , but especially those present during S phase of the cell cycle , lead to chromosomal translocations , deletions and gene amplifications that can promote B cell lymphomagenesis . | [
"Abstract",
"Introduction",
"Results",
"and",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Nbs1 ChIP-Seq Identifies Off-Target DNA Double-Strand Breaks Induced by AID in Activated Splenic B Cells |
Annual influenza epidemics and occasional pandemics pose a severe threat to human health . Host cell factors required for viral spread but not for cellular survival are attractive targets for novel approaches to antiviral intervention . The cleavage activation of the influenza virus hemagglutinin ( HA ) by host cell proteases is essential for viral infectivity . However , it is unknown which proteases activate influenza viruses in mammals . Several candidates have been identified in cell culture studies , leading to the concept that influenza viruses can employ multiple enzymes to ensure their cleavage activation in the host . Here , we show that deletion of a single HA-activating protease gene , Tmprss2 , in mice inhibits spread of mono-basic H1N1 influenza viruses , including the pandemic 2009 swine influenza virus . Lung pathology was strongly reduced and mutant mice were protected from weight loss , death and impairment of lung function . Also , after infection with mono-basic H3N2 influenza A virus body weight loss and survival was less severe in Tmprss2 mutant compared to wild type mice . As expected , Tmprss2-deficient mice were not protected from viral spread and pathology after infection with multi-basic H7N7 influenza A virus . In conclusion , these results identify TMPRSS2 as a host cell factor essential for viral spread and pathogenesis of mono-basic H1N1 and H3N2 influenza A viruses .
Annual influenza epidemics and unpredictable pandemics pose a severe threat to human health , exemplified by the estimated 30–50 million deaths caused by the 1918 pandemic . Current therapy targets viral proteins , neuraminidase and M2 , but is hampered by development of resistance [1] , due to the high mutation rate of the virus . Novel antiviral strategies are urgently required and invariable host cell factors essential for viral spread are attractive targets . The cleavage of the influenza virus hemagglutinin ( HA ) by host cell proteases is essential for viral infectivity [2] , [3] . The HA proteins of highly pathogenic avian influenza viruses harbor multiple basic amino acids at their cleavage site and are activated by furin [4] . In contrast , low pathogenic avian and human influenza viruses contain a mono-basic cleavage site in their HA proteins . Several studies showed that multiple secreted proteases can activate human influenza viruses for infection of cell lines ( see [5] , [6] for examples and [7] for a review ) . However , the analysis of cultured human respiratory epithelium demonstrated that influenza virus cleavage activation is a cell-associated process and no evidence for a role of secreted proteases was obtained [8] . Subsequently , the type II transmembrane serine protease ( TTSP ) family member TMPRSS2 , a membrane associated protease , was shown to activate HA proteins of diverse human influenza viruses in cell culture [9] , [10] , [11] , [12] . In addition , TMPRSS2 was found to be expressed in human respiratory epithelium positive for alpha 2 , 6-linked sialic acid [13] . However , the role of TMPRSS2 in influenza virus spread and pathogenesis in the infected host has not yet been studied . Therefore , we investigated if TMPRSS2 contributes to influenza virus replication and pathogenesis in experimentally infected mice . We focused our analysis on viruses of the H1N1 ( including the pandemic 2009 influenza virus ) and H3N2 subtypes , since viruses of these subtypes are presently circulating in the human population . Our study shows that deletion of Tmprss2 in knock-out mice strongly limits virus spread and lung pathology after H1N1 influenza A virus infection . The deletion of Tmprss2 also reduces body weight loss and mortality after H3N2 infection but to a much lower degree than for H1N1 infected mice .
To assess the role of TMPRSS2 during influenza virus infection in vivo , we used mice carrying a deletion of the Tmprss2 gene [14] . Non-infected Tmprss2 knock-out mice did not show a phenotype in the absence of infection , as described previously [14] and RT-PCR analysis of kidney tissue confirmed the absence of full length Tmprss2 transcripts . Upon intranasal infection of mice with mouse-adapted PR8M ( A/PuertoRico/8/34 H1N1 Münster variant , [15] ) , wild type mice lost weight significantly after infection and 50% of infected mice died , whereas Tmprss2 knock-out mice did not exhibit body weight loss and showed no signs of disease ( Figs . 1A; Fig . S1 ) . The same results were obtained after infection with a human isolate of the pandemic HA4 ( A/Hamburg/4/2009 H1N1 ) virus ( Figure 1B ) . Also , after infection with a lethal dose of the more virulent PR8F virus isolate ( A/PuertoRico/8/34 H1N1 Freiburg variant ) all wild-type mice died within seven days post infection whereas no knock-out mice showed symptoms of disease ( Figure 1C ) . Similar results were observed for blood oxygen saturation levels which provide a measurement for lung function . Whereas wild type mice exhibited a significant drop in peripheral blood oxygen saturation that peaked at day 8 post infection ( p . i . ) with PR8M virus , Tmprss2−/− mutant mice showed only a very mild change ( Figure 2 ) . Histological analyses of infected lungs revealed a similar onset of the influenza infection on day 1 post infection in Tmprss2−/− and wild type mice with infected epithelial cells in the bronchiole ( Figure 3A , B , E , F ) . However , at day 3 p . i . virus was spreading into the alveolar regions of wild type mice whereas it was only found in bronchioles in Tmprss2−/− mice ( Figure 3C , D and G , H , respectively ) . Furthermore , infected wild type mice showed a strong increase in lung infiltrates and also in the number of infected cells whereas in Tmprss2 knock-out mice a much lower number of infiltrating cells and infected cells were observed ( Figure 3 K , L and O , P , respectively ) . Thus , the absence of TMPRSS2 largely protects animals from virus spread and virus induced pathogenesis . Next , we assessed if protection from pathogenesis was due to reduced viral spread . After infection with PR8M , we could detect infectious viral particles in the lung in both homozygous Tmprss2−/− and wild type mice . However , the number of infectious particles was close to background at day 1 p . i . in Tmprss2−/− mice and was markedly reduced at days 2 and 3 p . i . compared to wild type mice ( Figure 4 ) . Influenza-specific antibodies were readily detectable in sera of Tmprss2 knock-out mice ( Figure S2 ) after infection with PR8M and PR8F virus demonstrating that the inoculated virus was able to infect lung cells and elicit a humoral immune response . In conclusion , these results show that TMPRSS2 is critical for efficient spread and pathogenesis of epidemic and pandemic H1N1 influenza viruses in vivo . Next , we sought to obtain direct evidence for the lack of proteolytic cleavage of HA in Tmprss2 deficient mice . For this , broncho-alveolar lavages ( BAL ) of infected mice were collected and the proteolytic processing of the HA precursor protein HA0 was analyzed . After infection with a high dose of PR8M virus , processed HA1 as well as non-processed HA0 protein were detected in infected wild type mice whereas only HA0 protein was found in homozygous Tmprss2 mutant mice ( Figure 5 ) . These results demonstrate that TMPRSS2 is essential for efficient HA cleavage activation in mice . At present , influenza viruses of the HA subtypes H1 and H3 are circulating in humans . Therefore , we investigated if a H3 virus was also dependent on expression of a functional Tmprss2 gene . After infection with a low dose ( 101 Focus Forming Units ( FFU ) ) of a mouse-adapted H3N2 virus ( A/HK/01/68 [16] ) , body weight loss is less severe and survival is increased in Tmprss2 knock-out compared to wild type mice ( Figure 6A , B ) . After infection with a higher dose ( 2×103 FFU ) of H3N2 virus , mortality was significantly lower in Tmprss2−/− mice compared to wild type mice ( Figure 7A ) . However , no significant differences were observed in the amount of infectious particles at day 1 to 3 p . i . ( Figure 7B ) . Thus , activity of TMPRSS2 is required for the processing of both H1N1 and H3N2 but H3N2 viruses may be cleaved by other proteases in addition to TMPRSS2 . In cell culture , TMPRSS2 is dispensable for cleavage activation of viruses with a multi-basic cleavage site [10] . Thus , if the resistance of Tmprss2 knock-out mice to H1N1 infection was indeed due to lack of HA processing , a virus with a multi-basic cleavage site should spread efficiently and cause disease . To investigate this , we infected mice with SC35M ( mouse-adapted A/Seal/Massachusetts/1/80 , H7N7 ) influenza virus which contains a multi-basic HA cleavage site . Mortality and body weight loss in infected Tmprss2−/− mice were not significantly different compared to wild type and Tmprss2+/− infected mice ( Figure 8 ) , suggesting that the presence of a multi-basic cleavage site renders viral spread and pathogenesis independent of TMPRSS2 expression . Finally , we investigated if murine Tmprss2 was able to activate HA proteins of H1N1 and H3N2 viruses . The co-expression of protease and the respective HA proteins of PR8M , HA4 , and H3N2 viruses in transfected cells facilitated HA cleavage of all viruses ( Figure S3A ) . Additionally , expression of TMPRSS2 in this cell culture system allowed the spread of PR8M and HA4 viruses in a trypsin-independent fashion ( Figure S3B ) . In contrast , spread of mouse-adapted SC35M virus , containing a multi-basic cleavage site did not depend on TMPRSS2 expression [10] . Thus , murine Tmprss2 , like its human homologue [11] , [17] can activate HA .
Cleavage activation of influenza virus HA by host cell proteases is essential for viral infectivity [2] , [3] . However , the nature of the proteases required for the cleavage activation of viruses with a mono-basic HA cleavage site in the infected host organism remains unclear . At least eight candidate enzymes from different protease families have been suggested based on cell culture studies [18] , leading to the concept that redundant proteolytic enzymes activate influenza viruses in the host . Here , we show for the first time that the deletion of a single protease , TMPRSS2 , in mice largely abrogates viral spread and protects animals from severe pathology and death after H1N1 and , to a lower extent , H3N2 influenza virus infection . After infection of Tmprss2−/− mice with H1N1 virus , no processing of the HA precursor protein HA0 was observed in BAL . However , an initial increase in viral titers was measured from day 1 to day 2 p . i . whereas at later times p . i . viral titers rapidly decreased . This initial increase in viral titers is expected because the virus used for infections had been produced in embryonated chicken eggs . It therefore carries an activated HA allowing it to enter cells and replicate [19] , [20] . In addition , it is conceivable that other proteases besides TMPRSS2 may facilitate low levels of H1 cleavage and allow limited viral spread which is rapidly cleared once the antiviral immune responses have been activated . However , the cleavage activation by such alternative enzymes must be very inefficient since viral titers in H1N1 infected Tmprss2−/− mice were markedly reduced compared to wild type animals and no weight loss was observed . Surprisingly , H3N2 virus which also carries a mono-basic cleavage site in the HA , was able to replicate in Tmprss2−/− mice . However , body weight loss and mortality were significantly reduced in knock-out mice compared to wild type mice in a dose-dependent manner . On the other hand , viral load was not significantly lower in mutant mice after infection with 2×103 FFU . The amino acid sequence in the HA loop which is recognized by proteases differs between HA subtypes H1 and H3 ( Figure 9 ) . A recent study demonstrated that different proteases , including TMPRSS2 , TMPRSS11d ( HAT ) and even trypsin , cleave HA from different subtypes and variants with varying efficiency [21] . In addition , ST14 ( matriptase ) has the capability to cleave HA of particular H1 subtype strains but only minimal cleavage was observed for H2 and H3 [22] . Furthermore , KLK5 and 12 ( kallikrein related-peptidase 5 and 12 ) have been described as host proteases that are capable of cleavage activation of viral H1 , H2 and H3 HA in vitro [23] . Thus , H3N2 appears to be processed to some extent by TMPRSS2 , resulting in the reduced pathology in Tmprss2 knock-out mice but also other proteases are able to cleave H3 hemagglutinin in vivo . Finally , it is noteworthy that Tmprss2−/− mutant mice were not protected from infection with an H7N7 virus which contains a multi-basic HA cleavage site and can be activated by ubiquitously expressed proteases [4] . Our findings may have potential for the development of future influenza virus therapeutics . Broad spectrum protease inhibitors have been shown to inhibit influenza virus in cell culture and in vivo [24] , [25] , [26] , [27] , [28] but unwanted side effects are a major concern . The results reported here suggest that targeting a single protease , TMPRSS2 , may be sufficient to achieve a notable therapeutic benefit against H1N1 influenza viruses and possibly other subtypes . Blocking of TMPRSS2 may not be associated with severe unwanted side effects , since Tmprss2−/− mice are healthy and do not show any phenotypic alterations in the absence of an infection [14] . Furthermore , TMPRSS2 inhibitors might exert activity against diverse respiratory infections . For example , human metapneumovirus [29] , the emerging MERS-coronavirus [30] , [31] and SARS-coronavirus [32] , [33] , [34] can also be activated by TMPRSS2 in cell culture and might use this protease to support their spread in the infected host . It should , however , be noted that our findings in the mouse model system require validation in humans .
All experiments in mice were approved by an external committee according to the national guidelines of the animal welfare law in Germany ( ‘Tierschutzgesetz in der Fassung der Bekanntmachung vom 18 . Mai 2006 ( BGBl . I S . 1206 , 1313 ) , das zuletzt durch Artikel 20 des Gesetzes vom 9 . Dezember 2010 ( BGBl . I S . 1934 ) geändert worden ist . ’ ) . The protocol used in these experiments has been reviewed by an ethics committee and approved by the ‘Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit , Oldenburg , Germany’ ( Permit Number: 33 . 9 . 42502-04-051/09 ) . Original stocks of viruses were obtained from Stefan Ludwig , University of Münster ( PR8M , A/PuertoRico/8/34 H1N1 , Münster variant ) , from Peter Stäheli , University of Freiburg ( PR8F , A/PuertoRico/8/34 H1N1 , Freiburg variant and SC35M [35] ) , from Otto Haller , University of Freiburg ( H3N2 , [16] ) and from Thorsten Wolff , Robert-Koch-Institute , Berlin ( HA4 ) . The different virulence of PR8M and PR8F virus isolates has been described before [15] . SC35M ( H7N7 ) was originally derived from the seal , adapted to the mouse by serial passages in the mouse lung [35] , and the laboratory strain which was used here was generated by reverse genetics using the plasmid rescue system [36] . Virus stocks of PR8 were prepared by infection of 10-day-old embryonated chicken eggs and for HA4 on MDCK cells as described [37] . Mutant Tmprss2−/− mice were on a mixed C57BL/6J – 129 background [14] . Animals were maintained under specific pathogen free conditions at the animal facility of the HZI in Braunschweig . Heterozygous mutant mice were interbred and wild type , heterozygous and homozygous mutant mice were genotyped by PCR analysis and then used for infections . Genotyping of Tmprss2 alleles was carried out using a three primer strategy ( P1 5′TGTGCCCTTGGACAGATGACTC3′ , P2 5′GGACTACAGATATGAGGTGTTC3′ , P3 5′AGGCCAGAGGCCACTTGTGTAG3′ ) that allows to distinguish between wild type ( yielding a 540 bp product ) and knock-out ( yielding a 400 bp product ) alleles , respectively . Expression plasmids encoding human and mouse proteases TMPRSS2 were published earlier [10] , [13] . Previously described expression plasmids for PR8M [38] were used as templates for amplification of the respective coding regions using oligonucleotides PR8-HA-5Acc: GGGGGTACCACCATGAAGGCAAACCTACTGGTCCTG , PR8-HA-3Nhe: GGGCGCTAGCTCAGATGCATATTCTGCACTG . The resulting PCR products were inserted into plasmid pCAGGS using the Acc65I and NheI sites . For cloning of HA4 HA protein , expression plasmid kindly provided by Prof . Klenk [39] was used as templates for amplification of the coding region using oligonucleotides swi09-HA-5Eco: GGGAATTCACCATGAAGGCAATACTAGTAGTTCTGC and swi09-HA-3Xho: GGGCTCGAGTTAAATACATATTCTACACTGTAGAG . The resulting PCR products were inserted into plasmid pCAGGS using EcoRI and XhoI . For infection experiments , female mice at the age of 8–11 weeks were anesthetized by intra-peritoneal injection of Ketamin-Xylazine solution in sterile NaCl ( 50 mg/ml Ketamine , Invesa Arzneimittel GmbH , Freiburg; 2% Xylazine , Bayer Health-Care , Leverkusen ) with a dose adjusted to the individual body weight . Infection was performed by intranasal application of virus solution in 20 µl of sterile phosphate-buffered saline . Subsequently survival and body weight loss were monitored until day 14 p . i . In addition to mice that were found dead , mice with a weight loss of more than 30% of the starting body weight were euthanized and recorded as dead . Viral load in infected lungs was determined on MDCK II ( Madin-Darby Canine Kidney II ) cells using the FFU assay as described [15] . Detection limit of the assay is at 80 infectious particles/lung . Thus , for samples where no virus was detected , the data points were set to 80 FFU/lung . For analysis of HA in viral particles , Broncho-alveolar lavages ( BALs ) of wild type and Tmprss2 knock-out mice infected with 2×105 FFU PR8M were harvested at day 1 p . i . , centrifuged at full speed and supernatant were loaded on a 20% sucrose cushion for concentration of viral particles for 2–3 h at full speed and 4°C . The pelleted viral particles were lysed in 2×SDS loading buffer . For immuno-blotting , the lysates were separated by SDS gel electrophoresis , blotted on a nitrocellulose membrane and HA was detected by staining with a rabbit anti-PR8HA antibody ( Sino biological ) at a dilution of 1∶500 , followed by incubation with a horseradish peroxidase ( HRP ) -coupled anti-rabbit antibody ( Dianova ) at a dilution of 1∶10 . 000 . As loading control , membranes were stripped ( Stripping Buffer: 1M Tris-HCl ( pH 6 , 8 ) , 10% SDS , 100 mM ß-Mercaptoethanol ) for 30 min at 50°C , washed 1 h with dH2O and stained for NA by using an anti-influenza A virus goat serum ( Millipore ) at a dilution of 1∶500 followed by incubation with a horseradish peroxidase ( HRP ) -coupled anti-goat antibody ( Dianova ) at a dilution of 1∶10 . 000 . Bands were visualized by using a commercially available kit ECL Prime Western Blotting Detection Reagents ( Amersham ) . Eight to twelve weeks old mice were infected intranasally with 2×105 FFU PR8M and the amount of oxygen saturation was determined by the MouseOx® system ( STARR Life Science Corp . ) over a period of 14 days . For oxygen measurements mice were anesthetized using an isoflurane inhalator . Lungs were prepared and immersion-fixed for 24 hours in 4% buffered formaldehyde solution ( pH 7 . 4 ) , dehydrated in a series of graded ethanol and embedded in paraffin . Sections ( 0 . 5 µm ) were cut from five evenly distributed levels of the paraffin blocks and stained with haematoxylin and eosin . For immunohistochemical studies , sections were stained with a polyclonal primary antibody ( against influenza A H1N1 virions; Virostat ) overnight at 4°C and subsequently tissue sections were incubated for 30 min with secondary antibody ( rabbit anti-goat-biotin; KPL; Gaithersburg , Madison , USA ) and counterstained with haematoxylin . 293T cells were transiently transfected with expression plasmids encoding human or mouse TMPRSS2 or control transfected with empty plasmid . At 24 h post transfection , cells were infected with either of the influenza viruses PR8M , HA4 or SC35M at a multiplicity of infection ( MOI ) of 1 . After 1 h incubation at 37°C , virus was removed and fresh MEM medium ( supplemented with 0 . 2% BSA and 1 µg/ml TPCK-trypsin or PBS ) was added to the cells . At 48 h p . i . , supernatants were harvested , cleared from debris by centrifugation for 5 min at 3 . 500 rpm and stored at −80°C until quantification of infectious virus particles by focus formation assay ( FFU ) as described previously [15] . | Seasonal influenza epidemics and pandemics represent a serious health threat to the human population . Resistance to presently available anti-viral drugs is frequently observed . Therefore , identification of new targets for anti-viral therapy is an urgent need . Host proteases are required for processing of the virus hemagglutinin and may thus represent a suitable target for intervention . Here , we report that the deletion of a single host protease gene , Tmprss2 , in mice protects the host against viral spread in infected lungs . Only very mild pathogenesis was observed in Tmprss2 mutant mice after infection with H1N1 virus and less severe pathogenesis was observed after infection with H3N2 virus . Thus , our results suggest that the host protease TMPRSS2 may be a prime target for antiviral intervention . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Tmprss2 Is Essential for Influenza H1N1 Virus Pathogenesis in Mice |
Viral diseases of the respiratory tract , which include influenza pandemic , children acute bronchiolitis , and viral pneumonia of the elderly , represent major health problems . Plasmacytoid dendritic cells play an important role in anti-viral immunity , and these cells were recently shown to express ChemR23 , the receptor for the chemoattractant protein chemerin , which is expressed by epithelial cells in the lung . Our aim was to determine the role played by the chemerin/ChemR23 system in the physiopathology of viral pneumonia , using the pneumonia virus of mice ( PVM ) as a model . Wild-type and ChemR23 knock-out mice were infected by PVM and followed for functional and inflammatory parameters . ChemR23−/− mice displayed higher mortality/morbidity , alteration of lung function , delayed viral clearance and increased neutrophilic infiltration . We demonstrated in these mice a lower recruitment of plasmacytoid dendritic cells and a reduction in type I interferon production . The role of plasmacytoid dendritic cells was further addressed by performing depletion and adoptive transfer experiments as well as by the generation of chimeric mice , demonstrating two opposite effects of the chemerin/ChemR23 system . First , the ChemR23-dependent recruitment of plasmacytoid dendritic cells contributes to adaptive immune responses and viral clearance , but also enhances the inflammatory response . Second , increased morbidity/mortality in ChemR23−/− mice is not due to defective plasmacytoid dendritic cells recruitment , but rather to the loss of an anti-inflammatory pathway involving ChemR23 expressed by non-leukocytic cells . The chemerin/ChemR23 system plays important roles in the physiopathology of viral pneumonia , and might therefore be considered as a therapeutic target for anti-viral and anti-inflammatory therapies .
A number of single-stranded RNA viruses , including influenza and respiratory syncytial virus ( RSV ) cause infections of the lower respiratory tract and pneumonias , representing thereby a major health problem worldwide . These viruses constitute a major cause of hospitalization for acute bronchiolitis in infants and young children ( especially RSV ) , exacerbation of chronic obstructive pulmonary disease ( COPD ) , and viral pneumonia in the elderly [1]–[4] . Animal models indicate that the severity of infections caused by RSV or its mouse counterpart , the pneumonia virus of mice ( PVM ) , is due essentially to an excessive primary immune response of the host , rather than the direct cytopathogenicity of the viruses [5]–[8] . In line with these observations , administration of anti-viral drugs was shown to have minimal impact on the clinical outcome . Following an early and non-specific inflammatory response characterized by the recruitment of neutrophils and cytotoxic natural killer cells [9]–[10] , a delayed adaptive immune response leads , 7 to 10 days after infection , to an important influx of CD8+ cytotoxic T cells in the lung , and the production of RSV- or PVM-specific antibodies by plasma cells [5] , [8] , [10] . Dendritic cells ( DCs ) play a central role in the initiation of this adaptive immune response . DCs capture viral antigens in the lung , undergo a maturation process , and migrate to lymph nodes where they activate naive CD4+ and CD8+ T cells [11]–[12] . In the lung , two major subsets of dendritic cells have been described: CD11c+ CD11b+ MHC IIhigh F4-80− conventional , or myeloid , DCs ( cDC or mDC ) and CD11clow CD11b− Gr1+ mPDCA-1+ plasmacytoid DCs ( pDC ) [13]–[16] . In contrast with myeloid DCs , plasmacytoid DCs display a relatively weak antigen-presenting capability but exert essential immuno-modulatory functions in asthma and viral diseases [17] . PDCs are major producers of type I interferons ( IFNs ) , including IFN-α and IFN-β , in response to the sensing of viral components through Toll-like receptors ( TLRs ) . Type I IFNs play a crucial role in innate anti-viral immunity , but also in the initiation of adaptive immune responses [18] . Nevertheless , the role of pDCs in the physiopathology of viral pneumonia is still unclear , as discrepancies exist among published data . Indeed , some studies performed with influenza virus or human RSV suggest a limited role for pDCs [16] , [19] , [20] whereas depletion and adoptive transfer experiments have demonstrated the ability of pDCs to inhibit viral replication and enhance viral clearance in RSV-infected mice . They also decrease airway hyperreactivity , lung inflammation and mucus production , properties attributed to inherent anti-inflammatory properties of pDCs [21] , [22] . Such anti-inflammatory activity of pDCs is also described in asthma , in which pDCs mediate tolerance through induction of regulatory T lymphocytes and suppression of effector T cell generation [14] , [23]–[25] . ChemR23 is a G protein-coupled receptor , expressed by immature DCs , macrophages and NK cells [26]–[28] . Its natural ligand , chemerin , was purified from human inflammatory fluids [29] . Chemerin , acting through ChemR23 , has chemoattractant properties at low and subnanomolar concentrations , particularly for immature plasmacytoid DCs ( pDCs ) that express ChemR23 at high levels [28] , [30] , [31] . Chemerin is synthesized as an inactive precursor , prochemerin , which is present at high concentration in plasma . Prochemerin can be rapidly converted into a full ChemR23 agonist by neutrophil-derived serine proteases ( elastase and cathepsin G ) and serine proteases of the coagulation and fibrinolytic cascades , released as a result of tissue injury , inflammation or infection [32] , [33] . Provided the expression of prochemerin in lung [28] , [29] , its processing by neutrophil proteases , the preferential expression of ChemR23 by pDCs , and the role of pDCs and neutrophils in the physiopathology of viral infections of the lung , we investigated the potential involvement of the chemerin-ChemR23 system in pDC recruitment and its functional consequences during viral pneumonia . We used wild-type ( WT ) and ChemR23 knock-out ( KO ) mice in the PVM-induced acute pneumonia model and demonstrated a crucial role of ChemR23 in the recruitment of pDCs to the lung , the control of viral replication and clearance , as well as in the dampening of excessive inflammatory responses . The enhanced inflammatory status and higher mortality rate of infected ChemR23-deficient mice is however not due to the impairment of pDC recruitment , but rather to the loss of an anti-inflammatory role of chemerin through non-leukocytic cells .
After intranasal inoculation of PVM ( 1000 PFU ) or control medium ( PBS ) , ChemR23−/− and wild-type C57BL/6 mice were monitored daily for survival and weight loss . A significantly lower survival rate was observed in ChemR23−/− mice as compared to WT mice ( 21% vs 62% respectively; p<0 . 01 ) ( Figure 1A ) . ChemR23−/− mice began to lose weight one day ahead of WT mice ( i . e . at days 7 and 8 respectively , Figure 1A ) . Thereafter , the weight curves were grossly parallel up to day 10 , when WT mice stabilized their weight before recovering , whereas ChemR23−/− mice continued to lose weight till day 12 included . Furthermore , at days 8 and 9 post-infection , infected ChemR23−/− mice presented a more severe pattern of illness signs , characterized by motor slowing , hunching , fur ruffling , and crackles . It should also be noted that the difference in weight loss remained significant at late time points ( days 12 to 16 ) , when the more affected animals had died , particularly in the infected ChemR23−/− group . The differences in mortality rate and weight loss between infected ChemR23−/− and WT mice correlated with the pathological findings made on lungs collected at day 9 post-infection ( Figure 1B and 1C ) . Macroscopic examination of infected lungs revealed numerous erythematous inflammatory foci , which were larger and more numerous in Chem23−/− mice than in controls ( Figure 1B ) . As expected , such foci were never seen in uninfected mice . At selected time points after infection , the lung weight was recorded , as a reflect of inflammation and edema . ChemR23−/− mice presented a higher lung weight increase ( 44% over that of WT mice on average between day 6 and day 11 ) ( Figure 1B ) . Microscopic examination of the lungs clearly showed stronger interstitial , peribronchiolar , and perivascular infiltrates rich in neutrophils , lymphocytes and macrophages in infected ChemR23−/− mice ( Figure 1C ) . Next , we investigated the respiratory function of the animals , using a whole body double chamber plethysmograph ( Figure S1 ) . No significant changes were observed up to day 6 post-infection . At days 7 and 8 post-infection , while WT mice exhibited limited changes of the various parameters , ChemR23−/− mice displayed a severe restrictive syndrome likely resulting from the strong lung edema and leukocyte infiltrate . This was characterized by a sharp decrease ( up to 50% ) in tidal volume ( p<0 . 01 ) , a 2 . 3-fold increase in end expiratory pause ( p<0 . 05 ) , a 3 . 4-fold increase in enhanced pause ( Penh ) ( p<0 . 001 ) , and a 65% reduction in expiratory balance ( p<0 . 05 ) ( Figure 1D and Figure S1 ) . No difference in airway resistance ( sRaw ) was observed between ChemR23−/− and WT mice ( Figure 1D ) . Plethysmographic analysis could not be performed after day 8 post-infection , as a result of bioethical limitations caused by the viral disease . Chemerin , the agonist of ChemR23 , was assayed by ELISA in the broncho-alveolar lavage at several time points after infection . Chemerin levels increased from day 8 to day 10 post-infection in both ChemR23−/− and WT mice ( Figure 2A ) . However , they were higher ( 4 . 5-fold on average ) in ChemR23−/− mice at each time point with the highest levels at day 10 ( 10 . 0±0 . 2 versus 2 . 1±0 . 4 ng/ml for WT , p<0 . 001 ) . In order to investigate whether elevated chemerin levels were the consequence of increased local expression , chemerin transcripts were assessed by qRT-PCR ( Figure 2B ) . The data were normalized using two housekeeping genes ( YWHAZ and CANX ) as references , and reported to the chemerin transcript level in uninfected control mice . No significant differences were observed according to infection status or genotype . As the ELISA assays both active and inactive forms of chemerin ( including prochemerin ) , we tested chemerin bioactivity in BAL fluids . Fractions resulting from a reverse phase HPLC were tested in an aequorin-based intracellular Ca2+ mobilization assay . A biological activity was recorded for fractions obtained from infected ChemR23−/− mice using CHO-K1 cells expressing mouse ChemR23 ( Figure 2C ) . Bioactivity was also retrieved from infected WT mice , following sample concentration , while no activity was observed in fractions from naive mice ( data not shown ) . To assess the role of ChemR23 in viral replication and clearance , immunofluorescence staining of lung sections with an anti-PVM antibody was performed at days 6 , 8 , and 9 post-infection . This allowed the observation of PVM-infected cells , which include bronchiolar epithelial cells , type I and type II pneumocytes , alveolar macrophages and eosinophils [34]–[36 and unpublished observations] ( Figure 3A ) . No noticeable difference was observed at day 6 between WT and ChemR23−/− mice . Thereafter , PVM staining remained stable in WT mice at day 8 , and decreased by day 9 , whereas it increased in ChemR23−/− mice up to day 9 . As a result , ChemR23−/− sections showed a stronger and more diffuse staining than WT mice at day 8 , and the difference increased further by day 9 . The enhanced viral replication and reduced viral clearance in ChemR23−/− mice were confirmed by the determination of the lung viral titers by plaque assay ( Figure 3B ) . The peak of viral titer was reached later in ChemR23−/− than in WT mice ( respectively at day 8 and 10 post-infection ) , and the titers were up to 100-fold higher in ChemR23−/− than in WT mice ( 2 . 7±0 . 9×106 versus 0 . 31±0 . 16×106 PFU/lung at day 8 , p<0 . 05; 5 . 2±4 . 5×106 versus 9 . 1±2 . 9×104 PFU/lung at day 9 , p<0 . 01; 2 . 2±2 . 1×104 versus 170±100 PFU/lung at day 10 , p<0 . 05 respectively ) . As for immunostaining , no difference in viral titers was apparent at day 6 ( 2 . 9±0 . 8×104 for ChemR23−/− mice and 2 . 9±0 . 6×104 PFU/lung for WT mice , p>0 . 05 ) . The delayed viral clearance in ChemR23−/− mice was associated with a mild decrease in anti-PVM antibodies in sera of ChemR23−/− mice at day 10 post-infection , whereas this difference was abrogated at day 11 ( Figure 3C ) . We next investigated if the differences in clinical outcome and viral clearance could be explained in part by differences in the production of cytokines involved in anti-viral defenses . Various cytokines were tested by qRT-PCR and/or ELISA in basal conditions and at days 6 , 8 and 10 post-infection . Most cytokines peaked at day 8 post-infection and decreased by day 10 ( Figure 4 and Figure S2 ) . Type I interferons , including IFN-α and -β , are known to play a central role in anti-viral immunity [18] . We first assayed IFN-α and -β transcripts in lung by qRT-PCR ( Figure 4A ) . In contrast to WT mice , IFN-α and –β transcript levels increased only slightly in ChemR23−/− mice and remained respectively 5-fold and 4-fold lower than in WT mice at day 8 post-infection ( p<0 . 05 ) . We confirmed by ELISA the lower levels of IFN-α in ChemR23−/− compared to WT mice in lung homogenates ( 49±16 versus 158±51 pg/ml respectively; p<0 . 05 ) , and BAL fluids ( 91±26 versus 239±46 pg/ml respectively; p<0 . 01 ) at day 8 post-infection . IFN-γ production was also assessed by qRT-PCR in lung , and using ELISA in lung , serum and BAL ( Figure 4B and Figure S2 ) . The levels peaked at day 8 post-infection , with no significant difference between ChemR23 KO and WT mice . Interleukin ( IL ) -12 ( or IL-12p70 ) , another important cytokine regulating the effector functions of lymphocytes , is composed of two subunits , p40 and p35 . The IL-12 p40 subunit was investigated in lung , and significantly higher levels were found in WT mice than in ChemR23−/− mice at days 8 and 10 post-infection , both for the protein ( 2 . 50±0 . 16 versus 1 . 40±0 . 30 ng/ml respectively at day 8 , 1 . 30±0 . 20 and 0 . 50±0 . 03 ng/ml at day 10; p<0 . 05 ) ( Figure 4B ) , and the transcript ( ∼3-fold difference; p<0 . 05 ) ( Figure S2 ) . As IL-12 p40 subunit is also shared by IL-23 , a heterodimeric cytokine promoting the synthesis of IL-17 , IL-17 synthesis was assessed in lung homogenates by ELISA and qRT-PCR , but no differences were observed between WT and ChemR23−/− mice ( Figure S2 ) . IL-12 p70 was assayed by ELISA and a cytometric bead array-based immunoassay , but remained undetectable in BAL fluids and lung homogenates ( data not shown ) . ChemR23−/− mice displayed higher levels of IL-6 than WT mice at day 8 and the difference became significant at day 9 ( 1447±664 versus 288±38 pg/ml respectively , p<0 . 01 ) and day 10 post-infection ( 1301±328 versus 454±124 pg/ml respectively , p<0 . 05 ) . Although all these cytokines were increased during the acute phase of the disease , no differences were observed between KO and WT mice for lung TNF-α , IL-1β , IL-5 , IL-10 , IL-13 , and IL-17 ( Figure 4B and Figure S2 ) . Moreover , as neutrophils are massively recruited during PVM infection , KC ( CXCL1 ) , a major chemokine for neutrophils in mouse , was measured in lung homogenates . KC levels increased to a peak value by day 8 post-infection and remained significantly higher in ChemR23−/− mice compared to WT mice at day 9 ( 86 . 4±20 . 5 versus 28 . 3±2 . 2 pg/ml respectively , p<0 . 01 ) and day 10 post-infection ( 50 . 3±5 . 4 versus 26 . 5±3 . 6 pg/ml respectively , p<0 . 01 ) ( Figure 4B ) . To investigate the different leukocyte populations recruited during the viral infection , cell suspensions were obtained from digested lungs at different time points after PVM inoculation and assessed by flow cytometry . As ChemR23 is highly expressed by pDCs and is one of the few functional chemoattractant receptors on these cells , the percentage and the absolute number of pDCs were determined ( Figure 5A ) . PDCs were identified as Gr-1+ mPDCA+ cells after a first gating on the CD11b− CD11c+ population . The number of pDCs in uninfected ChemR23−/− mice did not differ from that in WT mice . A peak of pDCs was reached at day 8 post-infection with , in WT mice , a 5 . 8-fold increase from 11 . 7±0 . 8×103 pDCs in basal conditions to 68 . 5±4 . 7×103 pDCs per lung . In ChemR23−/− mice however , the increase in pDCs was significantly lower than in WT mice ( Figure 5A ) . At day 8 post-infection , pDCs represented , respectively in ChemR23−/− and WT mice , 0 . 38±0 . 03% and 0 . 85±0 . 06% of the total cell number ( p<0 . 001 ) . The corresponding absolute numbers were respectively 42 . 0±4 . 4×103 and 68 . 5±4 . 7×103 pDCs per lung ( p<0 . 01 ) . Myeloid dendritic cells ( mDCs ) were identified as CD11c+ CD11b+ cells after a first gating on the MHC-IIhigh F4-80− population ( Figure 5B ) . MDCs also increased during viral challenge but , in contrast to pDCs , higher values were found in ChemR23−/− than in WT mice . Indeed , at day 8 post-infection , 143±8×103 mDCs were counted in ChemR23−/− mice and 88±7×103 cells in WT mice ( p<0 . 001 ) . This difference increased at day 10 with 175±13×103 mDCs in ChemR23−/− and 97±9×103 mDCs in WT mice . Lung macrophages were defined as F4-80+ CD11b+ CD11c− cells and neutrophils as Gr1+ CD11b+ CD11c− cells . The number of macrophages and neutrophils increased from day 6 to day 10 post-infection , reaching respectively 6 . 5- and 3 . 8-fold basal values ( Figure 5B ) . In ChemR23−/− mice , macrophages and neutrophils were significantly higher than in WT mice at days 8 and 10 post-infection . At day 10 , macrophage counts were respectively 3 . 6±0 . 4×106 and 2 . 4±0 . 1×106 cells in ChemR23−/− and WT mice ( p<0 . 05 ) , and neutrophil counts were respectively 2 . 5±0 . 3×106 and 1 . 5±0 . 1×106 cells ( p<0 . 01 ) . No differences were observed between ChemR23−/− and WT mice for NK cells ( NK1-1+ CD3− ) , T cells ( CD19− CD3+ ) and B cells ( CD19+ CD3− ) ( Figure 5B ) . Flow cytometry analysis was also performed on BAL fluids obtained at day 8 post-infection . Compared to uninfected mice , a significant increase in pDCs was observed in the WT group whereas this increase was not significant in ChemR23-deficient mice . In ChemR23−/− mice , pDC counts reached only 50% of the values observed in WT mice ( 5 . 6±0 . 7×103 versus 11 . 4±2 . 2×103 , p<0 . 05 ) ( Figure 6A ) . As pDCs play an important role in anti-viral immunity through their ability to synthesize type I IFN that promote the cytotoxic response , the number of CD8+ cytotoxic T cells was determined in BAL fluids . Lower CD8+ T cell counts were found in ChemR23 KO mice ( 35 . 2±2 . 6×103 versus 12 . 1±1 . 1×103 for infected WT and KO mice respectively; p<0 . 001 ) . No difference was observed for CD4+ T lymphocytes ( 7 . 7±0 . 8×103 versus 7 . 5±1 . 1×103 for WT and KO mice respectively; p>0 . 05 ) ( Figure 6A ) . Whereas neutrophil counts in BAL fluid did not differ between WT and KO mice at day 8 post-infection , a major difference was seen at day 10 . Indeed , neutrophils represented the main population in infected ChemR23−/− mice but not in WT mice ( respectively 53 . 5±7 . 6% and 6 . 5±1 . 1%; p<0 . 001 ) . Moreover , the phenotype of BAL neutrophils in ChemR23 KO and WT mice was evaluated at days 10 and 14 post-infection , using flow cytometry . No difference was seen in the neutrophil population ( F4/80− , Ly-6G+ , CD11b+ ) , in terms of FAS and ICAM-1 expression , which characterize the antitumoral/pro-inflammatory “N1” subset . The phenotype of BAL macrophages ( F4/80+ , Ly-6G− , CD11b+ cells ) was also evaluated by flow cytometry at days 10 and 14 post-infection and no difference between ChemR23 KO and WT mice were seen for CD209a ( DC-SIGN ) expression as a marker of the alternatively activated and anti-inflammatory “M2” subset ( data not shown ) . Altogether , these results do not support the recruitment of different subsets of neutrophils or macrophages to the lungs of infected ChemR23 KO and WT mice . The myeloperoxydase activity was also higher in ChemR23−/− mice than in WT mice ( Figure S3 ) . In contrast , lymphocytes were the main population in WT mice at day 10 post-infection ( 38 . 0±6 . 1% versus 8 . 5±1 . 5% for WT and ChemR23−/− respectively; p<0 . 01 ) . No significant differences were observed for cells categorized as macrophages on the basis of their appearance on cytospin preparations ( Figure 6B ) . Our data suggest the contribution of ChemR23 in pDC recruitment , and impaired pDC recruitment in ChemR23−/− mice might in turn explain lower synthesis of type I interferons and delayed viral clearance . In order to test the potential link between this cascade of events and the increased morbidity/mortality observed in KO mice , we performed pDC depletion and adoptive transfer experiments . Depletion of pDCs was achieved by intraperitoneal injections of the 120G8 monoclonal antibody , starting before PVM inoculation . Staining of spleen cells obtained at day 9 post-infection confirmed the efficacy of the protocol . With the exception of B cells , which were slightly but significantly ( p<0 . 05 ) decreased in the depleted WT group , no effect of the depleting antibody was observed on other subsets of inflammatory cells , including mDCs , macrophages and T cells ( Figure 7A and Figure S4 ) . Following PVM infection , a higher mortality rate ( 55% ) was observed for ChemR23−/− mice ( Figure 7A ) , in line with our previous experiments . However , pDC depletion in WT mice did not increase their mortality rate , compared to WT mice treated with a control IgG . Furthermore , pDC depletion did not reduce the difference in mortality rate between WT and ChemR23−/− mice . Adoptive transfer of pDCs was also performed . PDCs were purified from the spleen of WT and ChemR23−/− mice overexpressing Flt3 ligand . The generated pDCs were tested for their phenotype and functionality , and no difference in their ability to synthesize cytokines ( IFN-α and IL12p40 ) in response to CpG stimulation was detected in vitro ( Figure S5 ) . Moreover , pDCs from WT mice were able to migrate in response to chemerin with a typical bell-shaped curve culminating for concentrations around 1 nM , whereas pDCs from KO mice failed to migrate in response to chemerin ( Figure S6 ) . If defective pDC recruitment was causal in the differences observed between WT and ChemR23−/− mice , we expected that adoptive transfer of ChemR23-expressing pDCs to knock-out mice would reduce the mortality and weight loss in this group , as compared to knock-out mice receiving either pDCs from KO mice or a saline solution . Interestingly , these experiments showed the opposite effect ( Figure 7B ) . ChemR23−/− mice receiving a saline solution or pDCs from knock-out mice presented similar weight and survival curves , whereas ChemR23−/− mice receiving pDCs from WT mice displayed higher weight loss and mortality rate . We also observed more severe clinical signs in these animals , particularly reduced locomotor activity and increased crackles . These observations were complemented by the analysis of cells in BAL fluids , using flow cytometry . We showed that ChemR23−/− mice receiving pDCs from WT mice displayed an increase in the number of cells in BAL , as compared to ChemR23−/− mice receiving pDCs from KO mice , 14 days after infection/adoptive transfer of pDCs . The cells were mainly CD3+CD4+ and CD3+CD8+ T lymphocytes , while neutrophils were low at that time ( Figure 7C ) . In these settings , pDCs , NK cells and macrophages were low ( data not shown ) . Analysis of the cytokine levels in BAL fluid showed a sustained production of IL-6 compared to WT mice and higher levels of IFN-γ compared to WT and KO mice ( Figure S7 ) , while levels of KC were unchanged and no IL-10 was detected at this time-point ( data not shown ) . Finally , the restoration of the IFN-α production was detected using ELISA in the BAL fluid of ChemR23−/− mice receiving pDCs from WT mice , as compared with levels in the BAL fluid of WT mice ( Figure 7D ) . As lower pDC recruitment does not explain the higher immunopathology observed in ChemR23 KO mice , we further characterized the role of ChemR23 expression by leukocytes versus non leukocytic cells in PVM infection pathogenesis . To address this question , irradiated WT and KO mice were reconstituted with bone marrow ( BM ) cells harvested from KO and WT mice respectively , in order to generate WT mice having ChemR23-deficient leukocytes and conversely . Chimerism was assessed by flow cytometry 6 weeks post-irradiation and we observed that more than 90 percent of CD45+ spleen cells resulted from the graft ( GFP-positive cells; 94 . 0±1 . 5%; mean ± SEM for n = 6 independent experiments ) . Irradiated and reconstituted mice presented a lower susceptibility to PVM infection than sex- and age-matched non-irradiated mice ( data not shown ) . This observation is in line with previously published data showing lower asbestos-induced lung inflammation after myelo-ablative irradiation and BM transplantation [37] . As a result , all mice survived the PVM challenge and a relatively mild weight loss was observed . As shown in Figure 8 , KO mice reconstituted with WT leukocytes presented higher weight loss and displayed more severe illness signs than WT mice reconstituted with KO leukocytes . At the end of the observation period , mice were sacrificed and higher numbers of neutrophils , macrophages , CD8+ and CD4+ T cells were recovered in BAL fluids of KO mice reconstituted with WT leukocytes ( Figure 8A ) . These observations were confirmed in other experiments in which BAL fluids were harvested earlier ( day 14 post-infection ) , and including additional groups of WT and KO mice reconstituted respectively with WT and KO mice ( Figure 8B ) . Higher numbers of inflammatory leukocytes were found in BAL fluids of KO mice grafted with BM from KO mice , than in the two WT groups . However , as in the first experiment , KO mice reconstituted with WT leukocytes presented much higher BAL cell counts , as well as IL-6 and IFN-γ levels .
In this study , we investigated the role of the chemerin-ChemR23 system in a mouse model of viral pneumonia . Wild-type and ChemR23 knock-out mice were infected by PVM , the mouse counterpart of human RSV . The two viruses are closely related and evoke similar immune responses . The use of the natural mouse pathogen PVM was however preferred , because it replicates efficiently following a minimal virus inoculum , and recapitulates many of the clinical and pathologic features of the most severe forms of RSV infection in human infants . By contrast , mice are relatively resistant to infection by human RSV [10] . We demonstrated that ChemR23−/− mice develop a more severe inflammatory status than wild-type mice , resulting in a significant increase in morbidity and mortality rate . ChemR23−/− mice presented a higher recruitment of neutrophils and macrophages , higher myeloperoxydase activity , extended macroscopic and microscopic lesions , increased synthesis of pro-inflammatory cytokines ( e . g . IL-6 ) and chemokines ( e . g . CXCL-1/KC ) and higher levels of bioactive chemerin in the lung . As a direct consequence , ChemR23-invalidated mice showed a more severe respiratory dysfunction ( restrictive syndrome ) , evaluated by double chamber whole body plethysmography . We demonstrated a lower recruitment of pDCs to the lung of ChemR23−/− mice , despite a 100-fold higher viral load . In line with the high expression of ChemR23 on immature plasmacytoid dendritic cells and the chemoattractant activity of chemerin on pDCs in vitro , our data strongly suggest a major role of the chemerin/ChemR23 system in the recruitment of pDCs in vivo . This observation correlates with other studies reporting a link between chemerin expression and the recruitment of pDCs and NK cells in human inflammatory diseases of the skin [27] , [31] , [38] , [39] . The recruitment to lung of other ChemR23-bearing cells , including macrophages , myeloid dendritic cells and NK cells was apparently not affected in ChemR23−/− mice . This could be explained by the lower expression of ChemR23 on these cells and the redundancy with other chemoattractant molecules able to recruit these cell populations [31] , [40] , [41] . In this context , the higher number of mDCs and macrophages observed in lungs of infected ChemR23−/− mice might be explained by the higher inflammatory state and the higher production of chemokines acting independently of the chemerin/ChemR23 system . Chemerin levels were found to be increased in lung and airways during viral infection , reaching higher values in ChemR23−/− mice than in controls . As prochemerin transcript levels did not change during the course of the disease , the strong upregulation of chemerin in lung likely results from exudation of circulating prochemerin . The observation of bioactivity on ChemR23-expressing cells demonstrates that prochemerin was also processed to its active forms , presumably in part by neutrophil proteases . Chemerin and neutrophil recruitment increase indeed in parallel during the course of infection . The higher levels of chemerin immunoreactivity and bioactivity in ChemR23−/− mice may be explained in part by the stronger inflammatory response , resulting in more exudation and more efficient activation of prochemerin by proteases released by the higher number of neutrophils . However , basal immunoreactive chemerin levels are also increased in uninfected animals . Increased ( pro ) chemerin levels may therefore also result from defective removal of the protein from extracellular fluids , resulting from the lack of receptors ( ChemR23 ) able to bind and internalize chemerin . This hypothesis of chemerin scavenging by ChemR23-positive cells is supported by similar observations made on mice invalidated for chemokine receptors in which the circulating levels of the corresponding chemokine ligands were found to be elevated , as described for the CCL2/CCR2 axis [42] . PDCs play an important role in anti-viral immunity by their capacity to secrete large amounts of type I IFNs ( including IFN-α and -β ) and induce effector CD8+ T-cell response upon viral infection [13] . Type I IFNs have pleiotropic anti-viral functions: they increase the resistance of non-infected cells to viral infection , inhibit viral gene transcription , induce apoptosis of infected cells , induce B cell differentiation into antibody-secreting plasma cells , promote the cytotoxic activity of NK and CD8+ T cells , promote survival and proliferation of CD8+ T cells , as well as differentiation and maturation of DCs [18] , [43]–[46] . As a result , type I IFNs promote an efficient acquired immune response and facilitate viral clearance in PVM-induced mouse pneumonia models [7] . In PVM-infected ChemR23−/− mice , it is therefore tempting to link the defective pDC recruitment to the lower synthesis of type I IFNs , lower acquired immune response ( low CD8+ T cell counts and IL-12p40 production ) , and delayed viral clearance . In this context , parameters unaffected ( IFN-γ production ) or displaying mild differences ( anti-PVM antibodies production ) must be interpreted taking into account a 100-fold higher viral load in ChemR23 KO mice . In our study , no difference in viral loads were observed before day 8 post-infection , suggesting that intrinsic viral replication of PVM was not affected by ChemR23 deficiency . Moreover , replication of PVM was reported in type I and II pneumocytes , alveolar macrophages , eosinophils , and bronchiolar epithelial cells , which are not described to express ChemR23 [34]–[36] , [47 and unpublished observations] . Therefore , to investigate whether lower recruitment of pDCs in PVM-infected ChemR23−/− mice might explain the excessive morbidity , pDC depletion and adoptive transfer experiments were performed . In depletion experiments , mortality rates were essentially unchanged both in WT and ChemR23−/− mice , suggesting that pDCs do not play a major role in the poor clinical outcome of PVM-infected ChemR23−/− mice . Moreover , adoptive transfer of ChemR23-expressing pDCs to KO mice enhanced clinical symptoms and decreased survival . Interestingly , previous studies have shown that C57BL/6 mice lacking type I IFN receptor ( IFN-αβR−/− ) or functional TCR exhibit defective acquired response and higher PVM titers , but milder histological lesions , decreased PMN recruitment and lower mortality rate [7] , [8] . Taken together , these observations strongly suggest that the extent of lung tissue damage is not directly related to the viral load , but rather to the efficiency of the anti-PVM immune response , including the acquired Th1 component . In this context , improving the immune response by transferring ChemR23-expressing pDCs may indeed lead to a more severe lung disease . In other models however , pDC depletion resulted in an enhancement of the inflammatory state . In pDC-depleted BALB/c mice infected by human RSV , a higher Th2 response was observed , which was found to be IFN-independent [21] . PDC depletion in BALB/c mice also resulted in enhanced airway inflammation and eosinophilia in asthma models , whereas adoptive transfer of pDCs suppressed inflammation and the Th2 response [25] . This immunomodulatory role of pDCs on Th2 responses is type I IFN-independent and mediated by their ability to induce regulatory T lymphocytes and/or to suppress effector T cell generation [14] , [23]–[25] . The influence of pDCs may therefore affect the inflammatory state positively or negatively , according to the specific model considered and the main immune responses triggered in this model . As defective pDC recruitment did not appear to underlie the more severe clinical outcome observed in ChemR23 KO mice , we hypothesized that these mice might have lost a direct anti-inflammatory pathway involving ChemR23 . This hypothesis is indeed supported by previous results from our group , showing a ChemR23-dependent anti-inflammatory role of chemerin in a LPS-induced acute lung inflammation model [28] . To assess whether this anti-inflammatory effect is mediated by ChemR23-expressing leukocytes or other cell populations , chimeric mice were generated following lethal irradiation and BM adoptive transfer . These experiments showed that mice lacking ChemR23 only in BM-derived cells behaved essentially as wild-type animals , while restoring ChemR23-expression in leukocyte populations did not protect KO mice from excessive inflammatory response to PVM infection . It appears therefore that the chemerin/ChemR23 system has anti-inflammatory properties on a non-leukocytic cell population , which is presently not identified . Lung endothelial cells might constitute a candidate cell population , since ChemR23 expression was recently described in human endothelial cells [48] , [49] . Chemerin might therefore regulate trafficking of inflammatory cells such as PMNs , by modulating the release of chemokines by endothelial cells , or their interactions with leukocytes . Finally , our chimera experiments showed that mice expressing ChemR23 exclusively in leukocytes displayed the most severe inflammatory phenotype , with the highest levels of IFN-γ and CD8 T cells in BAL fluids . Similar results were obtained following adoptive transfer of ChemR23-expressing pDCs in KO mice . These observations suggest that ChemR23-expressing leukocytes , among which pDCs , contribute significantly to the innate and acquired cytotoxic responses and the resulting inflammatory insult to lung parenchyma . In conclusion , ChemR23-deficient mice are more susceptible to PVM infection . The recruitment of pDCs is impaired in these mice , resulting in a reduction of type I IFN synthesis and delayed viral clearance . However , the stronger inflammatory status and the resulting higher morbidity and mortality , are not the consequence of impaired pDC recruitment . Chemerin appears therefore to have anti-inflammatory properties , by acting on ChemR23 expressed by non-leukocytic cells , thereby dampening the inflammatory response promoted by the viral infection . Further analyses are needed to determine the precise underlying mechanisms and cell types involved in these processes .
This study was carried out in strict accordance with the national , european ( EU Directives 86/609/EEC ) and international guidelines in use at the Université Libre de Bruxelles . All procedures were reviewed and approved by the ethical committee ( Commission d'Ethique du Bien-Etre Animal , CEBEA ) of the Université Libre de Bruxelles ( Permit Number: 222N and 341N ) . All efforts were made to minimize suffering . Eight to twelve weeks-old C57BL/6 mice ( Harlan Netherlands ) were used throughout these studies . ChemR23-deficient mice ( KO ) were obtained from Deltagen ( CA , USA ) through Euroscreen S . A . ( Brussels , Belgium ) . They were backcrossed for 12 generations into the C57BL/6 background in a specific pathogen free environment . Wild-type littermates from F1 matings were used as controls . The J3666 strain of PVM ( initially provided by A . J . Easton ) was passed in BALB/c mice and grown once onto BSC-1 cells to produce the viral stock . Randomly selected aliquots of the stock yielded highly reproducible titers on BSC-1 cells , amounting to 1×106 PFU/ml . Mice were inoculated under brief anaesthesia ( ketamine , Pfizer , 50 mg/kg , and xylazine , Bayer , 10 mg/kg , i . p . ) by intranasal instillation of 50 µl of a viral suspension containing 1000 PFU and 1% BSA in PBS . Control mice were inoculated with PBS containing 1% BSA . At selected time intervals ( 6 , 8 and 10 days post-infection ) , groups of minimum 6 mice were sacrificed with sodium thiopental ( 5 mg/animal , i . p . ) . In some experiments , broncho-alveolar lavage fluids were obtained by flushing the lungs with sterile 0 . 9% NaCl , and differential cell counts were performed on cytospin preparations after Diff-Quick staining ( Dade Behring ) . Lung function was assessed in groups of five mice using a whole body double chamber plethysmograph ( Buxco ) and IOX software ( EMKA Technologies ) as described previously [50] . Briefly , at selected time points ( before inoculation and 1 , 3 , 5 , 6 , 7 and 8 days post-inoculation ) , awake mice were placed between two compartments ( nasal and thoraco-abdominal ) , in which flow variations were recorded during 5 minutes and used to determine the following parameters: tidal volume ( TV ) ; expiratory balance; RT/Te , in which RT is the relaxation time determined as the time needed to expire 64% of the inspired volume and Te the expiratory time; end expiratory pause ( EEP ) , Te – RT; enhanced pause ( Penh ) ; specific airway resistance ( sRaw ) . At day 9 post-infection , left lungs were insufflated with 500 µl of 4% paraformaldehyde , and embedded in paraffin . Sections ( 5 µm ) were stained with haematoxylin and eosin and assessed by light microscopy . Paraffin embedded lung sections ( 5 µM ) were also used for PVM antigen immunofluorescence staining using an anti-PVM antiserum ( dilution 1/100 , rabbit ) and an Alexa Fluor 488-conjugated anti-rabbit IgG antibody ( dilution 1/1000 , Invitrogen ) . Images were captured with an Axioplan 2 imaging fluorescence microscope , equipped with a Diagnostic Spot digital camera and analyzed by the Spot Advanced Soft Imaging System and Adobe Photoshop 7 . 0 . The lenses used were 20×/0 . 5 , ∞/0 . 17; 40×/0 . 75 , ∞/0 . 17 and 100×/1 . 30 , ∞/0 . 17 . At selected time points , viral titers were determined by standard plaque assay as previously described [34] . Briefly , the right lung was homogenized in PBS containing 1% BSA and successive 10-fold dilutions of the supernatant were used to infect BSC-1 cell cultures . The viral suspensions were left to adsorb for 3 h at 31°C , after which the cell monolayers were covered with 1 ml of 0 . 6% agarose in MEM containing 2% FBS . After incubation at 31°C for 12 days , the agar overlay was removed , the remaining cells were stained with crystal violet , and titers were determined by counting the number of plaque forming units ( PFUs ) . The serum was evaluated for anti-PVM antibodies on days 9 , 10 , and 11 post-inoculation and compared to serum obtained from control mice using the SMART-M12 kit ( Biotech Trading Partners ) . Lung total RNA was extracted and purified using the RNeasy Mini Kit ( Qiagen ) . After DNase I treatment ( Ambion ) , samples were reverse transcribed into cDNA using random hexamers ( Roche ) as primers and the Superscript II polymerase ( Invitrogen ) . RT-PCR products were analyzed by quantitative real-time RT-PCR . The sequence of primer pairs used for mouse chemerin , IFN-α , IFN-β , CANX , and YWHAZ is provided in Table S1 . Raw data were normalized for each sample using YWHAZ and CANX gene expression as references . At selected time points , right lungs were homogenized in PBS and supernatants were assayed for TNF-α , IL-6 , IFN-γ and KC/CXCL1 using cytometric bead array-based immunoassays ( CBA Flex set , BD Biosciences ) , a dual-laser flow cytometer ( FACSCalibur , BD Biosciences ) and the FCAP Array software ( BD Biosciences ) for analysis , following the manufacturer's instructions . IL-12p40 , IL-17 , chemerin , and IFN-α were determined by ELISA in lung homogenates ( R&D Systems , Abingdon , UK ) . Chemerin , as well as IL-6 , TNF-α , IL-10 , IFN-γ ( BD Biosciences ) , IFN-α and KC ( R&D Systems ) were also measured in BAL fluids using ELISA according to the manufacturer's instructions . BAL fluids from two infected WT and ChemR23−/− mice were pooled , filtered and loaded onto a C18 reverse phase column ( 2 . 1×250 mm; Vydac ) . Fractions from a 25–50% acetonitrile ( CH3CN ) gradient ( 0 . 5%/min ) in 0 . 1% TFA were collected . The biological activity was next measured by a calcium-mobilizing assay based on the bioluminescence of aequorin , using CHO-K1 cells co-expressing mouse ChemR23 , apoaequorin and Gα16 [30] . CHO-K1 cells co-expressing only apoaequorin and Gα16 were used as control . Fractions from infected WT mice were also concentrated and depleted of acetonitrile using a vacuum SpeedVac concentrator and then tested undiluted in the same assay . Results ( expressed as luminescence units ) were normalized to the response to 20 µM ATP . A dose-response curve for chemerin on ChemR23-positive CHO-K1 cells was also generated using 0 . 01 to 100 nM recombinant mouse chemerin ( R&D Systems ) . Lungs were perfused with 10 ml PBS through the right ventricle , dissected , minced , and incubated with 2 mg/ml collagenase D and 0 . 02 mg/ml DNAse I ( Roche ) for 1 hour at 37°C . In some experiments , cell suspensions were obtained from BAL fluids as previously described . After lysis of red blood cells and blockade of non-specific binding to FcR with anti-CD16/CD32 monoclonal antibodies ( mAbs ) , viable cells were counted by trypan blue exclusion . Cells were stained with mAbs directed against F4/80 ( FITC or APC ) , CD11b ( FITC or PerCp-Cy5 . 5 ) , CD4 ( FITC or Horizon V450 ) , CD11c ( PE or APC ) , I-A/I-E ( PE ) , CD19 ( PE ) , NK1-1 ( PE ) , Gr-1 ( PerCp-Cy5 . 5 ) , Ly6-G ( Alexa 700 ) , CD8 ( PerCp or Horizon V500 ) , ICAM-1 ( PE ) , CD95 ( PE-Cy7 ) , CD209a ( biotinylated ) , Streptavidin-FITC , CD3 ( APC ) , mPDCA ( APC ) , and isotype controls ( all from BD Biosciences except F4/80-FITC from AbD Serotec ) . All samples were analyzed using a dual-laser flow cytometer ( FACSCalibur ) using the CellQuest software ( BD Biosciences ) or a four-laser flow cytometer ( FACS LSRFortessa ) using FACS Diva software ( BD Biosciences ) . Depletion of pDCs was performed in WT and ChemR23−/− mice using a monoclonal antibody against pDCs ( 120G8 , Dendritics ) . Mice received four i . p . injections of 125 µg 120G8 , every other day , beginning the day before infection . Depletion efficacy and specificity was assessed by assaying cell populations in spleen . After mechanical disruption , isolated spleen cells were stained for pDCs , mDCs , macrophages , T and B cells using the following set of antibodies: B220 ( FITC ) , CD11c ( PE ) , CD11b ( PerCp-Cy5 . 5 ) , mPDCA ( APC ) ; or F4-80 ( FITC ) , CD19 ( PE ) , CD11b ( PerCp-Cy5 . 5 ) and CD3 ( APC ) . The generation , purification and transfer of isolated pDCs were performed as previously described [25] . Briefly , wild type and ChemR23−/− mice were injected subcutaneously with 5×106 melanoma B16 cells ( syngeneic to C57BL/6 mice ) expressing the dendritic cell growth factor FMS-like tyrosine kinase 3 ligand ( Flt3-ligand ) ( kindly provided by M . Moser , ULB , Gosselies ) . Three weeks later , mice were sacrificed and their spleen mechanically disrupted . Released cells were then passed through a 70 µM cell strainer ( BD Biosciences ) . Untouched pDCs from single-cell suspensions were then isolated by negative selection ( Plasmacytoid Dendritic Cell Isolation Kit II , Miltenyi Biotec ) . Briefly , non-pDC cells were labelled with a cocktail of biotin-conjugated monoclonal antibodies and depleted by retention on a MACS column . The viability and purity of the pDC preparations were respectively over 95% and 90% , as assessed by flow cytometry following staining using respectively propidium iodide ( PI ) and mAbs against B220 ( FITC ) , CD11c ( PE ) , CD11b ( PerCp-Cy5 . 5 ) and mPDCA ( APC ) ( all from BD Biosciences ) . Thereafter , 106 pDCs from WT mice in 0 . 9% NaCl were injected in the tail vein of ChemR23−/− mice infected the same day by PVM . Weight loss and mortality rate were compared between this group ( KO+pDC WT ) , infected ChemR23−/− mice receiving the saline carrier ( KO+0 . 9% NaCl ) , and infected ChemR23−/− mice receiving 106 pDCs purified from ChemR23−/− mice ( KO+pDC KO ) . Chimeric mice were generated to discriminate between the roles of ChemR23 expressed by leukocytes or other cell types . For this purpose , 8 week old WT and ChemR23 KO mice received a lethal irradiation dose of 800 Rad TBI delivered by a 137Cesium irradiator . Twenty four hours later , irradiated mice were reconstituted by i . v . injection with 20×106 syngeneic BM cells isolated from 6 to 8 week old WT or KO mice . BM cells were obtained by flushing BM from femurs and tibiae of donor mice . After 4 weeks , mice were infected and chimerism was assessed on spleen cells purified from mice reconstituted with BM cells from age-matched WT C57BL/6 mice expressing green fluorescent protein ( GFP ) ( % GFP+ cells among CD45+ spleen cells ) . Significance was determined using Student's t test or one-way analysis of variance , using the Prism4 software ( GraphPad ) . The Student-Newman-Keuls test was used for pairwise comparisons . Kaplan-Meier survival curves were compared using the logrank test . For all tests , p<0 . 05 was considered as significant . ( Source: http://www . ncbi . nlm . nih . gov ) [Mus musculus] Chemerin: 71660/NP082128; ChemR23/cmklr1: 14747/P97468; TNF-α: 21926/CAA68530; IL-10: 16153/NP034678; TGF-β1:21803/AAH13738; KC/CXCL-1: 14825/P12850; IL-6: 16193/P08505; IFN-α: 15962/P01572; IFN-β: 15977/P01575; IFN-γ: 15978/P01580; IL-5: 16191/P04401; IL-13: 16163/P20109; IL-17: 16171/Q62386; IL-12p40: 16160/P43432; IL-1β: 16176; CANX: 12330; YWHAZ: 22631 . | Infections of the lower respiratory tract by single-stranded RNA viruses represent a major health problem worldwide . Animal models indicate that the severity of infections caused by these viruses is due essentially to an excessive primary immune response of the host , rather than the direct cytopathogenicity of the viruses . Plasmacytoid dendritic cells have been reported to play an important role in anti-viral immunity , but the factors responsible for the recruitment of these cells to the infected lung were unknown . This study depicts the roles of the G protein-coupled receptor ChemR23 in the recruitment of plasmacytoid dendritic cells and anti-viral immunity in a mouse model of acute viral pneumonia . The data also highlight the role of ChemR23 in dampening the lung inflammatory response . This latter effect is independent of pDC recruitment but involves non-leukocytic cells . This observation is of particular interest considering the established role of airway endothelial and epithelial cells in the immune responses following bacterial , viral and fungal infections . Our results suggest therefore that the chemerin/ChemR23 system might be considered as a target for anti-viral and anti-inflammatory therapies . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"respiratory",
"infections",
"immunity",
"to",
"infections",
"immunology",
"pulmonology",
"viral",
"pneumonia",
"infectious",
"diseases",
"inflammation",
"respiratory",
"syncytial",
"virus",
"infection",
"immune",
"response",
"clinical",
"immunology",
"immunity"... | 2011 | ChemR23 Dampens Lung Inflammation and Enhances Anti-viral Immunity in a Mouse Model of Acute Viral Pneumonia |
In the murine model of Leishmania major infection , resistance or susceptibility to the parasite has been associated with the development of a Th1 or Th2 type of immune response . Recently , however , the immunosuppressive effects of IL-10 have been ascribed a crucial role in the development of the different clinical correlates of Leishmania infection in humans . Since T cells and professional APC are important cellular sources of IL-10 , we compared leishmaniasis disease progression in T cell-specific , macrophage/neutrophil-specific and complete IL-10-deficient C57BL/6 as well as T cell-specific and complete IL-10-deficient BALB/c mice . As early as two weeks after infection of these mice with L . major , T cell-specific and complete IL-10-deficient animals showed significantly increased lesion development accompanied by a markedly elevated secretion of IFN-γ or IFN-γ and IL-4 in the lymph nodes draining the lesions of the C57BL/6 or BALB/c mutants , respectively . In contrast , macrophage/neutrophil-specific IL-10-deficient C57BL/6 mice did not show any altered phenotype . During the further course of disease , the T cell-specific as well as the complete IL-10-deficient BALB/c mice were able to control the infection . Furthermore , a dendritic cell-based vaccination against leishmaniasis efficiently suppresses the early secretion of IL-10 , thus contributing to the control of parasite spread . Taken together , IL-10 secretion by T cells has an influence on immune activation early after infection and is sufficient to render BALB/c mice susceptible to an uncontrolled Leishmania major infection .
Infection with intracellular protozoan parasites of the genus Leishmania leads to a broad range of disease manifestations in humans , ranging from an asymptomatic carrier status or localized , self-healing cutaneous leishmaniasis to disseminating visceral disease ( kala azar ) [1] . The outcome of infection depends on the parasite species , but is also influenced by the host immune response [2] , [3] . In naturally resistant mouse strains such as C57BL/6 or C3H , IL-12 , secreted mainly by dendritic cells ( DC ) , has the essential role of inducing a Th1 immune response . The Th1 effector cytokine IFN-γ leads to an activation of infected macrophages and parasite killing . Conversely , the susceptibility of BALB/c mice has been attributed to a Th2 immune response characterized by the secretion of IL-4 , IL-5 and IL-13 . Accordingly , IL-4−/− BALB/c mice are able to control infection with some Leishmania major strains at least partially [4] and BALB/c mice treated with anti-IL-4 Ab at the time of challenge exhibit a healing phenotype [5] . There is also convincing evidence that the early IL-4 response is confined largely to an oligoclonal population of CD4+ T cells with a Vβ4Vα8 T-cell receptor that recognize the Leishmania antigen LACK ( Leishmania homologue of receptors for activated C kinase ) [6] . However , this classical Th1/Th2 paradigm has been challenged by recent findings in humans and some mouse models: for instance , IL-4−/− and IL-4Rα−/− BALB/c mice are not resistant against all L . major strains [7] , and , whereas IL-4−/− and IL-4Rα−/− BALB/c mice are resistant to infection with L . amazonensis , IL-4−/− C57BL/6 and IL-4−/− C3H mice are not [8] , [9] . In addition , IL-4−/− and IL-4Rα−/− BALB/c mice are more susceptible to infection with L . donovani [10] and also experimental visceral leishmaniasis in C57BL/10 mice is independent of IL-4 and the associated Th2 immune response [8] , [11] . In contrast , recent studies have emphasized the role of IL-10 as an important regulatory cytokine involved in parasite control in mice and humans . Originally identified as a Th2 cell-derived factor , it is now known to be secreted also by regulatory T cells ( Treg ) , Th1 cells , CD8+ T cells , B cells , macrophages , DC , mast cells , eosinophils , NK cells and some cell types not belonging to the immune system [12] , [13] , [14] . IL-10 has broad anti-inflammatory effects on several cell types: it inhibits phagocytosis , the expression of MHC class II and co-stimulatory molecules and the secretion of proinflammatory cytokines by macrophages and DC . It also limits the ability of macrophages to kill intracellular organisms . In addition to its influence on cells of the innate immune system , IL-10 has direct suppressive effects on T cell responses [12] . Accordingly , IL-10−/− mice develop excessive Th1 cell responses , resulting for example in spontaneous inflammatory bowel disease [15] or overwhelming immunopathology upon infection with parasites like Toxoplasma gondii [16] , Trypanosoma [17] , Schistosoma [18] or Trichinella spiralis [19] . IL-10 has also been shown to facilitate the spread of Leishmania parasites . IL-10−/− mice on a BALB/c background were able to control infection with L . major [20] , and IL-10−/− mice on a C57BL/6 background , in contrast to their wild-type littermates , achieved sterile immunity to this parasite [21] . In humans , cutaneous leishmaniasis , visceral leishmaniasis and post-kala-azar dermal leishmaniasis have been associated with increased levels of IL-10 [22] , [23] , [24] , [25] , [26] . Thus , IL-10 undoubtedly plays a crucial role in Leishmania disease progression . However , a variety of cell types is able to secrete IL-10 and there is no consensus about the cellular sources contributing to the IL-10-mediated suppression of the anti-leishmanial immune response . Belkaid et al . demonstrated that parasite persistence and the maintenance of immunity to re-infection in C57BL/10 mice are dependent on the CD4+ CD25+ FoxP3+ Treg cell-derived IL-10 [27] , [28] . In contrast , following infection of C57BL/6 mice with the L . major strain NIH/Sd , which produces nonhealing dermal lesions in a Th1-polarized setting , it was shown that IL-10-producing CD4+ CD25− FoxP3− Th1 cells rather than Treg cells are the major contributors to immune suppression [29] . This was also true for BALB/c IL-4 receptor-deficient mice infected with L . major . Similar results were obtained in a mouse model of visceral leishmaniasis , following infection with L . donovani [30] , as well as in human visceral leishmaniasis [26] . In a BALB/c IL-10−/− mouse model , however , the observed resistance to infection was attributed to the absence of IL-10 secreted by macrophages [20] . Furthermore , the overexpression of IL-10 under control of the MHC class II Ea promoter resulted in the increased susceptibility of the mice to infection with L . major , supporting a role for IL-10 derived from professional APC [31] . Recently , IL-10 secreting B cells have also been shown to influence the immune response in BALB/c mice infected with L . major [32] . Taken together , IL-10 secretion by CD4+ CD25+ FoxP3+ Treg cells , CD4+ CD25− FoxP3− effector T cells , B cells and professional APC has been connected with the suppression of the anti-leishmanial immune response . In the present study , we directly investigated the contribution of IL-10 from different cellular sources to L . major disease progression by using mice with a selective deficiency for IL-10 in T cells [33] or macrophages and neutrophils [34] , and comparing them with complete IL-10-deficient animals . The results show that the enhanced protection of complete IL-10-deficient mice is entirely attributable to the lack of T cell-derived IL-10 , while macrophage- or neutrophil-derived IL-10 has no effect on disease progression . In addition , we analyzed the mechanism underlying this enhanced protection and demonstrated that the suppression of the early antigen-dependent IL-10 secretion seems to contribute to the protection mediated by DC-based vaccination against leishmaniasis [35] , [36] .
Although IL-10 has been shown to play a role in protective immunity against L . major , the relative contributions of the different cellular sources of IL-10 are not clear . To directly investigate the role of T cell-derived and macrophage-derived IL-10 , IL-10fl/fl CD4-Cre+ T cell-specific IL-10-deficient mice were compared to IL-10fl/fl LysM-Cre+ macrophage/neutrophil-specific IL-10-deficient mice , IL-10fl/fl EIIa-Cre+ complete IL-10-deficient mice , and IL-10fl/fl Cre− control mice on a C57BL/6 background . The generation of these mice and the high specificity and efficiency of the cell type-specific deletion of the first IL10 exon have been described previously [33] , [34] . To investigate disease progression , these T cell-specific , macrophage/neutrophil-specific and complete IL-10-deficient mice were infected with L . major promastigotes into the right hind footpad and footpad swelling was monitored weekly ( Figure 1A ) . Surprisingly , T cell-specific and complete IL-10-deficient mice displayed a significantly ( p<0 , 01 ) increased footpad swelling , compared to macrophage/neutrophil-specific IL-10-deficient mice and Cre− control animals , as soon as one week after infection ( Figure 1B ) . In contrast , we could not observe any difference in footpad swelling at all later time points , including the peak of disease manifestation at 2 to 3 weeks after infection . Furthermore , there was no difference in the number of regional lymph node cells , draining the site of infection at any time point ( Figure 2C and data not shown ) . To rule out that the observed early footpad swelling of the T cell-specific IL-10 deficient mice is an unspecific reaction to injection trauma , we compared footpad swelling following injection of live L . major promastigotes or PBS respectively . One week after injection of PBS no significant footpad swelling could be observed ( Figure 1C ) . As footpad swelling reflects not only parasite replication but also local inflammation , we determined the parasite load in the regional lymph nodes of these mice at different time points after infection . Unexpectedly , we found only minor differences in the parasite load at one week after infection ( Figure 1D ) , with slightly reduced parasite numbers in the T cell-specific and the complete IL-10-deficient mice , indicating that the augmented footpad swelling of these mouse strains is due to an enhanced inflammation at this early time point . However , two weeks after infection , despite comparable footpad swelling , the T cell-specific and the complete IL-10-deficient mice displayed markedly reduced parasite loads within the draining lymph nodes , compared to the macrophage/neutrophil-specific IL-10 mutant and the Cre− control mice ( Figure 1E ) . At later time points , parasite loads were equally low in all investigated mouse strains , consistent with the healing phenotype of the naturally resistant C57BL/6 background ( Figure 1F ) . As the influence of IL-10 on the chronic phase of L . major infection and especially its role for sterile immunity has already been investigated before [21] , [27] , [37] , we further concentrated on the effects of T cell-derived IL-10 on the activation of the immune system early after infection . While all our mouse strains displayed an elevated total cell number in the draining lymph nodes already 3 days post infection , the increased footpad swelling of the T cell-specific and the complete IL-10-deficient mice was not detectable until 5 days post infection ( Figure 2A and C ) . To further characterize the early enhanced inflammation of the T cell-specific IL-10 mutant mice , the relative amounts of CD4+ and CD8+ T cells , CD4+ CD25+ FoxP3+ Treg cells , CD45R+ B cells , CD49b+ NK cells and F4/80+ macrophages in the popliteal lymph nodes of infected mice and in the infected feet were analyzed by flow cytometry . The early increase in the total lymph node cell count was associated with a relative increase of B cells and a decrease of CD4+ T cells , CD4+ CD25+ FoxP3+ Treg cells , CD8+ T cells and macrophages ( Figure 2D ) . In accordance with the uniform increase in absolute lymph node cell numbers , this shift in cell populations did not differ between the IL-10 mutant strains . All cell populations could also be found in low absolute numbers in the infected feet . However , cell type composition did not significantly differ between the IL-10 mutant strains nor change during the first week of infection ( Figure 2B ) . To investigate a functional correlate of the early increased footpad swelling of the T cell-specific and the complete IL-10-deficient mice , we restimulated lymph node cells of the different IL-10-deficient mouse strains with L . major lysate 7 days after infection and assayed for the production of IFN-γ , IL-4 , IL-10 and IL-17 . The increased footpad swelling was associated with a significantly enhanced antigen-dependent secretion of IFN-γ , while we found only a slight increase in IL-4 secretion by the complete IL-10-deficient mice ( Figure 2E and F ) . No significant IL-10 or IL-17 production was detectable in any mouse strain at that time point ( data not shown ) . As a variety of cell types is known to be able to secrete IFN-γ , we next determined the cell population which is suppressed by the early T cell-derived IL-10 . To this end , we performed intracellular cytokine staining for IFN-γ of lymph node cells of the different IL-10 mutant mouse strains 7 days after infection with L . major ( Figure 2G ) . Surprisingly , the early enhanced secretion of IFN-γ by lymph node cells of the T cell-specific and the complete IL-10 mutant mice is not only due to antigen-specific CD4+ T cells , reflecting a Th1 immune response , but also to CD8+ T cells , whereas there was no enhanced secretion of IFN-γ by macrophages or NK cells . It has been shown previously that complete IL-10-deficient mice on the naturally susceptible BALB/c background are able to control an infection with L . major . This has been attributed mainly to the lack of IL-10 secretion by macrophages [20] . In our model , using mice on the naturally resistant C57BL/6 background , only T cell-specific but not macrophage/neutrophil-specific IL-10-deficient mice displayed an altered disease progression . Therefore , we addressed the question whether abrogating the T cell-specific IL-10 secretion also has an effect on disease progression in the naturally susceptible BALB/c background . For this purpose , we backcrossed the C57BL/6 IL-10fl/fl CD4-Cre+ mice onto the BALB/c background . To obtain complete IL-10-deficient mice on a BALB/c background , IL-10fl/fl Cre− mice were crossed with CMV-Cre+ mice [38] . Following infection with L . major , the T cell-specific IL-10 mutant BALB/c mice displayed a healing phenotype , indistinguishable from the complete IL-10-deficient BALB/c mice ( Figure 3A ) . The reduced footpad swelling was accompanied by a marked control of parasite replication , as demonstrated by an up to three log reduced parasite load in the popliteal lymph nodes ( Figure 3B ) . Interestingly , the T cell-specific as well as the complete IL-10-deficient BALB/c mice also showed an increased footpad swelling early after infection , comparable to the respective IL-10 mutant mice on a C57BL/6 background ( Figure 3A and 4A ) . As the complete IL-10-deficient BALB/c mice were breeding very poorly , and no differences could be detected between the T cell-specific and the complete IL-10-deficient mice , only T cell-specific IL-10-deficient and Cre− control mice were used for the further experiments . Further investigation of the parasite load and immune response of BALB/c mice two weeks after infection with L . major showed that despite an increased footpad swelling of the T cell-specific IL-10-deficient mice ( Figure 4A ) there were again no differences in parasite load at that early time point ( Figure 4B ) . To determine the type of immune response underlying the increase in early footpad swelling in mice of the BALB/c background , cytokines in the supernatants of cultured lymph node cells were determined by ELISA . Surprisingly , not only the Th1 marker cytokine IFN-γ , but also the Th2 marker cytokine IL-4 was significantly upregulated in the T cell-specific IL-10 mutant mice ( Figure 4C–E ) . Since IL-10 secretion by CD4+ CD25+ FoxP3+ Treg cells [27] , [28] , [39] as well as CD4+ CD25− FoxP3− Th1 cells [29] , [30] , [40] has been implicated in the control of chronic infection with Leishmania , it was important to determine the T cell population secreting IL-10 early after infection . To this end , naïve wild-type BALB/c mice were infected with L . major promastigotes in the hind footpad . Two weeks later , the IL-10-secreting cells within the draining popliteal lymph nodes and the footpad lesions were identified by a cytokine secretion assay . Staining for CD4 and FoxP3 or CD25 showed that the majority of CD4+ IL-10-secreting cells in the draining lymph nodes were FoxP3− and CD25− . In the infected footpads , there was only a small number of CD4+ IL-10-secreting cells , the majority of which were FoxP3+ and CD25+ ( Figure 5A–F ) . Furthermore , T cells represented only a small proportion of the IL-10 secreting cells . Staining for CD8 , CD45R , CD11c and F4/80 revealed that all these cell populations contribute to IL-10 secretion in the draining lymph nodes and the footpads of L . major-infected wild-type BALB/c mice , emphasizing the profound differences in the effects of IL-10 from different cellular sources ( Figure 5G ) . According to current paradigm , a maximal secretion of the proinflammatory cytokines IFN-γ , TNF and IL-2 by antigen-specific T cells after pathogen contact is the major prerequisite for successful vaccination against L . major [41] . The early increase in inflammation accompanied by a better control of parasite replication in both T cell-specific IL-10-deficient C57BL/6 and BALB/c mice following infection with L . major raised the question whether a reduced antigen-specific secretion of IL-10 may contribute to effective vaccination against the disease . To test this hypothesis , we used a DC-based vaccination protocol which has been shown to induce highly effective and solid immunity against L . major infection [35] , [36] , [42] . To explore the effect of DC-mediated protection on the early secretion of IL-10 by antigen-specific T cells , we compared the cytokine activity of cells in the lymph nodes draining the lesions of vaccinated versus non-vaccinated mice at different time points after infection with L . major . Interestingly , the only significant differences in the antigen-dependent cytokine secretion between vaccinated and non-vaccinated mice were indeed observed the first two weeks after infection . In the very first week after infection the vaccinated mice produced less than half the amount of IL-10 and IL-4 compared to the non-vaccinated mice ( Figure 6B and D ) . At this time point , we did neither observe a difference in footpad swelling ( Figure 6A ) nor in parasite load between the two groups ( data not shown ) . Already two weeks after infection the vaccinated mice showed a significant increase in IFN-γ secretion ( Figure 6C ) . To further investigate if the reduced IL-10 secretion by lymph node cells of vaccinated mice one week after infection is accompanied by a reduced number of IL-10-secreting CD4+ T cells , an IL-10 cytokine secretion assay was performed . Interestingly , despite comparable numbers of all IL-10-secreting lymph node cells between the vaccinated and the non-vaccinated mice ( Figure 6E ) , the number of IL-10-secreting CD4+ T cells was reduced by half in the successfully vaccinated mice ( Figure 6F ) .
The cytokine IL-10 has a major impact on the regulation of inflammation and the progression of a multitude of infectious diseases ( reviewed in [14] , [43] , [44] ) . During infection with intracellular parasites of the genus Leishmania , IL-10 undoubtedly plays an important role in disease development and pathogen persistence [20] , [21] , [22] , [23] , [24] , [27] , [37] , [39] . However , the cellular source of immune suppressive IL-10 is less clear , as activated effector T cells [29] , [30] , [40] , Treg cells [27] , [28] , [39] , macrophages [20] , [31] , [45] , DC [42] as well as B cells [32] have been implicated in inhibiting the effective control and clearance of the parasite . Furthermore , the influence of IL-10 on the early induction of an immune response against L . major has been scarcely investigated . In the present study , we used T cell-specific , macrophage/neutrophil-specific and complete IL-10-deficient mice to directly address these questions . As previous studies on the role of IL-10 in leishmaniasis were mostly investigating differences in disease progression on the naturally resistant C57BL/6 background , this study is the first to show that disabling the secretion of IL-10 by T cells is sufficient to render otherwise susceptible BALB/c mice resistant to an infection with the parasite . The healing phenotype is accompanied by an elevated specific inflammatory immune response very early after infection . We further show that DC-based vaccination against leishmaniasis suppresses the early secretion of IL-10 following challenge infection . Following infection with L . major , the T cell-specific and complete IL-10-deficient mice on the C57BL/6 as well as the BALB/c background are characterized by an increased early inflammation , which is due to the enhanced secretion of IFN-γ by CD4+ as well as CD8+ T cells . This finding is in accordance with recent data of low dose infection models in C57BL/6 mice , where CD8+ T cells via secretion of IFN-γ significantly contribute to the induction of protection against L . major [46] , [47] . In contrast , these effects have not been observed in earlier studies with high dose infection of MHC class II−/− or β2-microglobulin−/− C57BL/6 mice [48] . Although a direct inhibitory effect of IL-10 on T cells has been described [12] , an increased secretion of IL-12p70 by lymph node cells of the T cell-specific and complete IL-10-deficient mice early after infection ( data not shown ) strongly argues for an indirect activation of CD4+ and CD8+ T cells via APC [49] . Similarly , the infection of IL-10−/− mice with the intracellular parasite T . gondii leads to enhanced IL-12 production by APC , resulting in an overwhelming and lethal Th1 immune response [16] . Interestingly , the only difference between the T cell-specific and the complete IL-10-deficient C57BL/6 mice occurred upon antigen restimulation of draining lymph node cells seven days post infection , resulting in an increased secretion of IFN-γ and IL-4 by the lymph node cells of the complete IL-10-deficient mice . As the number of IFN-γ-secreting T cells did not differ between the two mouse strains this could emphasize the importance of the T cell-derived cytokines on disease progression . We can not rule out , however , that the increased secretion of IFN-γ and IL-4 by the IL-10 deficient cells is a result of the in vitro incubation of the complete cell suspensions . At very late stages of infection the differences between individual mouse strains were less prominent than in previous studies [21] , [27] , [29] . These differences may be explained by different substrains of L . major used for the experiments , as this was shown to substantially influence disease progression [37] . Furthermore , the dose of parasites inoculated , the infection site as well as the injection route ( needle challenge vs . sand fly bite ) are known to influence Leishmania disease progression and vaccine efficiency [50] . To further elucidate the role of T cell-derived IL-10 in long-term infection , the conditional IL-10-deficient mice were backcrossed on the naturally susceptible BALB/c background . Intriguingly , T cell-specific IL-10-deficient , like complete IL-10-deficient BALB/c mice [20] , showed a protective phenotype . To the best of our knowledge , we showed for the first time that suppression of the T cell-specific IL-10 production not only improves the outcome of disease in naturally resistant mice , but even reverses progredient disease in susceptible mice into complete parasite control . This healing phenotype was associated with an increased inflammation early after infection , similar to T cell-specific IL-10-deficient C57BL/6 mice . On the BALB/c background , however , the inflammatory response was of a mixed Th1/Th2 type , characterized by an elevated secretion of IFN-γ and IL-4 . Our findings are in accordance with previous reports that antibody-mediated depletion of CD25+ cells and the reconstitution of SCID mice with splenocytes depleted of CD4+ CD25+ T cells leads to an early burst of IL-4 and a slight increase in IFN-γ secretion in the draining lymph nodes of mice on a BALB/c background [51] , [52] . However , this depletion led to disease exacerbation , which was attributed to the augmented early Th2 response , while the T cell-specific IL-10 mutant mice used in the present study were able to control the disease , thus indicating that the general down-regulation of the immune response by T cell-derived IL-10 is more relevant to disease outcome than the balance between a Th1 and Th2 immune response . The macrophage/neutrophil-specific IL-10-deficient mice on a C57BL/6 background never showed any significant difference in disease progression compared to IL-10-competent control mice , neither early after infection nor at any later time point . These findings contradict the model of an autocrine IL-10-based inhibition of macrophages due to the activation of Fcγ-receptors by antigen-antibody complexes as proposed by Mosser and colleagues [20] , [45] . The importance of macrophage derived IL-10 , secreted following IgG-FcγR engagement , is further called into question by the finding that mice lacking IgG1 are more resistant to infection with L . mexicana [53] . The studies by Mosser and colleagues , however , were performed with mice on the naturally susceptible BALB/c background , whereas we could examine only macrophage/neutrophil-specific IL-10 mutant mice on the resistant C57BL/6 background . Surprisingly , at later time points after infection , macrophage/neutrophil-specific IL-10-deficient mice had slightly higher parasite loads than IL-10-competent mice , while complete IL-10-deficient mice had somewhat higher parasite loads than the T cell-specific IL-10-deficient mice on both the C57BL/6 and the BALB/c background , although these differences never reached statistical significance . A cytokine secretion assay 7 days after infection with L . major revealed only a very small number of IL-10-secreting CD4+ cells in the draining lymph nodes . These appeared to be mainly FoxP3− , suggesting activated effector T cells as the primary source for the T cell-derived IL-10 . These results are in accordance with the above-mentioned studies which demonstrated a disease exacerbation in BALB/c mice depleted of CD25+ Treg cells [51] , [52] . In other mouse strains and in studies of human leishmaniasis , however , both CD4+CD25−FoxP3− as well as CD4+CD25+FoxP3+ T cells have been implicated in the long-term control of infection and development of sterile immunity [27] , [28] , [29] , [30] , [39] , [40] . To directly address this question , Treg cell-specific IL-10 mutants , which have recently been generated in our lab [54] could be used in a future study . Apart from T cells and macrophages , a variety of other cell types is able to secrete IL-10 [12] and could be found in the feet and draining lymph nodes of L . major infected mice . Infection of mice with Leishmania parasites is known to induce secretion of IL-10 by neutrophils . Our finding that disease progression is unaltered in the macrophage/neutrophil-specific IL-10 mutant mice , however , is in accordance with data showing that early after infection with L . major , secretion of IL-10 by neutrophils of C57BL/6 mice is higher than by neutrophils of susceptible BALB/c mice [55] . In an experimental model of visceral leishmaniasis , also NK cells were recently shown to produce IL-10 at late stages of infection , thus reducing host resistance [56] . Furthermore , regulatory B cells have been defined , which downregulate immune responses mainly by secretion of IL-10 . Notably , B cells secreting IL-10 have been shown to shape the development of Th2 immune responses and disease course in BALB/c mice infected with L . major [32] . Moreover , the greater susceptibility of CBA/J mice to L . amazonensis compared to L . major was found to coincide with an increased frequency of IL-10-positive splenic B cells [57] . Although these findings do not provide a clear proof , an effect of B cell-derived IL-10 on Leishmania infection cannot be ruled out . Furthermore , DC play a prominent role in orchestrating the induction of immune responses against pathogens and , thus , might contribute to susceptibility to infection by IL-10 secretion . Recently , Owens et al . demonstrated that during chronic L . donovani infection splenic CD11chi DC acquire a regulatory profile with elevated secretion of IL-27p28 and IL-10 . Following therapeutic depletion of all CD11c+ cells , enhanced host resistance and reduced disease pathology could be observed , accompanied by a diminished development of IL-10-producing CD4+ T cells [58] . Therefore , it would be tempting to investigate Leishmania disease progression in DC-specific IL-10-deficient mice . The finding that the early antigen-specific secretion of IL-10 by T cells influences the progression and outcome of leishmaniasis has clinically relevant implications . It clearly shows that an efficient vaccine against leishmaniasis should not only induce IFN-γ- , TNF- and IL-2-secreting memory T cells [41] , but also needs to prevent the development of antigen-specific IL-10-secreting T cells . Along that line , a DC-based vaccination approach developed in our lab significantly reduced the antigen-specific production of IL-10 early after infection . Consistent with our data , the ratio of IFN-γ/IL-10 was a reliable pre-challenge indicator of vaccine success in a heterologous prime-boost vaccination approach with two different antigens that both induced low levels of IL-4 [59] . In addition , recent reports demonstrated that the protection mediated by a non-persistent parasite vaccine was accompanied by a restriction of the early IL-10 production [60] , that IL-10 may account for the lack of efficiency of killed parasite vaccines [61] , and that IL-10 influenced the magnitude and quality of the Th1 response following immunization with leishmanial proteins and CpG [62] . It is not possible , however , to quantitatively distinguish the importance of the early reduced IL-10 secretion from the likewise reduced IL-4 secretion and increased IFN-γ secretion following infection . Taken together , we showed that the early secretion of IL-10 by antigen-specific FoxP3− T cells suppresses the development of an inflammatory response following infection with L . major . Thus , IL-10 secretion by T cells is not only involved in the long-term control of infection and prevention of overwhelming immune activation , but also has a crucial effect on immune activation early after infection , influencing disease outcome and vaccine efficiency .
All mice were kept under specific pathogen-free conditions . Mice experiments were performed in strict accordance with the German Animal Welfare Act 2006 ( TierSchG ) and the animal protocol was approved by the government of Lower Franconia ( permission no . 55 . 2-2531 . 01-16/09 ) . C57BL/6 IL-10fl/fl CD4-Cre+ and IL-10fl/fl LysM-Cre+ mice were generated as previously described [33] , [34] , [63] . C57BL/6 IL-10fl/fl CD4-Cre+ mice were backcrossed onto the BALB/c background for 7 generations . C57BL/6 IL-10fl/fl EIIa-Cre+ mice and BALB/c IL-10fl/fl CMV-Cre+ mice were generated by crossing the respective IL-10fl/fl Cre− animals with C57BL/6 EIIa-Cre+ mice [64] or BALB/c CMV-Cre+ mice [38] ( The Jackson Laboratory , Bar Harbor ) . Age- and sex-matched IL-10-deficient and IL-10fl/fl Cre− mice were used as controls . Female wild-type BALB/c mice were purchased from Charles River Breeding Laboratories ( Sulzfeld , Germany ) and were 6 to 8 weeks old at the onset of the experiments . All experiments were conducted according to the German animal protection law . The cloned virulent L . major isolate ( MHOM/IL/81/FE/BNI ) was maintained by passage in BALB/c mice . Promastigotes were grown in blood agar cultures . For the preparation of parasite lysate , stationary-phase promastigotes were subjected to six cycles of rapid freezing and thawing . Mice were infected intradermally with 2×105 ( BALB/c ) or 2×106 ( C57BL/6 ) stationary-phase L . major promastigotes into the right hind footpad . The course of infection was monitored by measuring the increase in footpad size of the infected versus the non-infected footpad . The frequency of parasitized cells in the popliteal lymph nodes of infected mice was determined by limiting dilution analysis as described elsewhere [65] . In brief , lymph nodes draining the site of infection of individual animals were passed through a 70-µm cell strainer to obtain single cell suspensions and cells were washed in PBS . Serial 1∶4 dilutions were prepared and 4 replicates per dilution and mouse were plated in 96-well blood agar plates and incubated for 10 days at 27°C , 5% CO2 in a humidified atmosphere . The frequency of L . major-infected cells was calculated at a fraction of 37% of negative culture wells . For flow cytometry , popliteal lymph nodes draining the site of infection and infected footpads were dissected and passed through a 70-µm cell strainer . Single cell suspensions were pretreated with anti-CD16/CD32 Fc block ( clone 2 . 4G2; BD Pharmingen , Heidelberg , Germany ) and subsequently stained with fluorochrome-conjugated mAb directed against CD3 ( 145-2C11 ) , CD4 ( RM4-5 ) , CD8 ( 53-6 . 7 ) , CD49b ( DX5 ) , CD45R ( RA3-6B2 ) ( all BD Pharmingen ) , or biotinylated mAb directed against F4/80 ( BM8; Caltag/Invitrogen , Karlsruhe , Germany ) , followed by PE-Cy5-labeled streptavidin conjugate ( BD Pharmingen ) . Intracellular staining for FoxP3 ( clone FJK-16s; eBioscience , San Diego ) was performed according to the manufacturer's instructions . For intracellular cytokine staining , single cell suspensions of lymph nodes were cultured overnight in the absence or presence of Leishmania lysate ( equivalent to 10 parasites/cell ) . Cytokine excretion was blocked by addition of 2 µM monensin ( BD Pharmingen ) for 4 h . Following staining of surface markers , cells were fixed with 4% paraformaldehyde in PBS , permeabilized with 0 . 1% saponin/1% FCS in PBS , and stained with PE-conjugated anti-mouse IFN-γ-mAb ( XMG1 . 2; BD Pharmingen ) . To identify IL-10-secreting cells , a cytokine secretion assay ( Miltenyi Biotec , Bergisch Gladbach , Germany ) was performed according to the manufacturer's instructions . Samples were analyzed on a FACSCalibur flow cytometer ( BD Pharmingen ) , using CellQuestPro software . Popliteal lymph nodes draining the infected footpads were collected and single-cell suspensions ( 1 or 2×106 cells/ml as indicated ) were cultured in the absence or presence of Leishmania lysate ( equivalent to 10 parasites/cell ) for 72 h . Culture supernatants were harvested for determination of the cytokines IL-2 , IL-4 , IL-10 , IL-17 and IFN-γ by sandwich ELISA as described previously [42] . The detection limits were 8 pg/ml for IL-2 , IL-4 , and IL-17 , 16 pg/ml for IL-10 , and 100 pg/ml for IFN-γ . Preparation of bone marrow-derived DC ( BMDC ) and vaccination of mice was performed as described previously with slight modifications [36] , [66] . In brief , freshly prepared bone marrow cells from 6- to 10-week-old female BALB/c mice were cultured in RPMI 1640 medium ( GIBCO Invitrogen ) supplemented with 10% heat-inactivated FCS , 2 mM L-glutamine , 10 mM HEPES buffer , 60 µg/ml penicillin , 20 µg/ml gentamycin and 200 U/ml GM-CSF ( PeproTech , London , UK ) . Fresh medium containing GM-CSF was added to the cultures at days 3 and 6 . After 10 days , non-adherent cells were collected and cultured at a density of 1×106 cells/ml with L . major lysate ( equivalent to 30 parasites/cell ) and the CpG oligodeoxynucleotide 1668 ( 5′-TCCATGACGTTCCTGATGCT-3′ , Qiagen Operon , Cologne , Germany ) at 25 µg/ml for 16 h . Antigen-loaded BMDC were washed twice with PBS , and subjected to four cycles of rapid freezing and thawing . Mice were immunized i . v . with an equivalent of 5×105 BMDC per animal . Control mice were treated with PBS only . One week after vaccination , mice were challenged intradermally with 2×105 stationary-phase L . major promastigotes into the right hind footpad . Cytokine secretion was measured by ELISA as described above . For ease of viewing , the values for cytokine secretion without antigen re-stimulation were subtracted from the values for the antigen-dependent cytokine secretion . Data were analyzed using the GraphPad Prism 4 . 00 software . All cytokine secretion data is shown as the mean of 3–4 independent experiments . In these cases no statistical analyses were performed . Footpad swelling following Leishmania infection as well as parasite frequencies are displayed per individual mouse with 4–12 mice each group . In these cases statistical analyses were performed . For the analyses of differences between more than two independent groups the Kruskal-Wallis-test followed by Dunn's posttest was performed . For comparison of two independent groups the unpaired t-test was used . For comparison of two groups with data from independent experiments the paired t-test was used . Differences were considered significant at p<0 . 05 . | The clinical symptoms caused by infections with Leishmania parasites range from self-healing cutaneous to uncontrolled visceral disease and depend not only on the parasite species but also on the type of the host's immune response . It is estimated that 350 million people worldwide are at risk , with a global incidence of 1–1 . 5 million cases of cutaneous and 500 , 000 cases of visceral leishmaniasis . Murine leishmaniasis is the best-characterized model to elucidate the mechanisms underlying resistance or susceptibility to Leishmania major parasites in vivo . Using T cell-specific and macrophage-specific mutant mice , we demonstrate that abrogating the secretion of the immunosuppressive cytokine IL-10 by T cells is sufficient to render otherwise susceptible mice resistant to an infection with the pathogen . The healing phenotype is accompanied by an elevated specific inflammatory immune response very early after infection . We further show that dendritic cell-based vaccination against leishmaniasis suppresses the early secretion of IL-10 following challenge infection . Thus , our study unravels a molecular mechanism critical for host immune defense , aiding in the development of an effective vaccine against leishmaniasis . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"infectious",
"diseases",
"clinical",
"immunology"
] | 2013 | T Cell-Derived IL-10 Determines Leishmaniasis Disease Outcome and Is Suppressed by a Dendritic Cell Based Vaccine |
Epidemic transitions are an important feature of infectious disease systems . As the transmissibility of a pathogen increases , the dynamics of disease spread shifts from limited stuttering chains of transmission to potentially large scale outbreaks . One proposed method to anticipate this transition are early-warning signals ( EWS ) , summary statistics which undergo characteristic changes as the transition is approached . Although theoretically predicted , their mathematical basis does not take into account the nature of epidemiological data , which are typically aggregated into periodic case reports and subject to reporting error . The viability of EWS for epidemic transitions therefore remains uncertain . Here we demonstrate that most EWS can predict emergence even when calculated from imperfect data . We quantify performance using the area under the curve ( AUC ) statistic , a measure of how well an EWS distinguishes between numerical simulations of an emerging disease and one which is stationary . Values of the AUC statistic are compared across a range of different reporting scenarios . We find that different EWS respond to imperfect data differently . The mean , variance and first differenced variance all perform well unless reporting error is highly overdispersed . The autocorrelation , autocovariance and decay time perform well provided that the aggregation period of the data is larger than the serial interval and reporting error is not highly overdispersed . The coefficient of variation , skewness and kurtosis are found to be unreliable indicators of emergence . Overall , we find that seven of ten EWS considered perform well for most realistic reporting scenarios . We conclude that imperfect epidemiological data is not a barrier to using EWS for many potentially emerging diseases .
There are numerous causative factors linked with disease emergence , including pathogen evolution , ecological change and variation in host demography and behavior [1–5] . Combined , they can make each pathogen’s emergence seem idiosyncratic . In spite of this apparent particularity , there is a recent literature on the possibility of anticipating epidemic transitions using model-independent metrics [6–14] . Referred to as early-warning signals ( EWS ) , these metrics are summary statistics ( e . g . the variance and autocorrelation ) which undergo characteristic changes as the transition is approached . In addition to infectious disease transmission , EWS have been investigated for transitions in a broad range of dynamical systems , including ecosystem collapse and climate change [15–21] . The motivation for EWS comes from the theories of dynamical systems and stochastic processes , in particular the slowing down that universally occurs in the vicinity of dynamical critical points [22–24] . Theoretical results for disease emergence are promising , and suggest that the transition from limited stuttering chains of transmission ( R0 < 1 ) to sustained transmission and outbreaks ( R0 > 1 ) is preceded by detectable EWS [8 , 13 , 14] . A major obstacle to deploying early-warning systems is the type of data available to calculate the EWS . Theoretical predictions assume the data will be sequential recordings ( or “snapshots” ) of the true number of infected in the population through time [8–13] . In this paper we refer to this as snapshots data . However , epidemiological data originate instead from notifications by public health practitioners whenever a new case is identified . Public health bodies aggregate individual cases into regular case reports ( e . g . the US Centers for Disease Control and Prevention’s Morbidity and Mortality Weekly Report ) , as shown in Fig 1 . Different combinations of serial interval ( difference in time of symptom onset between primary and secondary cases ) and aggregation period lead to time series which have very different appearances . Even assuming perfect reporting , variability in both the incubation period and onset of clinical symptoms mean that snapshots data cannot be reconstructed from case report data . In addition to aggregation , case reports are subject to reporting error ( see Fig 2 ) . Underreporting may occur due to asymptomatic infection , poorly implemented notification protocols , or socio-political factors [25–29] . Misdiagnoses and clerical errors in the compilation of reports can result in both under- and over-reporting [30–32] . Due to self reporting and contact tracing , once an index case has been positively identified secondary cases are more likely to be diagnosed , which may lead to clustering in case reports . The combination of case aggregation and reporting error results in a mismatch between snapshots and imperfect epidemiological data . EWS , such as the variance ( Fig 3 , top panel ) , are affected by imperfect data ( Fig 3 , bottom panel ) and may not display the characteristic trends that form the basis for detecting disease emergence . This provides reason to question the direct application of EWS to observed data . In this paper we report on a simulation study aimed at investigating the robustness of a range of EWS to case report data . We simulated a stochastic SIR model of a pathogen emerging via increasing R0 , and corrupted the simulated case reports by applying a negative binomial reporting error . The area under the curve ( AUC ) statistic was computed to quantify how well trends in an EWS identify emergence . We find that performance depends on both the EWS and the reporting model . Broadly , the mean , variance , index of dispersion and first differenced variance perform well . The autocorrelation , autocovariance and decay time perform well unless either i ) the data are highly overdispersed or ii ) the aggregation period is less than the infectious period . The coefficient of variation , kurtosis , and skewness have a more subtle dependence on the reporting model , and are not reliable . We conclude that seven of ten EWS perform well for most realistic reporting scenarios .
The dynamics of disease spread in a host population are modeled as a stochastic process using an SIR model with birth and death [37] . The model compartments and parameters are listed in Table 1 . Transition rates and effects are listed in Table 2 . The basic reproductive number for the SIR model is R0 ( t ) = β ( t ) / ( γ+ α ) , where β ( t ) varies due to nondescript secular trends in the transmissibility . Simulated data are generated using the Gillespie algorithm [33] , which simulates a sequence of transition events ( infection , recovery , birth and death ) , and returns the number of individuals in each model compartment through time . The SIR simulations are of a population with average size N0 = 106 . The parameter ζ gives the rate at which new cases arise due to external sources , and is set to ζ = 1 per week . The death rate , α , is the reciprocal of the life expectancy , set to 70 years . Case counts , Ct , are given by the number of recovery events ( at rate γIt ) within each aggregation period , and are included in the model as an additional variable ( see Table 1 ) . Reporting error is applied to the case count at the end of each aggregation period by sampling a negative binomial distribution , P ( K t = k | C t ) = Γ ( ϕ + k ) k ! Γ ( ϕ ) ( ρ C t ρ C t + ϕ ) k ( ϕ ρ C t + ϕ ) ϕ , ( 1 ) with reporting probability ρ and dispersion parameter ϕ [38] . Given Ct cases , the mean number reported is μt = ρCt . The variance is specified by the dispersion parameter via the relation σ t 2 = μ t + μ t 2 / ϕ . Increasing ϕ reduces the overdispersion of the data , so that for large ϕ the distribution of reports is approximately Poisson . Previous work has proposed a range of different EWS to anticipate dynamical transitions [8 , 12–15 , 17 , 18] . The ten candidate EWS considered in this paper are listed in Table 3 . We consider additional indicators to those most frequently studied in the EWS literature ( the variance , autocorrelation and coefficient of variation ) . As R0 approaches 1 , the mean number of cases caused by introductions rises , making it a potential EWS . The index of dispersion is a similar measure to the coefficient of variation , and is defined as the variance to mean ratio . The decay time ( or correlation time ) is a log-transform of the autocorrelation , which diverges as R0 approaches 1 ( the definition of critical slowing down ) . In addition to the autocorrelation , which is normalized by the variance , we consider the unnormalized autocovariance . As both the autocorrelation and variance increase , the autocovariance may outperform these two measures . Theoretical results show the increase in variance accelerates as R0 approaches 1 , suggesting the first differenced variance as a complementary EWS . Additionally we investigate the performance of two higher-order moments , the skewness and kurtosis . Functional expressions for the dependence of each EWS on R0 can be found using the Birth-Death-Immigration ( BDI ) process , a variation of the SIR model which neglects susceptible depletion ( i . e . St = N0 ) . The BDI process is a one-dimensional stochastic process , depending only on the number of infected It , and possesses an exact mathematical solution ( for full details see [13] ) . This allows expressions for the moments and correlation functions of It to be found ( Table 3 , fourth column ) . BDI theory predicts that most EWS ( the mean , variance , index of dispersion , autocovariance , decay time and first differenced variance ) are expected to grow hyperbolically as R0 approaches one . The autocorrelation is expected to grow exponentially , the kurtosis quadratically and the skewness linearly . The coefficient of variation is the only EWS which does not grow , instead remaining constant . We propose observing these trends in data as a basis for anticipating disease emergence . The numerical estimators used in this paper are listed in Table 3 , third column , discussed in more depth below . Theoretical predictions from the BDI process are based on It and do not take into account effects of reporting error and aggregation . The focus of this paper is to examine the robustness of each EWS to reporting process parameters , using simulated case report data , Kt . BDI theory predicts that 9 out of 10 EWS increase as the transition is approached . We quantify the association of each EWS with time using Kendall’s rank correlation coefficient [19] . A coefficient close to ( +/− ) 1 implies consistent increases/decreases of the EWS in time . As the underlying dynamics of the case reports are stochastic , the value of the rank correlation coefficient is itself a random variable . Multiple simulations of the test ( emerging ) and null ( stationary/not emerging ) scenarios result in two distributions of correlation coefficients for each EWS . We measure performance using the AUC statistic , defined as the overlap of the two distributions , and may be interpreted as the probability that a randomly chosen test coefficient is higher than a randomly chosen null coefficient , AUC = P ( τtest > τnull ) [39 , 40] . The name comes from one method of calculating it , the area under the receiver operating characteristic ( ROC ) curve , a parametric plot of the false positive rate against true positive rate as the decision threshold is varied [41] . Instead of explicitly calculating the ROC curve , the AUC can be efficiently calculated after ranking the combined set of test and null correlation coefficients by value [40] , AUC = [ r test - n test ( n test + 1 ) / 2 ] / ( n test n null ) , ( 2 ) where rtest is the sum of the ranks of test coefficients and ntest and nnull are the number of realizations of the test and null models respectively . In this paper the AUC statistic quantifies how successfully an EWS distinguishes whether or not a disease is approaching an epidemic transition . An AUC = 0 . 5 implies that an observed rank coefficient value conveys no information about whether or not the disease is emerging , i . e . the EWS is ineffective . If the AUC < 0 . 5 then a decreasing trend in the EWS indicates emergence , whereas if AUC > 0 . 5 an increasing trend indicates emergence . A larger |AUC − 0 . 5| implies better performance , if |AUC − 0 . 5| = 0 . 5 the rank coefficient value classifies the two scenarios perfectly . The mathematical definitions of the EWS depend on expectations of the stochastic process , E[f ( X ) ] ( Table 3 , second column ) . To calculate EWS from non-stationary time series data we use centered moving window averages with bandwidth b as estimators for expectation values . For example , the mean at time t is estimated using μ ^ t = ∑ s = t - ( b - 1 ) δ t + ( b - 1 ) δ X s 2 b - 1 , ( 3 ) where δ is the size of one time step . Near the ends of the time series ( t < bδ and t > T − bδ ) , the normalization factor 2b − 1 is reduced to ensure it remains equal to the number of data points within the window . Applying Eq 3 to the time series for X results in a time series for μ ^ . Certain EWS depend on others , for example the variance depends on the mean . EWS are therefore calculated iteratively , for example μ ^ is first calculated using Eq 3 , and then σ ^ 2 is found using σ ^ t 2 = ∑ s = t - ( b - 1 ) δ t + ( b - 1 ) δ ( X s - μ ^ s ) 2 2 b - 1 . ( 4 ) Estimators for each EWS are in Table 3 . For snapshots data Xt = It , and for case report data Xt = Kt . Throughout this paper we use a bandwidth of b = 35 time steps ( weeks or months depending on aggregation period ) . Results have been found to be similar for a bandwidth of b = 100 time steps . To quantify the sensitivity of each EWS to reporting process , we calculate the AUC from simulated data for a range of different model parameter combinations . The experimental design is fully factorial ( i . e . considers all parameter value combinations ) . The following four parameters are varied: ( i ) the infectious period , 1/γ , which can be either 7 or 30 days , ( ii ) the reporting probability , ρ = 2−8x for x in {0 , 0 . 05 , 0 . 1 , … , 1} , ( iii ) the dispersion parameter , ϕ , which is one of {0 . 01 , 1 , 100} , ( iv ) the aggregation period , δ , which is either weekly or monthly . For the test model , the disease emerges over T = 20 years , via an increase R0 . For the null model , R0 is constant . One epidemiological interpretation for the test scenario is it models transmission in a population with high vaccine coverage , where gradual pathogen evolution results in increasing evasion of host immunity . An alternative interpretation is it models zoonotic spillover , where pathogen evolution within an animal reservoir results in gradually increasing human transmissibility [42] . In both interpretations , the null model assumes no change in host-to-host transmissibility . The transmission dynamics were simulated using the Gillespie algorithm [33] . The Gillespie algorithm assumes all model parameters ( including the transmissibility ) are constant . To simulate disease emergence we modify the Gillespie algorithm , discretely increasing β at the end of each day and after each reaction to ensure an approximately linear increase in R0 over T = 20 years , from R0 ( 0 ) = 0 to R0 ( T ) = 1 . For the null model , transmission is simulated for 20 years at a constant rate , R0 = 0 . Our choice of null has no secondary transmission , making the classification problem easy under perfect reporting . This enables clearer identification of responses to reporting process effects as results span the full range of the AUC statistic . We repeated the experiment with null model R0 = 0 . 5 , and found no qualitative differences . For both scenarios transmission is subcritical , with disease presence maintained by reintroduction from an external reservoir . For each parameter combination 1000 replicates of both scenarios are generated . We perform these computational experiments in R using the pomp package [43] to simulate the SIR model and the spaero package [44] to calculate the EWS . Code was written to simulate aggregation and reporting error . All code to reproduce the results is archived online at doi:10 . 5281/zenodo . 1185284 .
Provided the data are aggregated monthly , with high reporting probability and low overdispersion , the coefficient of variation , skewness and kurtosis have similar AUC values when calculated from snapshots data ( Fig 4 ) and case report data ( Fig 5 , right column ) . Unlike the other seven EWS , this it is not the case for weekly data . If calculated from weekly snapshots data with 1/γ = 1 week , the coefficient of variation has an AUC = 0 . 18 ( Fig 4 , bottom right ) . With reporting , if ρ = 1 , ϕ = 100 the AUC = 0 . 005 ( Fig 6 , top right ) . By switching to case report data the performance of the coefficient of variation has improved dramatically . Similar improvements are seen for the skewness and kurtosis . In addition , and perhaps counterintuitively , these three EWS’s performances are further enhanced at lower reporting probabilities ( compare the top right and bottom right panels of Fig 6 ) . At low overdispersion and low reporting probability , the coefficient of variation ( |AUC − 0 . 5| = 0 . 5 ) is joint with the mean and variance as the best performing statistic , closely followed by the skewness ( |AUC − 0 . 5| = 0 . 497 ) and kurtosis ( |AUC − 0 . 5| = 0 . 491 ) . The improvement in performance at low reporting probability is acutely sensitive to other model parameters . Both overdispersion in the reporting ( for example Fig 6 , left column ) and larger aggregation period ( Fig 5 , right column ) severely dampen the sensitivity to ρ . All three EWS perform poorly if ϕ = 0 . 01 , regardless of the other model parameters . This group of EWS are all measures of the correlation between neighboring data points . At high reporting probability ( ρ > 0 . 33 ) and low overdispersion ( ϕ = 100 ) , all three perform well ( AUC > 0 . 77 ) , regardless of infectious and aggregation periods ( see Fig 5 ) . Performance is comparable with snapshots data ( Fig 4 ) . Overall , they perform best if 1/γ = 1 week ( Fig 5 , top row ) and worst if 1/γ > δ ( Fig 5 , bottom left ) . At low overdispersion , decreasing the reporting probability reduces the AUC ( compare the top right and bottom right panels of Fig 6 , AUC = 1 . 000 vs 0 . 831 ) . The performance drop is largest if 1/γ = 1 month and δ = 1 week . The performance of all three EWS is negatively affected by overdispersion . Sensitivity to overdispersion is least for 1/γ = δ = 1 week , performance is only poor if ϕ = 0 . 01 and/or ρ ≲ 0 . 036 ( Fig 5 , top left ) . These three EWS are reliable indicators of emergence provided δ ≥ 1/γ and ϕ = 100 . Unless reporting error is highly overdispersed ( ϕ = 0 . 01 ) , the mean , variance and first differenced variance perform extremely well ( AUC ≈ 1 , see Fig 5 ) . If case reports are aggregated weekly and have high overdispersion ( ϕ = 0 . 01 ) , they are among the best performing EWS . The mean and variance have AUC > 0 . 85 , and the first differenced variance has AUC ≈ 0 . 66 , but is largely unaffected by reporting probability and infectious period . However , if case reports are aggregated monthly and ϕ = 0 . 01 , then all three perform poorly . This holds regardless of reporting probability and infectious period , and is in line with the results for other EWS . The index of dispersion ( unrelated to the dispersion parameter ) has a similar performance to the previous group of EWS , however with certain differences . We first consider low overdispersion ( ϕ = 100 ) . At low reporting probabilities the index of dispersion performs best if 1/γ = 1 week and δ = 1 month ( Fig 5 , top right ) . For other combinations of infectious period and aggregation period , performance suffers a sharp drop as reporting probability decreases . This drop occurs at a reporting probability dependent on the infectious period and aggregation period , around ρ = 0 . 047 for δ = 1 week , and around ρ = 0 . 027 for δ = 1/γ = 1 month . Unique among the EWS , the index of dispersion performs best at intermediate overdispersion ( ϕ = 1 ) , in particular at small reporting probability . This is true for all infectious and aggregation periods , although most pronounced if 1/γ = 1 month and δ = 1 week . For ϕ = 0 . 01 the index of dispersion performs better if the data are aggregated weekly , and best if the infectious period is also one week , with AUC ≈ 0 . 71 for ρ = 0 . 047 ( Fig 6 , bottom left ) . Provided ρ ≳ 0 . 05 and ϕ > 0 . 01 , performance is good for all aggregation and infectious periods . Overall performance is best if 1/γ = 1 week and δ = 1 month . Taken in isolation , the mean and variance are the EWS least impacted by reporting . Unless the overdispersion in the observation process is high ( ϕ = 0 . 01 ) , their performance is largely unaffected by reporting process parameters . At low reporting probabilities they outperform the autocorrelation , autocovariance , decay time and index of dispersion , and are independent of aggregation period and infectious period . EWS sensitive to correlation between neighboring data points perform well unless i ) ϕ = 0 . 01 and/or ii ) 1/γ > δ and ρ ≲ 0 . 06 . While it is clear how high overdispersion in reporting reduces correlation in the data , an explanation for ii ) is less clear . If calculated from snapshots data , the coefficient of variation , kurtosis and skewness are the worst performing statistics ( |AUC − 0 . 5| ≈ 0 ) . Using case report data improves performance under certain conditions . If cases are aggregated weekly with low reporting probability and low overdispersion then they are among the best performing EWS , with |AUC − 0 . 5| ≈ 0 . 5 . In addition the trends of the skewness and kurtosis ( both decreasing ) are opposite those given by the BDI process ( both increasing ) . Overall , we conclude that these three EWS are unreliable indicators of disease emergence as their performance is conditional on a limited range of reporting process parameters .
For mathematical reasons , proposed EWS for disease emergence have assumed access to regular recordings ( “snapshots” ) of the entire infectious population [8–13] . However , epidemiological data are typically aggregated into periodic case reports subject to reporting error . To examine the practical consequences of this mismatch between theory and data , in this paper we calculated EWS from case report data . We performed extensive numerical simulations to determine the sensitivity of each candidate EWS to imperfect data . Case aggregation and reporting error change the statistical properties of the data , and can have subtle effects on an EWS’s performance . We identified four groups of EWS based on their sensitivity to the various reporting process parameters . The performance of one group , consisting of the EWS with either polynomial or no growth with R0 , has a nuanced relationship with the reporting process parameters . We therefore conclude that the coefficient of variation , kurtosis and skewness perform poorly as EWS . In general , the other EWS ( the mean , variance , first differenced variance , index of dispersion , autocorrelation , autocovariance and decay time ) all performed well and are strong candidates for incorporation in monitoring systems intended to provide early warning of disease emergence . Surprisingly , the combination of reporting error and aggregation of data does not always have a detrimental effect on EWS performance . The coefficient of variation , kurtosis and skewness perform best when both reporting probability and overdispersion are low . At first glance this result appears counter-intuitive: as an increasingly large fraction of cases are missed , performance improves . The point to stress here is that by changing the parameters of the reporting process we are systematically changing the statistical properties of the time series . For instance , the BDI process predicts no trend in the coefficient of variation , due to the standard deviation and mean increasing with R0 at an identical rate [13] . With aggregation and reporting error this identity does not necessarily hold , introducing a trend in the coefficient of variation and improving its performance . To fully explain this phenomenon requires an analytical solution for the statistics of Kt , which requires solving the stochastic process including aggregation and reporting error . However , it can be seen to be plausible if we focus only on stochasticity resulting from reporting error . For low overdispersion ( e . g . ϕ = 100 ) , the reporting probability distribution can be approximated by a Poisson distribution with parameter λ = ρCt . Ignoring demographic stochasticity , we replace Ct with E[Ct] = ηδ ( 1 − R0 ) −1 . Both the coefficient of variation and skewness for this distribution are λ−1/2 = { ( 1−R0 ) /ρηδ}1/2 and the kurtosis is λ−1 = ( ρηδ ) −1 ( 1 − R0 ) . These two expressions both decrease as R0 increases from 0 to 1 , consistent with the experimentally observed AUC < 0 . 5 . The improved performance at low ρ is a consequence of the increased stochasticity in reporting outweighing demographic stochasticity . Can these EWS be used to anticipate disease emergence ? If overdispersion and reporting probability are known to be low , then yes . However , it is unlikely that the reporting process is sufficiently understood for an emerging disease . We conclude that these three EWS are unreliable and therefore not good indicators of emergence . There is a similar reason for why the index of dispersion has a peak in performance at intermediate reporting overdispersion . The negative binomial reporting distribution , conditioned on E[Ct] as above , has index of dispersion given by σ2/μ = 1 + ρηδ{ϕ ( 1−R0 ) }−1 . Therefore increasing reporting overdispersion ( i . e increasing ϕ−1 ) amplifies the response of σ2/μ to changes in R0 . This leads to a greater differential , improving performance of the index of dispersion as an EWS . However , increased reporting overdispersion also implies increased volatility of data within a finite sized window , which reduces reliability . These two countervailing factors provide an explanation for the optimal performance at intermediate overdispersion values . In our analysis we considered an SIR model with epidemiologically plausible parameters . The negative binomial distribution is meant to provide a stringent test on EWS performance , and the parameter ranges are conservative ( especially for overdispersion ) . For instance , if there are 10 actual cases in a week , and reporting error is negative binomially distributed with ρ = 0 . 1 and ϕ = 0 . 01 , then the mean number of reported cases is 1 . However , the probability of no cases being reported is P ( K = 0 ) = 0 . 955 whereas the variance in reported cases is σ2 = 101 . The resulting time series is highly volatile , with little similarity in appearance to the underlying time series of actual cases . It is unlikely that case reports for an emerging disease will have such high overdispersion . In addition , for a highly pathogenic emerging disease , such as Middle East respiratory syndrome ( MERS ) or H7N9 avian influenza , the reporting probability is likely much higher than ρ = 1/256 ( the smallest value we studied ) . Nonetheless , one of the encouraging findings of this study is that high reporting is not essential for reliable early warning . Clear trends in the EWS can still be identified , provided there are sufficiently many introductions for cases to be sporadically detected prior to emergence . These dynamics are typical for a reemerging vaccine controllable disease , such as measles , where cases are continually introduced into disease-free regions from endemic regions [45 , 46] . Performance of EWS which depend on correlation between neighboring case reports was found to be contingent on the aggregation period being larger than the serial interval ( equal to the infectious period for the SIR model ) . If this is not the case , there is a smaller probability that successive links in a chain of transmission fall into neighboring case reports . We speculate that this reduces the impact of fluctuations in a particular report on the subsequent report , diminishing their correlation . This effect is exacerbated if the reporting probability is low . A more rigorous explanation requires a full solution to the stochastic process with aggregation and reporting error . For many known pathogens the serial interval is larger than one week , for example measles virus and Bordetella pertussis [47] . For other pathogens it is less than one week , such as SARS coronavirus [48] and influenza virus [47 , 49] . In order for the autocorrelation , autocovariance and decay time to be reliable EWS , our results suggest the data need to be aggregated by periods larger than the serial interval . The performance boost outweighs costs associated with having fewer data points . We expect that these three EWS will work best for pathogens with short serial intervals; for pathogens with extremely long serial intervals ( such as HIV ) reliable use of these EWS is unlikely . The purpose of this study was not to identify the best EWS , but to investigate the robustness of this approach to the reporting process . In order to isolate the effects of incomplete reporting and aggregation error , we ignored parameter uncertainty by fixing epidemiological parameters ( e . g . the infectious period and the introduction rate ) , rather than drawing them from a distribution . As shown in Table 3 , both the mean and the variance scale with the introduction rate , which is a product of the per capita introduction rate and the susceptible population size . On the other hand , the index of dispersion , autocorrelation and decay time are all independent of introduction rate . Uncertainty of important factors , such as the susceptible population size , is a key challenge to anticipating emergence , and these three EWS may outperform the mean and the variance if uncertainty is included . Thus , while the mean and the variance are most robust to imperfect data , they are not necessarily the best EWS . Instead , our results suggest that imperfect data is not a barrier to the use of EWS . One challenge to early-warning stems from the potential suddenness of novel pathogen emergence , for example SARS was unknown prior to the global outbreak in 2002-2003 . For known pathogens , intermittent data availability presents a separate challenge . Mumps was excluded from the US CDC’s MMWR in 2002 following a period of low incidence . Subsequently , there was a series of large outbreaks , notably in 2006 in the Midwest , and mumps was reincluded . Methods such as EWS are contingent on surveillance efforts being maintained . In addition to underlining the importance of disease surveillance , our work suggests ways it can be improved . Case reports sometimes include additional metadata , for example whether all suspected cases are counted or only clinically confirmed cases . The reporting error of case reports with differing case identification criteria is expected to be very different , as has been seen for instance with MERS [50] . This paper shows that EWS depend on the reporting process , and cross-validating EWS calculated from each data stream could improve performance . Provided it is available , how to leverage metadata is a promising avenue for future research into enhancing EWS . These results provide an essential stepping stone from previous theoretically focused works to implementable early-warning systems . Our findings further reinforce the hypothesis that disease emergence is preceded by detectable EWS . While epidemiological factors preclude early-warning for certain pathogens , for example Ebola virus ( estimates of R0 have consistently been greater than one [51] ) and HIV ( see above ) , they do not rule out many others , including reemerging childhood diseases [52] , H7N9 avian influenza virus [53] , and MERS coronavirus [54] . These pathogens all present public health risk , and EWS may be able to play an important role in monitoring for their emergence . | Anticipating disease emergence is a challenging problem , however the public health ramifications are clear . A proposed tool to help meet this challenge are early-warning signals ( EWS ) , summary statistics which undergo characteristic changes before dynamical transitions . While previous theoretical studies are promising , and find that epidemic transitions are preceded by detectable trends in EWS , they do not consider the effects of imperfect data . To address this , we developed a simulation study which assesses how case aggregation and reporting error impact on 10 different EWS’s performance . Case report data were simulated by combining a stochastic SIR transmission model with a model of reporting error . Temporal trends in an EWS were used as a method of distinguishing between an emerging disease ( R0 approaching 1 ) and a stationary disease ( constant R0 ) . We investigated the robustness of EWS to reporting process parameters , namely the aggregation period , reporting probability and overdispersion of reporting error . Seven of ten EWS perform well for realistic reporting scenarios , and are strong candidates for incorporation in disease emergence monitoring systems . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"infectious",
"disease",
"epidemiology",
"pathogens",
"mathematical",
"models",
"epidemiological",
"methods",
"and",
"statistics",
"probability",
"distribution",
"mathematics",
"stochasti... | 2018 | Anticipating epidemic transitions with imperfect data |
Breast cancer mortality is primarily due to metastasis rather than primary tumors , yet relatively little is understood regarding the etiology of metastatic breast cancer . Previously , using a mouse genetics approach , we demonstrated that inherited germline polymorphisms contribute to metastatic disease , and that these single nucleotide polymorphisms ( SNPs ) could be used to predict outcome in breast cancer patients . In this study , a backcross between a highly metastatic ( FVB/NJ ) and low metastatic ( MOLF/EiJ ) mouse strain identified Arntl2 , a gene encoding a circadian rhythm transcription factor , as a metastasis susceptibility gene associated with progression , specifically in estrogen receptor-negative breast cancer patients . Integrated whole genome sequence analysis with DNase hypersensitivity sites reveals SNPs in the predicted promoter of Arntl2 . Using CRISPR/Cas9-mediated substitution of the MOLF promoter , we demonstrate that the SNPs regulate Arntl2 transcription and affect metastatic burden . Finally , analysis of SNPs associated with ARNTL2 expression in human breast cancer patients revealed reproducible associations of ARNTL2 expression quantitative trait loci ( eQTL ) SNPs with disease-free survival , consistent with the mouse studies .
Breast cancer is the most common form of malignancy in women and is the second leading cause of cancer-related death for women in the United States [1] . Mortality for breast cancer , like most solid cancers , is due not to the primary tumor but instead to metastases , the secondary tumors that arise in distant anatomical sites from cells that have disseminated from the original tumor mass . If the tumor remains localized , the five-year survival rate for breast cancer approaches 99% , suggesting that current clinical interventions for localized breast cancer are highly effective . In contrast , for women with distant metastatic disease the survival rate plummets to 26% [1] , emphasizing the need for new approaches to deal with metastatic lesions . Metastatic breast cancer therefore remains a significant public health problem . It has been estimated that approximately 12% of women in the United States will be diagnosed with breast cancer during their lifetime [1] . At present , almost 240 , 000 women are diagnosed with breast cancer annually , and approximately 3 million women are living with breast cancer in the United States [1] . Although only six percent ( N~ 14 , 000 ) of new cases have metastatic involvement at the presentation of breast cancer , approximately 30% of women without evidence of disseminated disease may develop metastatic lesions later in life [2] . As a result , every year in the United States approximately 40 , 000 women die due to their metastatic breast cancer . While earlier detection and application of anti-metastatic adjuvant therapies have contributed to the increased survival of women over the past decades , metastatic disease remains a significant problem due to tumor cell dormancy and resistance of established metastatic lesions to therapy . Major improvements in patients’ long term survival will therefore come in part from better prevention of metastatic disease through enhanced adjuvant therapy and improved targeting of established metastasis . Advances in a variety of genomic tools have greatly enhanced our understanding of primary breast cancer . Next generation sequencing studies have provided detailed understanding of the common mutational events that drive tumorigenesis in breast cancer [3] . In addition , gene expression studies of primary tumors have revealed that the histochemically defined ER+ or ER- classes can be further subdivided based on molecular profiles and that these additional classes have distinct prognostic outcomes [4 , 5] . Targeted therapies exist for the luminal and HER2+ ( human epidermal growth factor receptor 2 ) subtypes of human breast cancers which significantly improve patient outcome by elimination of occult tumor cells through adjuvant therapies , thereby reducing the incidence of metastatic disease . Analysis of genome-wide primary tumor gene expression has also been successfully used to predict prognosis of breast cancer ( ex . [4] ) . This approach may allow for better application of targeted therapeutics to at-risk patients which could improve outcome while reducing unnecessary treatment-associated morbidity for patients with low risk of metastatic disease . Despite these successes , metastatic disease continues to be a major hurdle , particularly for patients with ER- tumors which include both the basal and triple negative ( estrogen- and progesterone receptor-negative , HER2-negative ) subtypes . These tumors have the worst prognosis among all of the breast cancer subtypes with rapid relapse after diagnosis and poor overall survival [6] . At present , no targeted therapeutic agents are available to combat progression of ER- tumors . As a result , most patients with ER- tumors are routinely treated with conventional chemotherapeutics , despite their considerable side effects , in an effort to improve outcome . ER- breast cancer patients may therefore benefit most from in-depth investigations into the etiology of metastatic disease through identification of better targets to reduce disseminated cells and subclinical lesions prior to the development of pathologic metastases . The FVB/NJ-TgN ( MMTV-PyMT ) 634Mul ( hereafter , MMTV-PyMT ) genetically engineered mouse is a highly aggressive , metastatic model of mammary tumors [7] . Expression analysis of this model suggests that it most closely resembles the Luminal subtype of human breast cancer . In earlier studies [8 , 9] , our laboratory has used this model to identify polymorphic genes within the mouse genome that influence metastatic progression [10 , 11] . Studies of several of these genes have indicated that the human orthologs of the mouse metastasis susceptibility genes are significantly associated with progression only in ER+ patients , consistent with the assignment of the MMTV-PyMT as a luminal cancer model [12] . However , our laboratory has recently demonstrated that the genetic background of this animal not only influences metastatic susceptibility but also the gene expression patterns used to assign molecular subtype [13] . In this study , we further extend those results to demonstrate that crosses between the MMTV-PyMT model and the Asian-derived mouse strain MOLF/EiJ generate gene signatures that are prognostic in ER- rather than ER+ breast cancers . Moreover , these analyses led to the identification of the circadian rhythm gene , Arntl2 , as a metastasis susceptibility gene , suggesting that circadian rhythms play an important role not only in the etiology , but also in progression of the most aggressive form of breast cancer .
Outcrosses were previously performed between the highly metastatic MMTV-PyMT model of breast cancer and members of different branches of the mouse phylogenetic tree to identify inbred strains harboring genetic tumor modifiers [14] . The strain MOLF/EiJ was found to have one of the most significant effects on tumorigenesis , extending tumor latency ( Fig 1A ) as well as suppressing tumor growth and development of pulmonary metastatic lesions ( Fig 1B ) . To map the chromosomal regions associated with these phenotypes , an [FVB/NJ x ( MOLF/EiJ x MMTV-PyMT ) ]N2 backcross ( N = 171 ) was generated and genotyped using the Illumina Mouse Medium Density Linkage Panel ( Fig 1C ) . A significant association was observed for all three phenotypes with distal chromosome 6 after genome-wide permutation correction ( Fig 1D ) [15] . Additional suggestive peaks for tumor growth and metastasis were observed on distal and proximal chromosome 10 , respectively . In contrast to previous genetics studies performed with the MMTV-PyMT model , the significant distal chromosome 6 peaks for all three tumor phenotypes were superimposable and peaked at the very distal end of the chromosome . This suggests the possibility of a common modifier for all three phenotypes in the MOLF/EiJ strain , in contrast to earlier studies where the modifiers of the individual tumor phenotypes were present on separate chromosomes . Based on the observation that the majority of single nucleotide polymorphisms associated with disease in humans are non-coding [16] , we performed an eQTL analysis of mammary tumors from the [FVB/NJ x ( MOLF/EiJ x MMTV-PyMT ) ]N2 backcross in order to identify potential candidate genes . Tumors from 134 animals were arrayed on Affymetrix ST 1 . 0 arrays ( GSE48566; [13] ) and screened for genes in the chromosome 6 interval that were associated with distant metastasis-free survival . Twelve genes on chromosome 6 were found to be significantly associated ( p <0 . 001 and FDR <0 . 05 ) with the presence or absence of metastasis in the MOLF/EiJ backcross ( Fig 1E and S1 Table ) . To determine whether these genes play a role in human breast cancer progression and metastasis , the human orthologs were identified and used as a gene signature to stratify breast cancer patients using GOBO . The individual genes within the signature were weighted using the mouse-derived hazard ratios to require similar direction and relative strength in the human patient cohort . As can be observed in Fig 1F , the weighted gene signature was able to discriminate outcome in human patients , consistent with one or more of the genes being associated with progression in both species . Surprisingly , despite the fact that the MMTV-PyMT tumor system is thought to be a model of ER+ tumors [8 , 9] , stratification of the patients by hormone receptor status demonstrated that the ability to discriminate metastasis-free survival was specific for ER- patients . This result is consistent with earlier observations that genetic background can significantly influence subtype assignments that are based on gene expression methods [13] . To begin to dissect the contributions of individual genes within the QTL interval , Arntl2 was selected due to its position at the apex of the QTL peak and it having the most significant association with metastasis ( based on the p-value ) among the candidate genes ( Fig 1E ) . To assess whether Arntl2 expression affects phenotypes related to tumor burden as well as metastasis in vivo , Arntl2 expression was knocked down in the 4T1 mouse mammary cell line using five shRNA constructs against murine Arntl2 ( Fig 2A ) . Since we were unable to validate a reliable antibody against mouse ARNTL2 , we transiently transfected HEK293 cells with each of the five shRNA constructs along with a myc-expressing Arntl2 overexpression plasmid to verify the shRNA efficacy . Similar to the mRNA levels in 4T1 cells , significant reduction in protein ARNTL2 levels were achieved by observation of myc expression ( Fig 2B ) . To test the effect of Arntl2 knockdown on cell phenotypes , in vitro cell proliferation and migration assays were performed . No significant differences were observed in either in vitro assay , indicating that Arntl2 does not alter these in vitro phenotypes ( S1A and S1B Fig ) . To specifically assess metastatic potential in vivo , control- and Arntl2 shRNA-transduced 4T1 cells were injected orthotopically into the fourth mammary fat pad of syngeneic BALB/cJ mice . The mice were assessed 28 days post-injection for primary tumor weight and pulmonary metastasis burden . Knockdown of Arntl2 did not produce a significant change in primary tumor growth ( Fig 2C ) . However , reduction of Arntl2 expression significantly decreased the number of pulmonary metastatic nodules ( Fig 2D ) . Similarly , overexpression of ARNTL2 in 4T1 mouse mammary tumor cells ( S1C and S1D Fig ) increased pulmonary metastases without significantly affecting primary tumor growth ( Fig 2E and 2F ) . Therefore , these in vivo data establish ARNTL2 as a metastasis-specific modifier . Unexpectedly , the metastasis promoting effect of Arntl2 in the 4T1 cells was the opposite of that predicted from the gene expression analysis from tumor tissue ( metastasis protective ) . We therefore performed a number of studies to determine whether the difference between the spontaneous/endogenous tumor and cell line-based tumor studies were due to the contribution of other factors/genes within the chromosome 6 locus or due to cell line-specific artifacts in the orthotopic transplant assays . The cell line-based shRNA and overexpression studies are consistent with Arntl2 functioning as a metastasis promoting factor . However , to rule out the possibility that the prognostic phase inversion seen in the MOLF/EiJ x MMTV-PyMT cross and human patients is not due to artifact introduced by selection for in vitro growth , in vivo validation was performed . Arntl2 knockout ( KO ) mice were obtained from the KOMP ( Knock Out Mouse Project ) repository [17] and were bred to MMTV-PyMT mice to generate Arntl2+/-; MMTV-PyMT+ or Arntl2+/+; MMTV-PyMT+ females . Animals were permitted to age for tumor initiation and progression , and pulmonary lung metastases were enumerated after euthanasia . In agreement with our previous data , suppression of pulmonary metastases was observed in the Arntl2+/-; MMTV-PyMT+ compared to the Arntl2+/+; MMTV-PyMT+ animals ( Fig 3B ) without affecting the primary tumor ( Fig 3A ) . Taken together , the cell line and in vivo animal studies suggest that Arntl2 is a metastasis-promoting inherited susceptibility factor . Furthermore , these studies indicate that the difference between the prognostic effect of Arntl2 alone and that of the entire chromosome 6 locus is due to additional factors acting either additively or epistatically with Arntl2 . Previous efforts in our laboratory have demonstrated that metastasis susceptibility factors can function as tumor autonomous factors [18–21] or by a tumor autonomous mechanism that results in engagement of stromal cells [10] . To test whether Arntl2-mediated metastasis promotion indirectly involves stromal or immune components other than the tumor cells , the Arntl2-/- mice were bred to BALB/cJ animals to make F1 hybrids . This allows for orthotopic injection of BALB/cJ-derived cell lines ( 4T1 ) without an immune rejection . Unmanipulated 4T1 mammary tumor cells were then implanted into the Arntl2+/+ x BALB/cJ or Arntl2+/- x BALB/cJ mice and pulmonary metastases enumerated after tumor growth and progression . No difference was observed between the wildtype or heterozygous knockout mice ( Fig 3C and 3D ) suggesting a tumor-autonomous effect on metastasis progression . While the effect of reduction of ARNTL2 in stromal cells or cells other than the tumor cells cannot be completely disregarded , these results indicate that the major role of Arntl2 in metastasis is due to its effect in the tumor cells themselves . To identify polymorphisms that might be responsible for the difference in Arntl2-mediated metastasis susceptibility , whole genome sequencing from the MOLF/EiJ genome ( Doran et al , Genome Biology , in press ) was performed . No sequence variants that distinguish FVB/NJ and MOLF/EiJ were identified in the Arntl2 coding region . The genomes were then examined to identify SNPs in regulatory regions that might alter Arntl2 expression . DNase hypersensitivity sites ( DHS ) from the 3134 mammary tumor cell line [22] were screened to identify potential gene regulatory regions near Arntl2 . Two hypersensitivity sites were identified approximately 10 and 12 kb upstream of the predicted transcriptional start site for the Arntl2 Refseq transcripts ( S2A Fig ) . Interestingly , no DHS was observed at the Refseq transcriptional start site , suggesting that the two upstream hypersensitivity sites were associated with the Arntl2 promoter rather than an enhancer element . Consistent with this possibility , searches of the Ensembl database identified an EST ( expressed sequence-tag ) that spans the proximal DHS and the first coding exon ( S2A Fig ) . RT-PCR in 4T1 cells confirmed the predicted spliced product , further suggesting that the two DHS are part of the Arntl2 proximal promoter region ( S2B Fig ) . Overlapping these data , ten SNPs were identified within the putative proximal promoter DHS that distinguished FVB/NJ from MOLF/EiJ ( S2A Fig ) . The SNPs were confirmed by direct target sequencing of this region in FVB/NJ and MOLF/EiJ DNA . In addition to the SNPs in the putative proximal promoter DHS , 12 polymorphisms in the 3’UTR of Arntl2 were also identified . While these SNPs could affect miRNA binding and therefore transcript expression , we focused on those that could alter transcription factor and/or chromatin regulator binding . To assess the potential effect of these SNPs on Arntl2 expression , publicly available RNA-seq and SNP data was examined . Brain RNAseq data ( http://csbio . unc . edu/gecco/ ) shows that wild-derived PWK/PhJ carries the same haplotype as MOLF/EiJ , and WSB/EiJ shares 9 of the 10 FVB/NJ SNPs at this location ( Fig 4A ) . The data demonstrated ~30% lower Arntl2 expression by PWK compared to WSB , supporting the hypothesis that these SNPs affect Arntl2 expression ( Fig 4B ) . To assess these SNPs in vivo in our cell lines , CRISPR/Cas9- mediated replacement of the ten SNPs across the putative promoter DHS was performed . The promoter replacement was performed in 6DT1 , a cell line derived from the MMTV-myc FVB-based genetically engineered model , to be certain that any effects observed were not due to a 4T1-specific cell line artifact . Single cell clones were assessed for successful integration of the MOLF region by PCR amplification and subsequent MOLF SNP-specific restriction enzyme digestion ( Fig 4C ) , followed by Sanger sequencing of the region . As predicted by the wild-derived mouse brain RNAseq data ( Fig 4B ) , replacement of the FVB/NJ DHS allele with the MOLF/EiJ allele resulted in a decreased expression of Arntl2 mRNA ( Fig 4D ) . Orthotopic implantation of the CRISPR-substituted cell line also resulted in a suppression of pulmonary metastasis without significant alteration of primary tumor growth , consistent with the shRNA experiments in the 4T1 cell line ( Fig 4E and 4F ) . Taken together , these data indicate that the differential expression of Arntl2 observed between FVB/NJ and MOLF/EiJ is most likely due to polymorphisms in the promoter of Arntl2 that change its expression and modulate metastasis burden . Validation of ARNTL2 as a bone fide metastasis susceptibility gene was performed by association studies in two human ER- breast cancer patient cohorts . Since the effect in mouse was seen at the gene expression level , we performed a human expression quantitative trait locus ( eQTL ) analysis to see whether inherited variation in human ARNTL2 expression was associated with prognosis in accordance with our hypothesis . ARNTL2 eQTL SNPs were selected based on linear regression analyses of all SNPs within breast tissue exhibiting ARNTL2 expression using two large public data sets: GTEx ( Genotype-Tissue Expression , gene expression data from 183 normal breast tissues ) and TCGA ( The Cancer Genome Atlas , gene expression data from 168 ER negative breast cancer tissues ) . Significant eQTL SNPs selected from the GTEx ( P<0 . 05 ) or from TCGA ( P<1e-7 , a more stringent criterion , was applied due to the concern that tumor characteristics may influence gene expression ) were evaluated for their association with disease free-survival ( DFS ) using a Cox regression model . For SNPs in strong LD ( linkage disequilibrium ) with each other ( r2>0 . 8 ) , only one SNP was selected for genotyping using the Sequenom platform in the validation cohort . Covariates adjusted in the analysis were age at diagnosis , PR status , TNM stage , and cancer treatment . Fourteen SNPs were examined for DFS in a cohort of 726 ER- patients included in a genome-wide association study ( GWAS ) and in an independent set of 1032 ER- patients with targeted Sequenom-based genotyping . The minor allele of SNP rs4964008 showed a consistent association with DSF in both cohorts , and the association reached significance when the two studies were combined ( Table 1 ) . Based on the data from GTEx , alternative allele C ( coded on human reference hg19 forward strand ) of SNP rs4964008 was associated with a lower expression level of ARNTL2 with a p-value of 0 . 0098 . This allele was also associated with better survival in the present study , with a HR ( 95% CI ) of 0 . 71 ( 0 . 52–0 . 97 ) and p-value of 0 . 03 . None of the other SNPs examined were significantly associated with DFS . While not conclusive , these results are consistent with the hypothesis that inherited variants that influence ARNTL2 expression are also associated with prognosis in humans , as they are in mice . The in vivo and allograft data demonstrated that Arntl2 is a metastasis modifier in both mouse and human and that polymorphisms in the promoter of Arntl2 contribute to the differential expression observed between FVB/NJ and MOLF/EiJ . We therefore examined whether the promoter polymorphisms resulted in altered binding of transcription factors or chromatin regulators using an in vitro pull-down assay with biotinylated FVB and MOLF promoter probes . Subsequent mass-spectrometry of the proteins associated with the promoter probes identified several chromatin modifier proteins ( S1 File ) . Interestingly , one of the proteins was PARP1 ( poly ( ADP-ribose ) polymerase-1 ) , a chromatin modifier previously shown to enhance transcription via binding to gene promoters [23] . Furthermore , it was recently demonstrated that PARP1 and CTCF are involved in regulating oscillation of circadian genes between active and repressed states [24] . Binding of both proteins simultaneously to actively transcribed chromatin loci results in circadian genes moving closer to the nuclear lamina , resulting in transcriptional inhibition . In addition , bioinformatics analysis of the DHS site/promoter identified a binding site for CTCF ( CCGCGNGGNGGCAG ) , suggesting that the SNPs could disrupt CTCF binding . These data suggested that differential binding of PARP1 and CTCF might contribute to the observed expression differences and subsequent effects of Arntl2 on metastatic disease . To determine whether this mechanism might be contributing to the difference in Arntl2-mediated metastasis efficiency , in vitro immunoprecipitation was repeated for validation . An increased binding of PARP1 to the MOLF promoter compared to the FVB sequence was observed ( Fig 5A ) , as expected from the initial pull-down experiment . Additionally , an increased interaction was also found between the MOLF promoter and CTCF , consistent with the hypothesis that the MOLF Arntl2 promoter might be more efficiently recruited to the transcriptionally repressive nuclear lamina than the FVB allele . To confirm that the differential binding of CTCF and PARP1 to MOLF was not due to uneven loading , we probed for RRP1B since it was shown to bind both promoters with equal affinity in the mass-spectrometry analysis . As shown in Fig 5A , there were no differences in the binding of MOLF and FVB to RRP1B , further demonstrating the stronger interaction of the MOLF promoter to the CTCF/PARP1 complex . Recruitment of the Arntl2 promoter to the transcriptionally repressive nuclear lamina might be expected to result in decreased expression of genes flanking the promoter by formation of a lamina associated domain ( LAD ) . To determine if the expression of the genes flanking Arntl2 are also altered due to chromosomal positioning , we performed qPCR of the flanking Stk38l and Smco2 genes in the CRISPR-engineered cell lines . Smco2 was not detected , suggesting that this gene is not expressed in our cell lines . However , Stk38l mRNA expression decreased in the heterozygous CRISPR 6DT1 cells compared to wildtype ( Fig 5B ) supporting the hypothesis that the chromosomal locus on which these genes lie may be undergoing a change in intra-nuclear positioning that results in transcriptional repression of the genes ( Fig 5C ) .
Metastasis is a highly complex process that requires the completion of a series of sequential steps to enable a tumor cells to successfully colonize a distant organ . These steps include acquisition of invasive and migratory abilities , penetration into the lymphatics and/or vasculature , surviving the shear forces and anchorage independent state during transit and initial arrest at the secondary site , extravasation out of the circulatory system into the secondary site parenchyma , and finally colonization and growth in an ectopic environment [25] . In addition to the plasticity of tumor cell-intrinsic processes required to complete these steps , it has become increasingly clear over the recent years that tumor non-autonomous factors including the immune system and distant tissues such as the bone marrow also play critical roles in the metastatic cascade [26] . As a result , factors that influence many different mechanisms or biological processes can have either positive or negative impacts on the ability of a tumor to generate life-threatening metastatic lesions . One of the factors that is frequently not considered in studies of metastasis is the influence of the genetic background [27] . Polymorphisms within populations are known to be associated with many disease conditions , including cancer , and can affect both cell autonomous and non-autonomous systems . Unlike cancer driver events , which are frequently somatically acquired missense or nonsense mutations that activate or inactivate oncogenes or tumor suppressors , polymorphisms associated with disease are usually non-coding and thought to be associated with variations in gene expression and pathway balance rather than constitutive gene activation or ablation . Disease-associated polymorphisms may therefore mark biological systems that might be exploited by more subtle manipulation than complete or near-complete suppression of activated oncogenes or their downstream effectors . In this study we provide evidence that polymorphisms that affect the expression levels of Arntl2 , a gene in the circadian rhythm pathway , has a significant effect on metastatic progression in ER- breast cancers . Previous studies have demonstrated an increased risk of breast cancer in women with circadian rhythm interruption/polymorphisms [28 , 29] . Some evidence has implicated circadian rhythm gene expression with metastasis-free survival in breast cancer , but the majority of the associations were not independent of other clinical variables [30] . In addition , these results did not address a causal role for circadian rhythm genes or the mechanisms of gene expression variation across the patient samples . Here we directly tested the causal role of Arntl2 using a variety of orthogonal methods . The results obtained are consistent with variation in Arntl2 expression due to polymorphisms in the proximal promoter , although we cannot presently rule out contributions from additional variants in more distant regulatory elements . In addition to direct effects on transcriptional efficiency , differential binding of PARP1 and CTCF to the putative promoter variants suggests that the polymorphisms may play a role in modifying higher order chromatin biology . A recent study demonstrated that CTCF and PARP1-bound regions of chromatin oscillate between the transcriptionally active nuclear interior and the transcriptionally repressive nuclear lamina compartments during the circadian cycle [24] . Polymorphisms that affect the binding of these two chromatin modifiers might therefore affect the efficiency of recruitment to or release of the Arntl2 promoter from the nuclear envelope , which in turn might alter the timing of Arntl2 expression and its downstream target genes . This possibility is particularly intriguing due to the association of other metastasis susceptibility genes with the nuclear envelope ( Sipa1 , Brd4 isoform 2; [31 , 32] ) . Further studies will be required to assess this possibility as well as the role of Arntl2 target genes in breast cancer metastatic progression . In concordance with our findings , a recently published study by Brady et al . also identified Arntl2 to be a metastasis promoter gene [33] . In this manuscript , the authors identify a secretory mechanism that drives metastatic self-sufficiency and enables lung adenocarcinoma cells to colonize and proliferate at the secondary site . Currently it is unclear whether or not this mechanism is also important in breast cancer . Previous analysis of transcription-based prognostic signatures in our lab has indicated that tissue-specific difference in metastasis biomarkers exist , suggesting different molecular mechanisms may be involved . Further investigations into the downstream mechanisms of Arntl2-mediated mammary metastatic progression are required to address this question and will be part of our future research endeavors . Finally , this study highlights the utility of incorporating polymorphism analysis with genetically engineered mouse models in studies of metastatic disease . Genetically engineered mouse models have been , and will continue to be important to our understanding of the etiology of human disease . However , as demonstrated here , incorporation of complex genetic architecture into an individual model enhances the ability of the model to represent human disease . In this case , the introduction of the polymorphic content from the Asian mouse strain MOLF/EiJ enabled the MMTV-PyMT mouse model , thought to be a model of ER+-like luminal tumors [8 , 9] , to at least partially model the biology of ER- breast cancers . While all models of human disease are at some level imperfect , introduction of polymorphism into the MMTV-PyMT and other models may result in a more accurate representation of the biology of disease that is seen in the genetically complex and diverse human population . This study also highlights additional difficulties in interpretation of quantitative trait genetics in cancer biology . Arntl2 was identified as a potential gene of interest based on 1 ) its presence at the apex of a QTL peak that spans the distal third of chromosome 6 , and 2 ) gene expression associations with metastatic disease . The initial identification of Arntl2 suggested that this gene would function as a metastasis-protective factor , based on both the mouse and human gene expression patterns . However , manipulation of Arntl2 as a single factor in both cell lines and knockout animals consistently supported the role of Arntl2 as a metastasis-promoting , rather than protective , factor . The most plausible explanation for this discrepancy is the presence of an additional gene or genes within the interval that interact with Arntl2 , resulting in a prognostic phase inversion . Additional mammary cancer genes exist within the chromosome 6 interval since the locus was also associated with tumor latency and burden , neither of which was significantly affected by Arntl2 . Multiple genes within a QTL interval affecting the same trait has also been observed [34] , including metastasis modifying genes in the MMTV-PyMT model system ( [10] , [35] ) . These factors indicate that careful validation and interpretation is necessary for the dissection of metastasis-associated QTLs to appropriately interpret how genes individually and collectively contribute to the complex metastatic cascade . Overall , this study has important implications regarding the role of the circadian rhythm in cancer progression and provides a potential mechanism to explain the previously suggested increased risk of breast cancers in nightshift workers . Furthermore , this provides the first evidence that transcriptional control elements can be engineered using CRISPR/Cas9 to establish the causative role of SNPs in inherited susceptibility to cancer metastasis . Further studies into the downstream targets of Arntl2 will be needed to identify the exact mechanisms by which Arntl2 modulates breast cancer cell metastasis .
The generation of the FVB/NJ x [MOLF/EiJ x MMTV-PyMT] backcross was performed as previously described ( Cancer Res December 15 , 2001 61; 8866 ) . Genotyping was performed by the Center for Inherited Disease Research ( http://www . cidr . jhmi . edu/ ) . QTL mapping was performed with the R/QTL program using the J/QTL interface [15] . QTL peaks were considered significant if the p value was less than 0 . 05 after 10 , 000 permutations of the data to correct for genome-wide significance . The research described in this study was performed under the Animal Study Protocol LCBG-004 , approved by the NCI Bethesda Animal Use and Care Committee . Animal euthanasia was performed by cervical dislocation after anesthesia by Avertin . The Shanghai Breast Cancer Study and Shanghai Breast Cancer Survival Study were approved by the institutional review boards of Vanderbilt University , the Shanghai Cancer Institute , and the Shanghai Center for Disease Prevention and Control , and written informed consent was obtained from all participants . The Arntl2 knockout mouse strain used for this research project was created from ES cell clone ( Arntl2_F05 ) originally generated by the Wellcome Trust Sanger Institute . ES cells containing a targeted , non-conditional allele ( tm1e ( KOMP ) Wtsi ) for Arntl2 were obtained from the KOMP ( Knock Out Mouse Project ) repository and knock out mice were generated in a C57BL/6 background by SAIC-Frederick ( Frederick , MD ) . Mice were genotyped using cassette-specific primers . Mouse mammary carcinoma cell lines 4T1 and 6DT1 were a generous gift from Dr . Lalage Wakefield ( NCI , Bethesda , MD ) . All cell lines were cultured in Dulbecco’s Modified Eagle Medium ( DMEM ) , supplemented with 10% Fetal Bovine Serum ( FBS ) , 1% Penicillin and Streptomycin ( P/S ) and 1% Glutamate , and maintained in 37°C degrees with 5% CO2 . Short hairpin RNA ( shRNA ) -mediated knockdown and overexpression cells were cultured in the same conditions with an addition of 10ug/ml puromycin and 5ug/ml blasticidin , respectively . 1 . 1 x 106 293T cells were plated in 6 cm dishes 24 hours prior to transfection in P/S-free 10% FBS DMEM media . Cells were transfected with 1ug of shRNA/cDNA and 1ug of viral packaging plasmids ( 250ng pMD2 . G and 750ng psPAX2 ) using 6ul of Xtreme Gene 9 transfection reagent ( Roche ) . After 24 hours of transfection , media was refreshed with 10% DMEM , supplemented with 1% P/S and 1% Glutamine . The following day , virus-containing supernatant was passed through a 45um filter to obtain viral particles , which were then transferred to 100 , 000 4T1/6DT1 cells . 24 hours post-transduction the viral media was removed and fresh 10% DMEM was added . Finally , 48 hours after transduction , the cells were selected with 10ug/ml puromycin- or 5ug/ml blasticidin-containing complete DMEM . One day prior to scratch assay , 25 , 000 cells/well were plated in triplicates in an Essen ImageLock 24-well plate and allowed to grow to confluence . On the day of the assay , cells were treated with 10ug/ml Mitomycin C ( Sigma ) for 3–4 hours to inhibit cell proliferation . Scratch wounds were made using an Essen 4-channel scratch instrument loaded with Eppendorf 10uL micropipette tips and displaced cells and debris removed by washing with Phosphate-buffered saline ( PBS ) three times . The cells were placed into the IncuCyte Kinetic Live Cell Imaging System ( Essen BioScience ) in complete DMEM media and cell motility imaged for 24 hours . Cells were counted and 5000 cells/well were plated in quadruplicates in 24-well cell culture plates ( Corning , Inc . ) and placed into the IncuCyte Kinetic Live Cell Imaging System ( Essen BioScience ) and programmed to image each well at 2-hour intervals . Samples were imaged until they reached 100% confluence . Data analysis was conducted using IncuCyte 2011A software . RNA was isolated cell lines using TriPure ( Roche ) and reverse transcribed using iScript ( Bio-Rad ) . Real-Time PCR was conducted using VeriQuest SYBR Green qPCR Master Mix ( Affymetrix ) . Peptidylprolyl isomerase B ( Ppib ) was used for normalization of expression levels . Expression of mRNA was defined from the threshold cycle , and relative expression levels were calculated using 2- deltaCt after normalization with Ppib . Primer sequences were as follows: Arntl2 Fwd: GTCTTCCCCAGAATCCCTTT; Arntl2 Rev: TTGTCTCTCCGACGCTTTTC Stk38l Fwd: TTCCTATGAGCAACCATACCCG; Stk38l Rev: TCTAGGCCAAGTCTGGTCCTC; Ppib Fwd: GGAGATGGCACAGGAGGAAAGAG; Ppib Rev: TGTGAGCCATTGGTGTCTTTGC Protein lysate from one million cells were extracted on ice using Golden Lysis Buffer ( 10 mM Tris pH 8 . 0 , 400 mM NaCl , 1% Triton X-100 , 10% Glycerol+Complete protease inhibitor cocktail ( Roche ) , phosphatase inhibitor ( Sigma ) ) . Protein concentration was measured using Pierce’s BCA Protein Assay Kit and analyzed on the Versamax spectrophotometer at a wavelength of 560nm . Appropriate volumes containing 20ug of protein lysates combined with NuPage LDS Sample Buffer and NuPage Reducing Agent ( Invitrogen ) were run on 4–12% ( or otherwise indicated ) NuPage Bis-Tris gels in MOPS buffer . Proteins were transferred onto a PVDF membrane ( Millipore ) , blocked in 5% milk ( TBST + dry milk ) for one hour and incubated in the primary antibody ( in 5% milk ) overnight at 4°C . Membranes were washed with 0 . 05% TBST ( TBS + 5% Tween ) and secondary antibody incubations were done at room temperature for one hour . Proteins were visualized using Amersham ECL Prime Western Blotting Detection System and Amersham Hyperfilm ECL ( GE Healthcare ) . The following primary antibodies were used: mouse anti-Actin ( 1:10 , 000; Abcam ) , mouse anti-myc-tag ( 1:1000; Cell Signaling ) . Secondary antibodies goat anti-rabbit ( Santa Cruz ) and goat-anti-mouse ( GE Healthcare ) were used at concentrations of 1:10 , 000 . Single-guided RNA ( sgRNA ) against the Arntl2 regulatory region was designed using the Massachusetts Institute of Technology CRISPR algorithm ( crispr . mit . edu ) . The top sgRNA with the following binding sites was selected: 5’-GGAATCCCCCTCGCGACCGT-3’ . sgRNAs were annealed and ligated into the pSpCas9 ( BB ) -2A-Puro ( PX459 ) ( Addgene; 48139 ) vector ( Nat Protoc . 2013 Nov;8 ( 11 ) :2281–308 . doi: 10 . 1038/nprot . 2013 . 143 ) . To create single CRISPR clones , 6DT1 cells were plated in 10cm dishes and transfected with 5ug sgRNA-containing PX459 vector and 5ug of linearized 900bp promoter region . Cells were transfected with Xtreme Gene 9 and 24 hours after transfection , cells were FACS-sorted by GPF fluorescence . A total of 1000 cells were sorted and single cell clones were manually plated into 96-well plates . Clones were allowed to grow up until they reached confluence and then gDNA was isolated to check for integration of the MOLF region by PCR-amplification with the following primers: Fwd- 5’-TAAGCA-ACGCGT-GGGCTGGCTAGGGCTG-3’ and Rev- 5’- TGCTTA-AGATCT-TACAAGAGAGTTGACAGGTCCAG-3’ . PCR-amplified DNA was purified using QIAquick PCR Purification Kit ( Qiagen ) and digested using MseI at 37°C overnight . Female virgin FVB/nJ or BALB/cJ mice were obtained from Jackson Laboratory at 6–8 weeks of age . Two days prior to in vivo experiments , cells were plated at one million cells/condition into T-75 flasks ( Corning ) in non-selective DMEM . A total of 100 , 000 cells per mouse was injected into the fourth mammary fat pad of FVB/nJ or BALB/cJ mice for 6DT1 or 4T1 cells , respectively . For experimental metastasis assays , 100 , 000 4T1 tumor cells were injected into the tail vein of BALB/cJ mice . The mice were euthanized between 28–30 days post-injection . Primary tumor was resected , weighed and lung metastases counted . The MOLF and FVB promoter regions were amplified using strain-specific gDNA and the following primers: Fwd /5BiosG/GAAGGTCCACACCCTCTTGC and Rev- /5BiosG/CCTGGACTTGGCCATTGGAA . Phusion taq polymerase was used according to the manufacturer’s recommendation with the following PCR conditions: 98°C for 15 sec , 35 cycles of 98°C for 10 sec , 72°C for 1 . 5 min followed by 72°C for 10 min and a final step of holding in 4°C . PCR products were purified and 5ug of DNA was used for each pull-down experiment along with nuclear lysate from 6DT1 cells ( FVB background ) . Nuclear lysate was isolated using Pierce’s NER Extraction kit according to the provided protocol . Following nuclear extraction , 200ug of lysate was incubated with 5ug of biotinylated DNA probes along with 100ul of streptavidin magnetic beads ( Dynabeads M270 , Thermo-Fisher ( former Invitrogen ) ) . The final volume of 500ul was adjusted using NER buffer from the extraction kit . The mixture was placed on a rotator and incubated at room temperature for 1 hour ( [36] ) . The samples were placed on a magnetic stand and washed with ice cold PBS three times followed by one wash with NER buffer . The beads were resuspended in 25ul of 2x SDS buffer , boiled at 95°C for 5 min and the proteins were separated on a 4–12% SDS-Page minigel ( Invitrogen ) . The gel was stained using Pierce’s Silver Stain for Mass Spectrometry according to the manufacturer’s protocol . Protein bands were sent to the ATRF facility at NCI–Frederick and identified via mass spectrometry . For validation of the pulldown , the same conditions were used to bind the probes to the beads . The beads were washed with PBS three times followed by one wash with NER buffer . The proteins were separated on a 7% Tris-acetate SDS-Page minigel ( Invitrogen ) and transferred onto a PVDF membrane at 4°C for 1 hour . The membranes were probed with the following antibodies overnight at 4°C: rabbit-anti-PARP1 ( Santa Cruz; 1:5000 ) , rabbit-anti-RRP1B ( Santa Cruz; K-19: 1:1000 ) and rabbit-anti-CTCF ( Cell Signaling; 1:1000 ) . Secondary antibody goat anti-rabbit ( Santa Cruz ) was used at concentrations of 1:10 , 000 at room temperature for 1 hour . Samples included in the epidemiological study came from participants of the Shanghai Breast Cancer Study ( SBCS ) and Shanghai Breast Cancer Survival Study ( SBCSS ) . Details on the methodology of the parent studies have been described previously [37–39] . Briefly , the SBCS is a population-based , case-control study that recruited 3 , 448 incident breast cancer patients and controls in urban Shanghai between August 1996 and March 1998 and again between April 2002 and February 2005; 90 . 6% of participants provided a blood or exfoliated buccal cell sample [37] . The SBCSS , also conducted in urban Shanghai , recruited 5 , 042 breast cancer patients between March 2002 and April 2006 with 98% of patients providing an exfoliated buccal cell sample [38] . All participants of both studies provided written informed consent before participating in the study and the Institutional Review Boards of all institutes involved approved the study protocols . Medical charts for breast cancer patients were reviewed to verify cancer diagnosis and obtain tumor characteristic ( including estrogen receptor , ER ) and treatment information . Cancer patients have been followed for survival status and breast cancer recurrence through a combination of record linkages with the Shanghai Vital Statistics Registry and in-person surveys . A total of 1 , 758 women with estrogen receptor negative breast cancer and DNA samples from the SBCS and SBCSS were included in the current study . Among them , 726 samples were genotyped using the Affymetrix Genome-Wide Human SNP Array 6 . 0 [39] and imputed according to the 1000 Genomes Project Asian data . Genotype from the remaining 1032 samples were genotyped using the iPLEX Sequenom MassARRAY platform . A total of 137 disease progression events were observed in the GWAS patients and 193 events were observed in the Sequenom genotyped patients . | Estrogen receptor-negative ( ER- ) breast cancer is a highly malignant form of breast cancer with poor prognosis . Like most solid tumors , the mortality associated with ER- breast cancer is due to the development of secondary tumors , or metastases , in vital organs distant from the original breast tumor . ER- breast tumors , particularly those also lacking HER ( human epidermal growth factor receptor ) expression , are particularly deadly because unlike ER+ or HER+ breast cancers , targeted therapies have not yet been developed that can effectively reduce or eliminate tumor cells that have disseminated throughout the patient . Therefore , better understanding of the etiology of metastasis in ER- patients would potentially have a large clinical benefit by providing new targets to eradicate single tumor cells before they develop into metastases . A better understanding of metastasis etiology may also provide methods to prevent the conversion of these disseminated cells into life-threatening lesions . In this study , we demonstrate that a commonly used model of metastatic breast cancer is capable of identifying genes that play a role in the metastatic progression of ER- breast cancers . Furthermore , we identify the circadian rhythm gene , Arntl2 , as a gene associated with inherited susceptibility for the development of metastatic lesions . These studies provide additional information regarding the mechanisms associated with metastasis in ER- breast cancers . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"breast",
"tumors",
"quantitative",
"trait",
"loci",
"cancers",
"and",
"neoplasms",
"basic",
"cancer",
"research",
"oncology",
"chronobiology",
"mammalian",
"genomics",
"epigenetics",
"chromosome",
"6",
"chromatin",
"chromosome",... | 2016 | The Circadian Rhythm Gene Arntl2 Is a Metastasis Susceptibility Gene for Estrogen Receptor-Negative Breast Cancer |
Knockdown resistance ( kdr ) to dichlorodiphenyltrichloroethane ( DDT ) and pyrethroids is known to link amino acid substitutions in the voltage-gated sodium channel ( VGSC ) in Aedes aegypti . Dengue fever primarily transmitted by Ae . aegypti is an annual public health issue in Taiwan . Accordingly , pyrethroid insecticides have been heavily used for decades to control mosquito populations in the summer and autumn . In Taiwan , an Ae . aegypti population with two VGSC mutations , V1016G and D1763Y , was described previously . Aedes aegypti ( G0 ) were collected in Tainan and Kaohsiung in southern Taiwan . The VGSC gene polymorphisms of the kdr mutations and the intron flanked by exons 20 and 21 were verified . The first generation offspring ( G1 ) were used to measure the resistance level to cypermethrin , a pyrethroid insecticide currently used in Taiwan . In addition to V1016G and D1763Y , we describe two new mutations , S989P and F1534C , which have not been reported in Taiwan . Moreover , we also identify two types ( groups A and B ) of introns between exons 20 and 21 . Intriguingly , the kdr mutations S989P , V1016G and D1763Y are strictly located on the haplotype harboring the group A intron , whereas F1534C links to the group B intron . When those data were taken together , we proposed the following six haplotypes for VGSC genes in Taiwan today: ( i ) S989-intron A-V1016-F1534-D1763 , ( ii ) S989-intron A-V1016G-F1534-D1763 , ( iii ) S989P-intron A-V1016G-F1534-D1763 , ( iv ) S989-intron A-V1016G-F1534-D1763Y , ( v ) S989-intron B-V1016-F1534-D1763 and ( vi ) S989-intron B-V1016-F1534C-D1763 . Triple heterozygous mutations of either S989P/V1016G/F1534C or V1016G/F1534C/D1763Y can be found in one single Ae . aegypti mosquito . The proportions of the VGSC mutations were relevant to cypermethrin resistance . Notably , the presence of S989P and V1016G in the population could be a helpful reference to predict the resistance level to cypermethrin . This is the first study to demonstrate the coexistence of four kdr mutations in a population of Ae . aegypti . Four kdr mutations ( S989P , V1016G , F1534C and D1763Y ) and two intron forms ( Group A and B ) were commonly found in local Ae . aegypti populations in Taiwan .
The yellow fever mosquito , Aedes aegypti ( L . ) , is a metamorphic dipteran species capable of spreading chikungunya virus , dengue virus , Rift Valley fever virus , yellow fever virus and Zika virus via feeding on human blood . Its larval and pupal stages are aquatic and rely heavily on anthropogenic containers [1] . This species originated in Africa [2] . Through human activities such as transportation and urbanization in new areas , Ae . aegypti is today found in tropical and subtropical regions throughout the world [3] . In Taiwan , yellow fever mosquito habitats are strictly distributed over the southern area and Penghu ( a group of islands at west side of Taiwan ) , whereas Asian tiger mosquitoes , Aedes albopictus , can be found throughout Taiwan , from sea level to 1 , 760 m [4] . Dengue fever contributes annually as a public health burden in Taiwan . There were 15 , 492 indigenous cases in 2014 , and in 2015 , the case number hit a record high of 43 , 419 . The identity of geographical distribution between Ae . aegypti habitat and most indigenous cases strongly suggests that Ae . aegypti is the primary vector of dengue fever in Taiwan , whereas occasionally rare indigenous cases occurring elsewhere point out a secondary role of Ae . albopictus in the epidemiology of dengue fever [5] . For the control of mosquito-borne diseases , community engagement for habitat management and the use of insecticides are currently used . Undoubtedly , habitat management is a reliable and promising approach to lower mosquito population number with almost no disadvantages . However , the application of insecticides is a quicker method , particularly during the action to deal with imported cases and outbreaks of mosquito-borne diseases . The long-term use of insecticides promotes the development of resistance in mosquitoes . This issue is considered one of the hardest obstacles to mosquito control throughout the world [6 , 7] . Among various insecticides , dichlorodiphenyltrichloroethane ( DDT ) and pyrethroid compounds primarily target voltage-gated sodium channel ( VGSC ) , namely the voltage sensitive sodium channel ( VSSC ) ( reviewed by [8] ) . The resistance of insects to DDT and pyrethroid is linked to knockdown resistance ( kdr ) [8–10] . Several amino acid substitutions in Ae . aegypti VGSC were functionally confirmed to be associated with kdr by using expressed VGSC protein in Xenopus oocytes [11 , 12] . I1011M , V1016G and F1534C were reported to confer increasing resistance to pyrethroid compounds . S989P alone did not alter the resistance level of the recombinant protein [11 , 12] . When coexpressed with V1016G , S989P was shown to enhance V1016G-mediated resistance to deltamethrin [12] . Saavedra-Rodriguez et al . reported that an ~250 bp intron separates codons 1011 ( exon 20 ) and 1016 ( exon 21 ) in the Ae . aegypti VGSC genomic region [13] . Afterward , Martins et al . then described a polymorphism of that intron . Based on their length , they were classified into two groups , A ( 250 bp ) and B ( 234 bp ) [14] . The variance in intron polymorphism could serve as a marker to study the origins of mutations in the VGSC gene . I1011M was found to coexist with the group A intron [14] . Subsequently , V1016I was reported to locate at the haplotype harboring the group A intron [15] . In other studies , the mutation F1534C was found in alleles possessing either the group A or group B intron , but apparently had a strong link to the group A intron [16] . After malaria eradication in Taiwan was certified by the World Health Organization ( WHO ) [17] , dengue fever replaced malaria to become the most serious mosquito-borne disease in Taiwan . There are hundreds to thousands indigenous cases annually in Taiwan . For instance , 200–2000 indigenous cases of dengue fever were reported each year from 2004–2013 according to Taiwan CDC’s surveillance data ( http://www . cdc . gov . tw/rwd/ ) . To reduce the damage caused by dengue fever , pyrethroid insecticides have been used for decades . A surveillance study from 2002–2012 reported that the Ae . aegypti population in Taiwan displayed resistance to various pyrethroid insecticides [18] . In 2009 , Chang et al . reported two VGSC mutations , V1016G and D1763Y ( referred to as V1023G and D1794Y in [19] , respectively ) , from a permethrin resistant strain , originally collected in Kaohsiung in 1990 [19] . In 2013 , a study found that the two mutations were present in Ae . aegypti collected in the field from Tainan and Kaohsiung [20] . During 2014–2015 , Taiwan suffered severe damage from dengue fever , mainly in Tainan and Kaohsiung: 15 , 492 and 43 , 419 indigenous cases were reported in 2014 and 2015 , respectively . The failure in the fight against dengue fever during those two years might be due to an inability to efficiently control the vector , which possibly had developed new kdr mutations in the population . Hence , it became very important to be aware of the current VGSC gene status in Ae . aegypti population in Taiwan . In this study , we investigated the VGSC gene information for Ae . aegypti collected in 2016 in the high-risk areas of Tainan and Kaohsiung . To investigate the VGSC gene , we focused on the two previously reported amino acid sites of kdr mutations , V1016 and D1763 , along with the other kdr sites with functional confirmation , including S989 , I1011 and F1534 . Moreover , we also characterized the polymorphic status of the intron between exon 20 and 21 , in order to further clarify the relationship among those kdr mutations . Bioassays to examine the resistance to cypermethrin , a pyrethroid insecticide currently used in Taiwan , were carried out as well . The links between VGSC gene traits and cypermethrin resistance are also discussed .
We drove along Tainan ( West Central District , South District , Yongkang District and North District ) and Kaohsiung ( Sanmin District , Xiaogang District , Qianzhen District and Fengshan District ) in southern Taiwan ( Fig 1 ) to collect Ae . aegypti larvae and pupae in March and October 2016 . Since dengue fever cases are frequently reported during the summer and autumn in Taiwan , a large amount of pyrethroid insecticides is used during that period . Therefore , we selected March and October for mosquito collection in order to verify the impact of insecticide use . Mosquito collection was carried out in public areas or private residences/lands with residents’/owners’ permission . The mosquito larvae and pupae were identified under the microscope . Species belonging to Ae . aegypti were brought back to the laboratory and reared supplied with sufficient amount of food [yeast extract/pig liver powder; 1:3 ( w/w ) ] daily in an insectary at 20–30°C . Adults were maintained in an acrylic cage ( 30 × 30 × 30 cm; MegaView Science , Taichung , Taiwan ) and were provided with a 10% sucrose solution . Males of the parental generation ( G0 ) were used for VGSC gene sequencing . Eggs were collected and reared to the next generation . Females of the first generation ( G1 ) were collected for insecticide bioassay . To avoid male sperm DNA contamination in female mosquitoes , we selected only males to verify the VGSC information . The sex determination factor in Ae . aegypti is located on the first chromosome [21] . Since the VGSC gene is mapped on the third chromosome [22] , theoretically , the results derived from male DNA should not cause sexual bias . Each mosquito was placed in a 1 . 5 ml Eppendorf tube with 80 μl phosphate-buffered saline and one glass bead ( diameter 2 . 5 mm ) . Samples were homogenized using a TissueLyser ( Qiagen , Hilden , Germany ) while shaking for 30 sec 3 times . The supernatant was processed using a QIAamp DNA Mini Kit ( Qiagen ) according to the procedure supplied with the kit . Finally , the genomic DNA was eluted in 80 μl Tris-EDTA buffer for immediate use and stored at -20 °C . Sequences of primers to detect VGSC point mutations and intron polymorphisms are based on a previous study [23] . Briefly , the 630/614 bp [the length variance depends on the intron polymorphism ( 250/234 bp ) ] segment at domain II was amplified with the primers AaSCF20 ( 5’-GACAATGTGGATCGCTTCCC-3’ ) and AaSCR21 ( 5’-GCAATCTGGCTTGTTAACTTG-3’ ) and then sequenced with either AaSCF3 ( 5’-GTGGAACTTCACCGACTTCA-3’ ) or AaSCR22 ( 5’-TTCACGAACTTGAGCGCGTTG-3’ ) ; the 748 bp segment at domain III was amplified with primers AaSCF7 ( 5’-GAGAACTCGCCGATGAACTT-3’ ) and AaSCR7 ( 5’-GACGACGAAATCGAACAGGT-3’ ) and then sequenced with AaSCR8 ( 5’-TAGCTTTCAGCGGCTTCTTC-3’ ) ; and the 280 bp segment at domain IV was amplified with the primers AlSCF6 ( 5’-TCGAGAAGTACTTCGTGTCG-3’ ) and AlSCR8 ( 5’-AACAGCAGGATCATGCTCTG-3’ ) and then sequenced with AlSCF7 ( 5’-AGGTATCCGAACGTTGCTGT-3’ ) . The locations of the kdr mutations and primers are illustrated in Fig 2A . The polymerase chain reaction was carried out using i-pfu DNA polymerase ( iNtRON Biotechnology , Seongnam , Korea ) at a 55 °C annealing temperature and 1 min elongation time . The amplicons were visualized by ethidium bromide staining after electrophoresis in 2% agarose gels ( Fig 2B ) and sent out for direct sequencing . The sequences were aligned and analyzed using GeneStudio ( http://genestudio . com/ ) . For measuring the resistance to cypermethrin in mosquitoes from different districts , three-to-five-day-old G1 female adult mosquitoes were transferred in web cages ( 25 × 11 × 11 cm , 25 insects per cage ) from acrylic cages . During the transfer , the mosquitoes were constantly supplied with a 10% sucrose solution . For each treatment , eleven cages of females ( 275 mosquitoes ) were required . Ten cages of mosquitoes ( 250 ) were evenly hung in an ~30 m3 room ( eight corners and two at top-middle ) , while one cage ( 25 ) was left behind as the untreated control . Different concentrations of 60 ml diluted cypermethrin solution were sprayed into the room by an ultra-low volume fogger . After cypermethrin treatment , the mosquitoes were kept in the room full of cypermethrin air particles for 30 min and then were moved into a collecting chamber ( BioQuip Products , Inc . , Rancho Dominguez , CA , USA ) supplied with a 10% sucrose solution . The mosquitoes were then held in a growth chamber at 28±2°C and 75±10% relative humidity ( RH ) with the photoperiod of 10:14 ( L:D ) for 24 h . The mortality was calculated according to the criteria whether the mosquito can stand ( with six or less legs ) or not . Finally , the LC99 in each group was calculated with Microsoft Excel based on treated cypermethrin concentrations . The correlation of LC99 to either proportion of kdr mutations or VGSC haplotypes was examined using Pearson’s correlation coefficient model for the estimation of r2 value and Student’s t-test with Microsoft Excel .
We analyzed five VGSC mutation sites , S989P , I1011M/V , V1016G/I , F1534C and D1763Y , of the Ae . aegypti VGSC gene [the amino acid positions are numbered based on the house fly ( Musca domestica ) VGSC protein sequence] . In 157 mosquitoes collected in southern Taiwan , we observed four mutation types , namely S989P ( TCC to CCC ) , V1016G ( GTA to GGA ) , F1534C ( TTC to TGC ) and D1763Y ( GAC to TAC ) . The four amino acid mutations resulted from DNA point mutations . The mutations I1011M/V nor V1016I were not detected ( Fig 3 ) . The four mutations were all found in eight districts from either Tainan or Kaohsiung . The most frequent mutation is V1016G ( 28 . 03% ) and the rarest is D1763Y ( 6 . 69% ) ; the proportions of S989P and F1534C are 17 . 83% and 21 . 97% , respectively ( Table 1 ) . Furthermore , V1016G can occur independently and can also accompany S989P or D1763Y . In contrast , in the presence of S989P or D1763Y , we always found V1016G ( Fig 3 ) . In the mosquitoes carrying F1534C in both alleles ( F1534C/F1534C ) , there is no coexistence of any other mutation ( S989P , V1016G or D1763Y ) ( Fig 3 , ten F1534C/F1534C individuals among 157 mosquitoes ) . In both Tainan and Kaohsiung , the mosquitoes collected in October possessed more mutations of all four mutation types than those collected in March ( Table 1 ) . This phenomenon may be due to the frequent use of pyrethroid insecticides for mosquito control during the summer and autumn in Taiwan . Polymorphism of the intron between exon 20 and 21 of the VGSC gene IIS6 region has been reported in Brazilian Ae . aegypti . Two forms were classified as group A ( 250 bp ) and B ( 234 bp ) based on their length and sequence differences [14] . In the Taiwanese Ae . aegypti population , we detected both forms . The genomic position of this intron inserts between V1015 ( GTA ) and V1016 ( GTA ) . The form belonging to group A is the majority ( 74 . 52% ) in Taiwanese Ae . aegypti ( Table 1 ) . Remarkably , we noticed that there is a link between VGSC intron polymorphism and kdr mutations . In the group A homologous genotype ( A/A ) , we never found F1534C; in the group B homologous genotype ( B/B ) , S989P , V1016G and D1763Y were not found ( Fig 3 ) . This phenomenon presumably suggests that S989P , V1016G and D1763Y link the haplotype harboring the group A intron; F1534C then links the group B intron haplotype . According to our description above , we proposed that there should be six VGSC haplotypes in the Taiwanese Ae . aegypti population ( Fig 4A ) . We identified six VGSC haplotypes from four amino acid positions and polymorphic intron types of two lengths in the genotyped mosquitoes ( Fig 4A ) . Sequencing results were deposited to NCBI GenBank ( accession numbers: MK495869~ MK495876 ) . The six haplotypes were deduced from the four existing homozygotes ( S989-intron A-V1016-F1534-D1763 , S989P-intron A-V1016G-F1534-D1763 , S989-intron A-V1016G-F1534-D1763Y and S989-intron B-V1016-F1534C-D1763 ) and several present heterozygotes ( Fig 3 ) . The haplotype number is the smallest that can compatibly reconstitute all the genotypes observed; thus , we proposed six haplotypes in Fig 4 , based on the six VGSC haplotype-segregating model . According to the component of these six VGSC haplotypes , we speculated that S989P-intron A-V1016G-F1534-D1763 and S989-intron A-V1016G-F1534-D1763Y were derived from S989-intron A-V1016G-F1534-D1763; S989-intron A-V1016G-F1534-D1763 then originated from S989-intron A-V1016-F1534-D1763 . On the other hand , we also proposed that S989-intron B-V1016-F1534C-D1763 was generated from S989-intron B-V1016-F1534-D1763 ( Fig 4B ) . The haplotype harboring group A intron and no mutation ( Table 2 , S-A-V-F-D ) is the majority ( 46 . 50% ) . In both Tainan and Kaohsiung , the proportions of the three haplotypes containing double mutations ( S989P+V1016G or V1016G+D1763Y ) or a single mutation F1534C ( Table 2 , including P-A-G-F-D , S-A-G-F-Y and S-B-V-C-D ) are higher in October than in March . This phenomenon may be due to pyrethroid insecticide use to control mosquitoes during the summer and autumn in southern Taiwan . Curiously , the proportion of the haplotype containing the single mutation V1016G is higher in March than in October , in both Tainan and Kaohsiung ( Table 2 , S-A-G-F-D ) . Mosquitoes caught in October were thought to be more resistant to pyrethroids than those caught in March , owing to the selection by chemical control with pyrethroid insecticides . V1016G plays a critical role in kdr resistance against pyrethroid compounds . The coexistence of S989P was demonstrated to strengthen V1016G-dependent resistance [12] . We speculated that the loss of the V1016G haplotype from March to October possibly resulted from the gain of the two haplotypes harboring V1016G along with either S989P or D1763Y ( Table 2 , P-A-G-F-D and S-A-G-F-Y ) . The data support the hypothesis that the coexistence of S989P enhances V1016G-mediated resistance and implies that D1763Y might play a role in V1016G-dependent resistance to cypermethrin . Cypermethrin is a widely used pyrethroid insecticide in Taiwan . We examined the mosquitoes’ resistance level to cypermethrin . Fig 5A shows cypermethrin LC99 values for mosquitoes collected from eight districts in Tainan and Kaohsiung . Due to the limited number of insects caught , we could not accomplish a bioassay of mosquitoes collected in March while the North District of Tainan [8 ( locations ) x 2 ( months ) —1 ( miss ) = 15 datasets] . In general , the mosquitoes collected in October are more resistant to cypermethrin than those collected in March , except in West Central District of Tainan . The phenomenon could be interpreted by the heavy insecticide use during summer and autumn in Taiwan . Notably , the mosquitoes collected in Kaohsiung are more resistant to cypermethrin than those collected in Tainan . The temporal and spatial differences in insecticide resistance level also reflect the proportions of VGSC mutations and haplotypes in the genotyped mosquitoes ( Fig 3 and Tables 1 and 2 ) . To verify the role of VGSC mutations in cypermethrin resistance in Taiwanese Ae . aegypti populations , we analyzed the relationship between insecticide resistance level and the proportion of these four mutations in each district from March or October by regression analysis . There is a positive correlation between them ( r2 = 0 . 3268 , p < 0 . 05 ) ( Fig 5B ) . We also verified the relation between insecticide resistance level and proportions of each mutation individually . S989P ( r2 = 0 . 4661 , p < 0 . 01 ) and V1016G ( r2 = 0 . 3398 , p < 0 . 05 ) have a higher correlation to cypermethrin LC99 ( Fig 5C and 5D ) , whereas the frequency of mutations at F1534C ( r2 = 0 . 0880 , p = 0 . 2805 ) and D1763Y ( r2 = 0 . 0362 , p = 0 . 4955 ) have lower correlations to the cypermethrin resistance level ( Fig 5E and 5F ) . From another point of view , since we could classify VGSC haplotypes into six categories , we also analyzed the relationship between cypermethrin resistance level and haplotype proportion in different mosquito populations . Not surprisingly , the haplotype with both S989P and V1016G mutations ( S989P-intron A-V1016G-F1534-D1763 ) displays the highest correlation to cypermethrin resistance level with r2 = 0 . 4661 and p < 0 . 01 ( Fig 5I ) , whereas the r2 values of the other three haplotypes carrying mutation ( s ) range from 0 . 0264 to 0 . 0880 , with p values ranging from 0 . 5612 to 0 . 2805 ( Fig 5H , 5J and 5L ) . These results suggested that the presence of the two kdr mutations , S989P and V1016G , may provide a reference to assess cypermethrin resistance level in Taiwanese Ae . aegypti populations .
In Taiwan , two Ae . aegypti VGSC point mutations were reported previously [19 , 20] . In those two articles , the authors described the two mutations as V1023G and D1794Y . The amino acid numbers of both were based on the yellow fever mosquito VGSC protein sequence . When the codons are transformed to the house fly VGSC protein sequence , V1023G and D1794Y refer to V1016G and D1763Y , respectively ( eight corresponding positions of VGSC mutations between yellow fever mosquito and house fly were clearly annotated in [8] ) . In the present study , we detected V1016G and D1763Y as well . Moreover , we identified two new VGSC mutations , S989P and F1534C , in Taiwanese Ae . aegypti populations; however , the causation of these mutations is not clear and requires further in depth research [24] . Our primer sets , following the previous report [23] are also capable of detecting the I1011M/V mutation . In the 157 mosquitoes collected , we found no examples of I1011M/V . The absence of I1011M/V in Taiwanese Ae . aegypti is not a surprising result . I1011M/V substitutions strictly distribute in the Americas [13 , 14 , 25–27] . In other regions , the I1011V mutation was only detected in Thailand and Vietnam [28 , 29] . Chang et al . was the first group to identify Ae . aegypti VGSC mutations in Taiwan [19] . They sequenced the entire coding region of VGSC genomic DNA from a permethrin resistant strain derived from mosquitoes collected in Kaohsiung in 1990 . When compared with two susceptible strains , Bora and NS , the authors found two mutations , V1016G and D1763Y . At that time , V1016G had been identified previously [25] , whereas D1763Y was a novel mutation . The permethrin resistant strain had been maintained in the laboratory under constant selection with permethrin for 35 generations [30] . It is of interest whether the novel mutation originated during the pressure of long term permethrin treatment inside the laboratory . In a later work [20] , Lin et al . surveyed Ae . aegypti collected from Tainan and Kaohsiung in 2008 . From the mosquitoes caught in the field , they indeed could detect D1763Y . Our surveillance of eight districts in Tainan and Kaohsiung demonstrates that D1763Y distributed in all zones as well . Altogether , the data suggest that the mutation D1763Y originated from a wild population in the field but was not a selected product under long-term exposure to permethrin in the laboratory . Among introns of the VGSC gene , the intron between exons 20 and 21 is polymorphic . Based on their length , these various intron types were classified into group A ( 250 bp ) and B ( 234 bp ) [14] . Intriguingly , the intron polymorphism may serve as a marker to track the origins of kdr mutations . I1011M and V1016I were reported to be concurrent with the group A intron in Brazilian Ae . aegypti [14 , 31] . In Africa , F1534C was found to possess a strong link to the group A intron but was rarely coupled with the group B intron [16] . In this paper , we discovered that S989P , V1016G and D1763Y strictly coexist with the group A intron; F1534C was with the group B intron ( Figs 3 and 4A ) . When previous reports [14 , 16 , 31] and our results are taken together , they suggest that the F1534C kdr mutation might originate from multiple historical events , whereas S989P , I1011M , V1016G/I and D1763Y might individually come from one single occurrence . Curiously , the etiology of most Ae . aegypti VGSC mutations being exclusively concurrent with the group A intron remains an interesting puzzle to verify . In our results and previous investigations [14 , 16] , alleles harboring the group A intron are the majority in the population . We speculated that this may be because most kdr mutations are located on alleles with group A intron , since yellow fever mosquitoes in the world are constantly under the pressure of pyrethroid insecticides . The selection force presumably keeps the alleles carrying kdr mutations . In our results , all S989P and D1763Y mutations are concurrent with V1016G ( Figs 3 and 4A ) . The coexistence of S989P and V1016G was reported previously [32 , 33] , as well as D1763Y and V1016G [19] . The combination of the membrane protein of site-directed mutagenesis expressed in Xenopus oocytes and the two electrode voltage clamp technique constructs a platform to examine the function of the VGSC mutation [9] . V1016G was found to reduce the sensitivity of expressed protein to permethrin and deltamethrin [11 , 12] . S989P did not alter the sensitivity of recombination protein to permethrin and deltamethrin [11 , 12] , nor did D1763Y [11] . It is of great interest to examine whether a kdr mutation can assist another mutations resistance to pyrethroid , particularly when a mutation , such as S989P or D1763Y , needs to be concurrent with V1016G rather than alone . When S989P was cointroduced with V1016G , S989P did not change the response of V1016G to permethrin [11 , 12] , but reduced the sensitivity to deltamethrin [12] . D1763Y was coupled with V1016G in the permethrin resistant strain [19] , implying that D1763Y might confer an assistant role to enforce or strengthen V1016G’s resistance to permethrin . Indeed , we observed an increasing proportion of the haplotype harboring V1016G and D1763Y from March to October ( Table 2 ) , strong evidence that D1763Y was involved in V1016G-dependent resistance to cypermethrin . However , the coexpression of D1763Y could not alter V1016G’s resistance to permethrin and deltamethrin [11] . The function of D1763Y in pyrethroid resistance remains to be further investigated . The association between VGSC haplotype and pyrethroid resistance was clearly demonstrated in this study . The haplotype harboring both S989P and V1016G was positively correlated with Pyrethroid resistance ( Fig 5I ) , which is concordance with the study of Kasai et al . [23] . They revealed the reduced susceptibility accompany with the increased frequency of S989P and V1016G after repeated pyrethroid selection in the laboratory . It is not surprisingly to see no other haplotypes had significantly correlation with insecticide resistance alone . However , the VGSC genotype is comprised of 2 haplotypes in one mosquito; the role of VGSC genotypes in pyrethroid resistance deserves further understanding . In our results , in the mosquito groups with stronger resistance to cypermethrin , more S989P and V1016G are present . This phenomenon was supported by both views of the VGSC mutation proportions of either S989P or V1016G ( Fig 5C and 5D ) and the S989P+V1016G haplotype ( Fig 5I ) . Being aware of pests’ resistance level to insecticides will be helpful in pest control strategy . However , the bioassay to probe certain population’s resistance level to insecticides requires numerous live mosquitoes and is usually time-consuming . For the areas that need to use cypermethrin to control Ae . aegypti , our data may propose an alternative method where the proportion of S989P and V1016G in the population perhaps can serve as a reference to estimate the cypermethrin resistance level . The complexity of genetic components allows organisms to survive through various challenges during natural selection . The kdr mutations in the Ae . aegypti VGSC gene play a vital role to help mosquitoes resist the disturbance of pyrethroid molecules targeting the neural VGSC protein [8–10] . The accumulation of kdr mutation types may benefit insect fitness to resist pyrethroid insecticides . After Bregues et al . initially identified VGSC mutations from strains resistant to pyrethroid and DDT [25] , to date at least ten mutations have been reported . In various regions around the world , more than one mutation in certain mosquito populations was widely recorded [8 , 10] . More recently , the coexistence of three mutations ( S989P , V1016G and F1534C ) was reported from Southeast Asia [34–37] . In this paper , we reported that currently there are four kdr mutations , namely S989P , V1016G , F1534C and D1763Y , in Ae . aegypti populations in Taiwan . These four mutations likely would be an obstacle to the control and prevention of diseases transmitted by Ae . aegypti . In summary , the present study is the first article to report the coexistence of four kdr mutations in a population . | VGSC mutations of Aedes aegypti threaten vector control programs and have been brought to attention in dengue endemic areas . Taiwan has suffered dengue outbreaks , which usually begin from an imported case . Pyrethroid insecticides were used to kill infectious females and adults in the surrounding environment of each suspected case . In Taiwan , V1016G and F1763Y mutations of VGSC have been described previously . Here , we further describe two additional amino acid substitutions ( S989P , F1534C ) and two forms of the intron between exon 20 and 21 . We also propose six haplotypes for VGSC genes in Taiwan today . In conclusion , four kdr mutations ( S989P , V1016G , F1534C and D1763Y ) and two intron forms ( Group A and B ) are commonly found in local Ae . aegypti populations in Taiwan . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"geographical",
"locations",
"animals",
"genetic",
"mapping",
"mutation",
"insect",
"vectors",
"infectious",
"diseases",
"genome",
"complexity",
"agrochemicals",
"taiwan",
"aedes",
"aegypti",
"disease",
"vectors",
... | 2019 | Voltage-gated sodium channel intron polymorphism and four mutations comprise six haplotypes in an Aedes aegypti population in Taiwan |
Bone and lung metastases are responsible for the majority of deaths in patients with breast cancer . Following treatment of the primary cancer , emotional and psychosocial factors within this population precipitate time to recurrence and death , however the underlying mechanism ( s ) remain unclear . Using a mouse model of bone metastasis , we provide experimental evidence that activation of the sympathetic nervous system , which is one of many pathophysiological consequences of severe stress and depression , promotes MDA-231 breast cancer cell colonization of bone via a neurohormonal effect on the host bone marrow stroma . We demonstrate that induction of RANKL expression in bone marrow osteoblasts , following β2AR stimulation , increases the migration of metastatic MDA-231 cells in vitro , independently of SDF1-CXCR4 signaling . We also show that the stimulatory effect of endogenous ( chronic stress ) or pharmacologic sympathetic activation on breast cancer bone metastasis in vivo can be blocked with the β-blocker propranolol , and by knockdown of RANK expression in MDA-231 cells . These findings indicate that RANKL promotes breast cancer cell metastasis to bone via its pro-migratory effect on breast cancer cells , independently of its effect on bone turnover . The emerging clinical implication , supported by recent epidemiological studies , is that βAR-blockers and drugs interfering with RANKL signaling , such as Denosumab , could increase patient survival if used as adjuvant therapy to inhibit both the early colonization of bone by metastatic breast cancer cells and the initiation of the “vicious cycle” of bone destruction induced by these cells .
Breast cancer metastasizes to bone , lung , liver , brain , and lymph nodes . Among these metastases , those targeted to bone are preponderant and observed in approximately 70% of breast cancer fatalities [1] . They are predominately osteolytic and responsible for virtually all breast cancer deaths [2] . Currently available treatments are unable to eradicate metastatic cancer [3] and are limited to the treatment of bone symptoms and complicating fractures . There is thus a critical need for identification of therapeutics that curtail the metastatic process . The process of cancer metastasis is multifactorial , influenced by a combination of genes [4] , and dependent upon intrinsic cancer cell characteristics that dictate how cells migrate , survive , and proliferate , as well as on the cellular and cytokine profile of the tissue from which the cells initially egress . This process is also driven by the microenvironment to which metastatic cancer cells ultimately home [5] . The mechanisms underlying the organ-specific nature of bone metastasis are governed by chemoattractants ( e . g . CXCL12/SDF1 ) , attachment molecules ( e . g . ALCAM , annexin II ) , and cytokines regulating cell growth and angiogenesis ( e . g . IL6 and VEGF ) [6] , [7]; however , the conditions and factors that regulate the expression or activity of these critical molecules to affect metastatic cancer cell bone colonization , establishment , tumor growth , metastatic progression , and recurrence remain unclear . Characterization of such is critical , not only to understanding why some patients are more prone than others to bone metastasis or relapse following treatment of the primary cancer , but also for the design of therapeutic interventions to prevent metastasis to distant organs . The bone microenvironment is a dynamic compartment in which bone is continuously remodeled for proper maintenance of skeletal properties and calcium serum levels , and where hematopoiesis takes place . Thus , it is richly vascularized , but also abundantly innervated by sympathetic , sensory , and glutaminergic nerves [8] . Sympathetic neurons are found in the bone marrow and within cortical bone , and it has become clear during the last decade that they significantly affect both the mesenchymal and hematopoietic lineages that constitute the bone marrow . Norepinephrine ( NE ) -releasing sympathetic nerves , activated by brainstem and hypothalamic centers , stimulate the formation of osteoclasts , thus favoring bone resorption [8]–[12] . In addition , sympathetic nerves inhibit osteoblast proliferation and regulate hematopoietic stem cell ( HSC ) proliferation , survival , and trafficking [13] , [14] . The osteoblastic niche and the β2 adrenergic receptor ( β2AR ) appear to be central and necessary mediators of such sympathetic-driven skeletal processes , generating cytokines that play pivotal roles in stimulating osteoclast formation and hematopoietic cell trafficking , including SDF1 [15] , [16] and RANKL ( receptor activator of NFκB ligand ) [17] , [18] . Beyond the demonstration that many aspects of skeletal biology are under the control of neuroendocrine factors , these findings suggest that drugs and emotional or pathophysiological conditions that affect sympathetic outflow may influence bone biology and contribute to bone pathologies . Sympathetic activation is triggered by prolonged or severe emotional stress . It does not affect overall tumor incidence , yet is associated with shorter patient survival and increased recurrence of breast cancer , which suggests that sympathetic activation may contribute to breast cancer metastasis [19] , [20] . This hypothesis is supported by mouse models that reveal a linkage of neuroendocrine signaling to increased tumor vascularization , invasiveness , and metastasis in soft tissues , via a direct effect on ovarian or breast tumor cells [21] , [22] . The effects of sympathetic activation on bone homeostasis , however , suggest that emotional stress could also indirectly control the behavior of metastatic cancer cells by acting on bone marrow stromal cells , particularly β2AR-expressing osteoblasts . The observation that sympathetic activation increases bone remodeling and bone marrow HSC trafficking via cytokines that are also involved in cancer metastasis further lends credence to this hypothesis . Indeed , SDF1 has clearly been implicated in the mechanisms underlying the homing of metastatic cancer cells , including breast and prostate carcinomas , and RANKL is increasingly recognized as a crucial factor for cancer cell motility , in addition to its well-established role in tumor-induced osteolysis [23] . We show here that sympathetic outflow alters the cytokine profile of the bone microenvironment and promotes the incidence of metastatic colonization by breast cancer cells . We identify RANKL , secreted by osteoblasts in response to sympathetic activation , as a stimulatory factor for breast cancer cell migration and bone colonization in vivo and demonstrate that breast cancer metastasis to bone , induced by increased endogenous sympathetic outflow triggered by restraint stress , can be inhibited by the β-blocker propranolol .
Metastasis to bone is a multistep process starting from the egress of metastatic cells from the primary tumor , contingent upon their subsequent survival in the bloodstream , followed by arrest within the bone capillaries , and colonization of the bone marrow microenvironment . To characterize the effects of sympathetic activation on bone metastasis , we utilized an established model of bone metastasis in which osteotropic GFP-tagged MDA-MB-231VU human mammary carcinoma cells selected at Vanderbilt by in vivo passage of a clone from Dr . T . Yoneda ( called MDA-231 cells herein ) were inoculated via intracardiac ( IC ) injection into athymic nude mice [24] . This model is relevant to the late stages of the bone metastasis process , when metastatic cancer cells egress from blood capillaries and reach the bone marrow microenvironment . Chronic immobilization stress ( CIS ) was chosen as a model of endogenous sympathetic activation ( and depression ) in rodents [21] . In this experimental paradigm , mice are submitted to bodily restraint for 2 h a day , 6 d a week , for the duration of the trial . Six weeks of CIS treatment , including 2 wk of CIS treatment prior to tumor cell inoculation , induced a significant 2-fold increase in the number of osteolytic lesions ( and lesion area ) , measured by Faxitron , compared to control ( no CIS ) ( Figure 1A–D ) . Endogenous sympathetic activation also significantly increased bone tumor number ( 2 . 5-fold increase ) , measured by histomorphometry ( Figure S1A ) . In an independent experiment , daily administration of the non-selective β1/β2 adrenergic receptor agonist isoproterenol ( ISO , 3 mg/kg i . p . ) , utilized as a pharmacological surrogate model of sympathetic activation , had a similar stimulatory effect on bone lesion number ( Figure 1E ) , bone lesion area ( Figure 1F ) , and bone tumor number and burden compared to PBS control ( Figure S2B–D ) . The observed increase in bone lesion area induced by CIS and ISO was not surprising considering sympathetic activation is known to promote osteoclastogenesis and bone resorption [8] . The increase in bone lesion and tumor number , on the other hand , supports the hypothesis that sympathetic activation also promotes tumor metastasis and/or growth in bone . Furthermore , the observation that ISO has a similar effect to CIS is evidence that sympathetic activation , rather than activation of the hypothalamic-pituitary-adrenal axis , triggers these effects of CIS on tumor metastasis . An implication of these findings is that blockade of βAR signaling by β-blockers should inhibit the colonization or growth of metastatic breast cancer cells in bone . To address this question , the β-blocker propranolol was used daily , concomitantly with CIS treatment , to block sympathetic activation induced by CIS . Upon propranolol administration , no significant decrease in bone lesion area and number was observed in control ( no CIS ) mice . By contrast , in mice subjected to CIS treatment , propranolol significantly decreased the number of osteolytic lesions , their surface , and the number of bone tumors compared to CIS control mice ( Figure 1A–C and Figure S1A ) . These data thus indicate that endogenous sympathetic activation by chronic stress in mice increases the incidence of breast cancer cell metastasis to bone , and that β-blockers have the potential to inhibit this effect , which has obvious clinical implications . Both osteoblasts and breast cancer cells , including MDA-231 and 4T1-592 cells ( an osteotropic clone derived from 4T1 cells [25] ) , express the β2AR ( Figure 2A ) . Therefore , the stimulatory effect of sympathetic activation on breast cancer cell metastasis to bone could be mediated by a direct effect on cancer cells , or by an indirect effect through host bone cells . To assess a possible direct effect of adrenergic stimulation on cancer growth , we first treated MDA-231 and 4T1-592 tumor cells with ISO ( 1 uM ) , then measured cell number over a 4-d period in vitro . Surprisingly , ISO decreased MDA-231 and 4T1-592 cell growth over time , in a dose-dependent manner ( Figure 2B , C , E , F ) . We then used two distinct in vivo models of tumor growth to further assess the direct effect of β2AR stimulation in breast cancer cells . In athymic nude mice inoculated subcutaneously with 106 MDA-231 cells and treated daily with ISO for 3 wk , tumor volume at end point was not increased but rather decreased compared to the PBS control group ( Figure 2D ) . Furthermore , no increase in tumor growth was observed in wild type ( WT ) BALB/c mice that received mammary fat pad inoculations of 5×103 4T1-592 cells and subsequent ISO treatment for 4 wk ( Figure 2G ) . These results suggest that direct adrenergic stimulation of breast cancer cells inhibits their growth or promotes cell death in extraskeletal sites , and that the increase in bone tumor lesion and number observed in vivo following CIS treatment is not likely due to the promotion of cell proliferation , but rather to an effect of sympathetic nerves on bone colonization . To further understand the mechanism whereby sympathetic activation promotes breast cancer bone metastasis , we administered ISO to nude mice before MDA-231 cell intracardiac inoculation ( “pre”-treatment for 2 wk ) or after MDA-231 cell inoculation ( “post”-treatment for 4 wk ) ( Figure 3A ) . We reasoned that if sympathetic activation promotes cancer cell colonization in bone via an indirect effect on the stroma , treating mice with ISO prior to tumor inoculation ( with no further treatment afterward ) may increase the incidence of tumors in bone and possibly the number of osteolytic lesions , whereas treating mice after tumor inoculation should promote bone resorption and increase the area of osteolytic lesions , but not bone colonization . In tumor bearing bones , ISO treatment for 4 wk post-MDA-231 cell inoculation significantly increased the size , but not the number , of osteolytic lesions , assessed by radiographic analyses ( Figure 3A–D ) . Accordingly , ISO “post”-treatment also increased tumor burden , but not tumor number , as measured by histomorphometry ( Figure 3E–G ) . In contrast , ISO “pre”-treatment increased bone lesion and tumor areas , but also increased bone lesion and tumor numbers . More metastatic tumors and bony lesions were thus formed in the “pre”-ISO treatment group , even though MDA-231 cells were never directly subjected to ISO stimulation , indicating that ISO promotes metastasis to bone via its effect on the bone marrow environment and not the tumor cells themselves . In agreement with studies in euthymic mice , non-tumor-bearing bones from athymic mice treated with ISO for 4 wk displayed a 32% decrease in tibial trabecular bone volume ( BV/TV ) , assessed by 3-D microtomography ( uCT ) ( Figure S2A ) . The surface of TRAP-positive osteoclasts was significantly increased in these mice ( Figure S2B ) , and accordingly , Rankl expression in long bones was increased 17-fold in response to ISO when compared to PBS controls ( Figure S2C ) . Tumor bearing tibiae , on the other hand , displayed a 76% decrease in BV/TV ( Figure S2D ) . Collectively , these results indicate that sympathetic activation alters the bone marrow environment to make it more hospitable for breast cancer cell metastatic colonization , establishment , and growth , thus priming the “vicious” cycle of bone destruction induced by cancer cells , and exacerbating this cycle once tumor burden increases . RANKL is a cytokine well-known for its osteoclastogenic properties and for being a critical mediator of the feed-forward cycle of bone destruction induced by bone metastatic cancer cells . It is also expressed by normal mammary gland epithelial cells , contributing not only to the development of the lactating mammary gland during pregnancy [26] but also to cancer cell migration in melanoma [27] . In agreement with other reports [27] , [28] , we found that parental non-osteotropic breast carcinoma-derived ATCC MDA-231 cells express RANK , the receptor for RANKL , but at a lower level than bone metastatic MDA-231 cells ( Figure S3A ) . In contrast , RANK expression was not detected in MCF-7 cells . These results , coupled with the observation that ISO strongly increases the expression of Rankl in bones—to a greater extent than in any other organs tested ( Figure 4A ) —led us to explore whether this cytokine contributes to the effect of sympathetic activation on breast cancer metastasis to bone . First , as observed in bone in vivo , ISO significantly increased Rankl expression in primary bone marrow stromal cells cultured in vitro ( BMSCs , Figure 4B ) as well as in MC3T3 osteoblasts ( Figure S3B ) , which suggests that among bone marrow cells , the osteoblast lineage represents a main target . Second , endogenous sympathetic activation by CIS , like ISO treatment , significantly increased Rankl expression in bone 2 h post-stimulation in mice that have been subjected to CIS for 2 wk , indicating that the chosen CIS regimen does not lead to desensitization , at least within the early critical period of bone colonization focused on in this study ( Figure 4C ) . Third , BMSC expression of Sdf1 , a major cytokine promoting breast cancer cell migration , was unaffected by ISO treatment for 2 , 6 , or 24 h ( Figure S3C ) ; in breast cancer cells treated with ISO , the expression of RANK and CXCR4—the receptors for RANKL and SDF1 , respectively—remained unchanged as well ( Figure S3D , E ) . To functionally demonstrate the role of RANKL in the migration of metastatic breast cancer cells , transwell migration assays were performed . When MDA-231 cells were plated in the transwell filter without osteoblasts in the bottom chamber , ISO treatment did not increase their migratory properties ( unpublished data ) . Contrastingly , in co-culture transwell assays , primary BMSCs plated in the bottom chamber increased the migration of MDA-231 cells plated on the transwell filter , and ISO treatment significantly exacerbated this effect ( Figure 4D , E ) . ISO treatment of osteoblasts , prior to the addition of propranolol-treated MDA-231 cells to the transwell filters ( in order to block β2AR signaling specifically in cancer cells ) , did not inhibit cell migration ( Figure 4E ) , signifying that the effects of ISO on the migration of MDA-231 cells are mediated by β2AR stimulation in osteoblasts . Addition of recombinant OPG , a soluble decoy receptor for RANKL , blocked the effect of ISO on MDA-231 cell migration in this transwell co-culture assay , indicating that RANKL is the main cytokine involved in ISO-mediated stimulation of MDA-231 cell migration toward BMSCs ( Figure 4F ) . Similar results were obtained by co-culture of MDA231 cells with MC3T3 osteoblasts ( unpublished data ) . Inhibition of SDF1-CXCR4 signaling by AMD3100 did not block the ISO-induced increase in MDA-231 cell migration , but further reduced migration when used in combination with OPG ( Figure 4G ) . These data are in agreement with the observation that pharmacological blockade of the SDF1-CXCR4 axis by AMD3100 , a CXCR4 antagonist , does not fully prevent bone metastasis [27] . Lastly , recombinant soluble RANKL ( rRANKL ) dose-dependently stimulated the migration of MDA-231 cells ( Figure 4H ) , and consistent with the above observations , high RANK-expressing MDA-231 cells , but not low RANK-expressing parental ATCC MDA-231 cells , responded to rRANKL with a significant increase in migration ( Figures S3A and S4A ) . Recombinant soluble RANKL did not affect MDA-231 cell proliferation ( Figure S3F ) . In sum , these results demonstrate that β2AR stimulation in osteoblasts in vitro promotes breast cancer cell migration via RANKL and via an SDF1-independent mechanism . To address whether the pro-migratory activity of RANKL observed in vitro contributes to the stimulatory effect of sympathetic activation in breast cancer metastasis to bone in vivo , we used a loss-of-function strategy to specifically reduce RANK expression in metastatic MDA-231 cells . The advantage of this strategy , compared to using a RANKL blocker like OPG , was that it allowed us to assess whether sympathetic activation promotes breast cancer bone colonization via the pro-migratory effect of RANKL on metastatic cancer cells or via an indirect stimulatory effect on bone turnover , since sympathetic activation increases bone remodeling , potentially increasing the expression , activity , and/or availability of other cell- or ECM-derived cytokines promoting cancer cell bone colonization , establishment , and growth . A shRNA knockdown approach was used to generate stable clones of MDA-231 cells expressing reduced levels of RANK . We selected a clone whose RANK expression was decreased by 85% ( RANKlow ) compared to scramble shRNA control ( RANKscramble ) ( Figure 5A ) . No difference in cell proliferation or PTHrP expression was detected between MDA-231 RANKscramble control cells and RANKlow cells , treated or not with rRANKL ( Figure S4B , C ) . In contrast , RANK knockdown significantly reduced MDA-231 cell migration in response to rRANKL in a transwell assay ( Figure 5B ) , demonstrating the necessity of a functional RANKL-RANK signaling pathway for the migratory properties of MDA-231 cells . We then inoculated MDA-231 RANKscramble control cells or RANKlow cells intracardially to nude mice treated daily with ISO for 6 wk to mimic sympathetic activation . Confirming the first set of results ( Figure 1 ) , when control MDA-231 RANKscramble cells were used , ISO treatment significantly increased the incidence of GFP-positive tumors in bone , the number and area of bone lesions measured by Faxitron , and bone tumor burden measured by ex vivo GFP imaging , when compared to PBS-treated mice ( Figure 5C–G ) . In contrast , selective inhibition of RANK expression in MDA-231 RANKlow cells blunted the effect of ISO on each of these parameters . No significant difference was observed between the two clones in absence of ISO treatment . These results demonstrate that RANK expression in breast cancer MDA-231 cells , independently of increased bone turnover , is required for their migratory response toward RANKL-expressing bone cells in vivo in response to sympathetic activation .
Although the genetic and phenotypic make-up of a tumor determines its metastatic efficiency , a receptive microenvironment is a prerequisite for tumor colonization , establishment , and growth in distant sites . In the case of breast cancer , the interaction of cancer cells with the bone microenvironment is crucial for their preferential colonization of bone , as well as their subsequent survival , growth , and osteolysis-promoting activity [23] . Characterization of the conditions and factors that transform the bone microenvironment to a state more favorable for cancer cell colonization , dormancy escape , and growth is thus of great interest . In this study , we show that activation of the sympathetic nervous system , a hallmark of severe stress and depression , promotes breast cancer cell colonization in bone . We demonstrate that this neuronal effect on bone metastasis is mediated via the β2AR in bone-forming cells of the host bone marrow environment , and not by a direct effect on metastatic cancer cells , and furthermore that RANKL , whose expression is induced in osteoblasts by sympathetic activation , mediates this effect in vitro and in vivo via its pro-migratory activity . Additionally and of clinical importance , we show that RANKL signaling and the β-blocker propranolol can inhibit the stimulatory effect of endogenous sympathetic activation on breast cancer bone metastasis . Sympathetic nerves releasing NE are present within bone in the vicinity of bone cells , and the β2AR is broadly expressed in bone cells of the mesenchymal , monocytic , or immune lineages , as well as in several cancer cell lines , including mammary carcinomas [27] . Additionally , epinephrine released into the circulation following stress is also an agonist for the β2AR . Sympathetic activation can thus directly stimulate the β2AR in metastatic cancer cells to promote their survival during anoikis , their colonization of bone , and their growth following bone establishment . Such direct stimulatory mechanisms have been reported for ovarian [29] and prostate [30] cancer cell lines . A stimulatory effect of chronic stress has been observed as well on the in vivo growth of parental ( non-osteotropic ) MDA-MB-231 breast cancer cells implanted orthotopically in the mammary fat pads ( MFP ) [21] . In contrast , however , β2AR stimulation of the bone metastatic MDA-231 clone , which is derived from MDA-MB-231 cells following in vivo selection of bone osteotropic cells [31] , reduced tumor growth when cells were implanted subcutaneously . Catecholamines may thus promote or restrain breast cancer cell proliferation depending on either the site of cell growth or the genetic make-up of each cancer clone . On the other hand , the results of our study indicate that β2AR activation in host stromal osteoblasts predominantly accounts for the stimulatory effect of sympathetic activation on MDA-231 breast cancer cell bone colonization . This is supported by the observation that cancer cell inoculation in mice pre-treated for 2 wk with ISO can increase the number of bone tumors and lesions . In that experimental setting , cancer cells are not directly subjected to ISO stimulation; therefore the stimulatory effect of ISO on breast cancer cell bone metastasis must occur via stimulation of the β2AR in host stromal cells , and not via a direct effect on breast cancer cells . It is further reinforced by the observation that selective deletion of the β2AR in osteoblasts recapitulates the high bone mass induced by global β2AR deletion [32] , suggesting that β2AR signaling in osteoblasts , and not vascular cells or immune cells for instance , contributes to the regulation of bone remodeling and of the bone marrow cytokine profile . It is also of note that the increase in RANKL expression induced by ISO was more pronounced in the MC3T3 osteoblastic cell line when compared to BMSCs , suggesting that among the adherent stromal cells constituting the bone marrow , osteoblasts represent the main target for the effect of sympathetic nerves on RANKL expression and breast cancer cell bone colonization . Of interest is that osteocytes also secrete RANKL [33] and thus might also represent a critical bone mesenchymal cell population associated with the aforementioned sympathetic effect . Lastly , the absence of effect of recombinant RANKL on MDA-231 cell proliferation reinforces the notion that the release of RANKL by bone marrow osteoblasts promotes the colonization or retention of metastatic cancer cells , and not tumor growth . It thus appears that sympathetic activation has significant effects on both the host bone marrow stroma and metastatic cancer cells , and that the direct effect of β2AR stimulation on cancer cells depends on the intrinsic characteristics of these cells and on their tissue localization . The observation that the inhibitory effect of ISO on MDA-231 cancer cell proliferation was prevented by coculture with bone marrow osteoblasts ( and not fibroblasts ) further supports the importance of the tumor microenvironment on the growth of various metastatic cancer cell clones ( Figure S5 ) . Stress and severe depression activate both sympathetic outflow and the HPA axis , and thus the potential role of HPA activation in the behavior of metastatic cancer cells in CIS-treated mice cannot be excluded . The observation that bone metastasis induced by CIS could be replicated by βAR stimulation and be blocked by a selective βAR antagonist like propranol strongly suggests that the contribution of adrenergic signaling is very significant and possibly predominant in this process . It is possible that the stimulatory action of glucocorticoids on β2AR signaling in osteoblasts [34] could in fact augment adrenergic signaling in this cell lineage to promote bone metastasis following CIS-induced sympathetic and HPA axis co-activation . This putative mechanism may have important implications for the effects of exogenous glucocorticoids given in conjunction to chemotherapy to cancer patients . The use of adrenergic agonists in our experimental strategy raised the possibility that sympathetic activation could influence breast cancer bone colonization by a stimulatory effect on blood flow or angiogenesis , as observed in other types of cancers [35]–[37] , which would lead to higher likelihood of metastatic cell engraftment . The fact that mice were taken off treatment ( ISO or CIS ) 24 h before intracardiac injection of cancer cells and that cardiovascular parameters returned to normal by 2–3 h following ISO injection in nude mice , as measured by Visen ultrasound studies ( Figure S6A–F ) , exclude , however , a contribution of blood flow on this effect of sympathetic activation on tumor metastasis to bone . Immune cells may be involved as well since they respond to sympathetic signals , but the demonstration of an effect of sympathetic activation on breast cancer bone metastasis in immunodeficient nude mice precludes any contribution of T cells to this model . Lastly , the evidence that sympathetic activation promotes breast cancer cell metastasis to bone via a host-mediated mechanism suggests that such pathophysiological conditions influence other types of osteotropic solid tumors , which warrants further research . RANKL is well known for its osteoclastogenic properties during typical bone turnover . The importance of RANKL in promoting tumor growth in bone and osteolysis induced by metastatic breast cancer cells is also well established [38] and has led to the use of RANKL blockade to prevent fracture in patients with breast and prostate cancer [39] . Over the last few years , it has become increasingly recognized that this cytokine plays a broader role in the process of cancer cell metastasis [40] . The protective effect of RANKL blockade by recombinant OPG on melanoma bone metastasis reported by Jones and collaborators was the first to suggest that RANKL , by enhancing the migration of B16F10 melanoma cancer cells , was implicated in promoting metastatic bone colonization and establishment . In our study , the finding that RANK knockdown in metastatic breast cancer cells reduces bone metastasis indicates that this signaling axis is used by metastatic breast cancer cells as well for their bone metastatic potential , independently of the increased bone turnover induced by sympathetic activation . These findings differ from the Jones study in that this effect was only observed in mice subjected to β2AR stimulation ( and high levels of RANKL in bone ) , and not in non-challenged mice . Regardless of the pathophysiological factor ( s ) increasing its expression or activity , our findings indicate that RANKL is one of the important “soil” factors that promote the colonization of bone by breast cancer cells . Whether RANKL promotes the recruitment of circulating metastatic breast cancer cells from the circulation into bone , or the retention of these cells within the bone marrow environment following egress , remains to be determined . The links encompassing severe depression , high sympathetic tone , increased RANKL expression in bone , and higher incidence of bone metastasis reported in our study , as well as the poor prognosis of breast cancer patients with high tumor RANK expression [41] , provide an explanation for the observed correlation between emotional stress and reduced survival of patients with breast cancers [42]–[44] , which will obviously have to be confirmed by further clinical or epidemiological studies . Because PTHrP and other cytokines converge on the RANKL pathway once tumor load increases , the effect of sympathetic activation on RANKL expression by host bone marrow osteoblasts is likely to be most relevant to the early phases of bone colonization by metastatic cancer cells and to the initial steps of the osteolytic cycle of bone destruction induced by metastatic cancer cells , when bone tumor burden is still low . The derived therapeutic implication is that drugs targeting RANKL , including Denosumab ( which is currently only approved for the prevention of bone fracture in patients with breast and prostate cancer ) , could be efficacious not only to treat the bone symptoms associated with bone cancer metastasis , but also to prevent or limit the development of the disease itself , by reducing breast cancer bone metastasis and/or the activation of dormant breast cancer cells , if administered as adjuvant therapy . The observation that blockade of RANKL in prostate cancer patients increases metastasis-free survival according to a recently published clinical trial supports this prediction [45] . The use of β-blockers may have similar effects to RANKL blockade and could represent an alternative to the current standard of care , with a possible milder effect on bone turnover and a proven safety profile . This is compelling when considering preventative treatments and thus long-term use , and when taking into account treatment cost . Importantly , three recent studies [42]–[44] reported a beneficial effect of β-blocker drug therapy on secondary breast cancer formation and patient survival . These clinical studies support the contribution of sympathetic signaling to the bone metastatic process , and the use of β-blockers as possible adjuvant therapy for breast cancer patients , especially in combination with chemotherapy [46] . Prospective clinical trials will be needed to ascertain the efficacy of β-blockers or RANKL blockers to increase survival in breast cancer patients . If successful , this clinical translation may impact the treatment of millions of women word-wide with an alternative cost-effective treatment .
All procedures were approved by the Institutional Animal Care and Use Committee at Vanderbilt University Medical Center . Mice were group housed in plastic cages ( n = 5/cage ) under standard laboratory conditions with a 12-h dark , 12-h light cycle , a constant temperature of 20°C , and humidity of 48% . Mice were fed a standard rodent diet ( Pharma Serv , Purina Rodent Laboratory Chow 5001; Framingham , MA ) . Nude mice were housed in sterile conditions and fed autoclaved standard chow . Isoproterenol ( ISO ) treatment was given as daily intraperitoneal injections ( 3 mg/kg in 100 uL sterile PBS ) . Control mice were not given injections . Propranolol groups received propranolol ad libitum ( 0 . 5 g/L ) via drinking water . Chronic Immobilization Stress ( CIS ) was carried out by placing mice in 50 mL laboratory conical tubes , perforated for adequate air supply , for 2 h daily . MDA-231VU used in this study were derived from MDA-231SA , a highly bone metastatic GFP-tagged clone developed by Dr . T . Yoneda by in vivo passage via intracardiac injection and subsequent culture of tibial metastases . MDA-231VU cells were then FACS sorted for GFP to enrich for GFP-positive cells . MDA-231VU cells were cultured in 10% FBS DMEM with 1% penicillin streptomycin . Cells were trypsinized at 70%–90% confluence , rinsed , and re-suspended in cold PBS at 10^7 cells/mL . Athymic nude Foxn1nu female mice aged 4–6 wk were anesthetized and injected in the left cardiac ventricle with 100 uL of cell suspension ( 106 MDA-231VU cells ) . Bone metastasis was assessed weekly for 4 wk with Maestro in vivo fluorescence and Faxitron radiographic imaging . In order to prevent cardiac and blood flow complications , all treatments were stopped at least 12 h prior to the intracardiac injection procedure . The 4–6-wk-old Foxn1nu BALB/c ( nude ) mice were inoculated subcutaneously with 100 uL of PBS solution containing 1×106 MDA-MB-231 at the dorsal midline between the scapulae . Measurements of tumor dimensions were made every 2–3 d with calipers or via Maestro GFP imaging . For 4T1-592 , 50 uL of a PBS solution containing 5×103 cells were injected into the 4th mammary fat pad . Tumor size was assessed longitudinally with caliper measurements or volume and post-mortem by weighing on a balance . Osteolytic lesions were quantified bilaterally in the humeri , femora , and tibiae at end point from Faxitron images . Presence of tumor within the bones was confirmed with GFP imaging or by histology . Lesion area was calculated as the average of total osteolytic area per mouse ( sum of all six bones counted ) . Lesion number was calculated as total number of long bones with a visible lesion per mouse ( e . g . 0/6 , 1/6 , etc . ) . Data were double-blinded and calculated by at least two independent researchers . Tibiae from each animal were dissected , cleaned , and fixed for 48 h in 10% formalin/PBS , transferred to 70% EtOH , then loaded into 12 . 3-mm-diameter scanning tubes , and imaged ( μCT 40; Scanco Medical , Bassersdorf , Switzerland ) . The scans were integrated into three-dimensional ( 3-D ) voxel images . A Gaussian filter ( sigma = 0 . 8 , support = 1 ) was used to reduce signal noise , and a threshold of 300 was applied to all analyzed scans . Scans were done at 12 µm resolution ( E = 55 kVp , I = 145 µA ) . Two hundred transverse slices of the proximal tibia were taken from the growth plate and extended distally . All trabecular measurements were made by manual determination of appropriate slices to exclude growth plate , and automated contouring using voxel counting and sphere-filling distance transformation indices . Following uCT scanning , bones were decalcified in 20% EDTA pH7 . 4 at room temperature for 3–4 d . Decalcified samples were then dehydrated and embedded in paraffin . A modified H&E-phloxine-Orange G stain was used to quantify tumor burden and tumor number in 5 um paraffin sections . Metastatic tumor foci were identified by morphology and pink staining in contrast to blue bone marrow cells . Tumor number was counted as number of long bones present with a tumor . Tumor burden was quantified using the Bioquant system imaging software ( Nashville , Tennessee ) as total tumor area per combined area of the six long bones averaged from three sections per bone . Osteoclasts were visualized and counted following Tartrate Resistant Acid Phosphatase ( TRAP ) staining using a standard protocol and the Bioquant system . MDA-231VU cells were detached with trypsin and resuspended in 10% FBS for 1 h prior to assay . Cells were plated in serum-free DMEM in the top well of a 96-well Boyden chamber apparatus ( Neuroprobe ) and allowed to migrate through a semipermeable ( 8 uM pore size ) membrane towards 2 . 5% FBS DMEM in the bottom chamber for 4–6 h . Co-culture migration assays were conducted with two separate transwell systems: a 24-well format using Corning 8 um pore transwell inserts and a 96-well Boyden chamber ( Neuroprobe ) . Primary BMSCs or MC3T3 osteoblasts were grown to confluence in 24- or 96-well tissue culture plates . 24 h before migration , fresh 2 . 5% FBS DMEM containing 10 uM isoproterenol or PBS was added to the cells . On the day of migration , semipermeable membranes were added to the plates , on top of which cancer cells were plated in serum-free media . In both types of assays , unmigrated cells were removed with wet kimwipe from the top of the membrane and were then subsequently fixed with 10% PBS buffered formalin and stained with crystal violet . Total cell area per well was then quantified with an area scan on a 96-well plate reader ( Synergy2 , Biotek ) . MDA-231VU and 4T1-592 cell growth was assessed in Phenol-free DMEM containing 2 . 5% serum , in a 96-well format . Medium was changed daily with fresh ISO or PBS . Cell number was quantified daily by GFP signal measurements ( MDA-231VU cells ) or crystal violet staining and OD592 reading after fixation ( 4T1-592 ) , using a Synergy2 plate reader ( Biotek ) . Growth curves were normalized against day 0 . Each treatment contained eight replicates . Dose-response was calculated based on cell number after 4 d of growth . For coculture growth assays , BMSCs , MC3T3 , or NIH3T3 cells were grown to confluence in 96-well plates in 10% FBS DMEM . 2 . 5×103 MDA-231VU cells were then plated onto these cells in Phenol-free DMEM containing 2 . 5% serum , which was changed daily along with treatment . Total GFP signal per well was quantified with a Syngery2 plate reader . Total counts at each time point were reported as raw numbers or normalized against day 0 reading as indicated . For all gene expression assays , total RNA was extracted from tissues or cells using TriZOL . Tissue RNA extraction was performed following tissue snap-freezing in liquid N2 and power generation using a N2-chilled mortar and pestle , prior to homogenization in TriZOL . RNA quality and quantity were then checked by spectrophotometer and 28S/18S band integrity on a 2% formaldehyde/agarose denaturing gel . cDNA was generated using the High Capacity Reverse Transcriptase Kit ( Applied Biosystems #438814 ) . Real-time PCR was performed using TaqMan gene expression assays on a BioRad CFX96 Real Time System . Taqman probes/primers were from Applied Biosystems [Adrβ2 , Mm02524224_s1; CXCR4 , Hs00237052_m1; Cxcr4 Mm01292123_m1; Cxcl12 ( Sdf1 ) , Mm00445553_m1; RANK ( TNFRSF1 ) , Hs00187192_m1; Rankl , Mm00441908_m1; 18s RNA ( DQ ) MIX probe dye: FAM-MGB , 4352655; Hprt1 , Mm00446968_m1] . Results were analyzed using standard curve quantification or ddCt methods . RT-PCR for βAR1-3 expression was performed as described previously [18] . MDA-231 ( 1×106 ) cells were transfected with HuSH-29 ( Origene , Rockville , MD ) shRNA silencing vectors against human RANK ( TNFRSF11A ) 5′GGAAAGCACTCACAGCTAATT3′ or scramble control: 5′GGAATCTCATTCGATGCATAC3′ , which are placed behind a U6 promoter . Transfections were carried out with the shRNA Nucleofector Kit V ( Lonza , Walkersville , MD ) and program X-013 per manufacturer's instructions . Selection with 1 µg/mL puromycin was begun following overnight recovery . Individual colonies were isolated and maintained with 1 µg/mL puromycin . QPCR was performed as described above to evaluate RANK knockdown efficiency and verify permanent , stable knockdown over ( >5 ) multiple passages . Cardiac ultrasounds were performed on athymic nude mice before , during , and after IP injection of ISO under isoflurane anesthesia ( 3% induction 1 . 5% maintenance ) . Mice were positioned dorsally on a heated manipulation table and measurements taken with Visualsonics Vevo 770 Ultrasound . Three diameteric measurements of the ascending aorta , closest to the semilunar valve , were obtained in short parasternal axis view , b mode . Three measurements of heart rate were obtained in m mode , using distance between diastolic peaks in the left ventricle . Detailed hemodynamic data were collected at baseline and 24 h after stimulation with ISO ( 3 mg/kg ) . Heart rate measurements were repeated at −5 min pre-injection time 0 , 1 min post-injection , and then 2 , 5 , 8 , 10 , 20 , 30 , 60 , and 240 min post-injection . All data are presented as means ± SEM . Statistical analyses were performed using one-way ANOVA for multiple comparisons and two-tailed Student's t tests , either paired or unpaired with Welch's correction for two-group comparisons . For all analyses , p<0 . 05 was considered significant . | Improved detection programs and better drugs to eradicate breast tumors have increased survival in women with breast cancer . However , pain and metastasis to distant organs , including bone , remain significant clinical problems . Understanding why and how metastatic cancer cells colonize specific organs is therefore critical if we are to further improve morbidity and mortality for these patients . Using a mouse model of breast cancer bone metastasis , we present evidence that activation of sympathetic nerves , which is typical in chronic stress or depression , promotes the colonization and establishment of metastatic cancer cells within the bone marrow , leading to an increase in bone osteolytic lesions . We show that this effect is mediated via a β-adrenergic receptor-dependent response of the host bone marrow stroma to catecholamines , that are released upon sympathetic activation , and via the pro-migratory activity of RANKL , a cytokine that is well known to promote bone resorption . Of importance clinically , blocking sympathetic activation with a β-blocker , or blocking RANKL signaling in cancer cells , inhibited the stimulatory effect of sympathetic activation on bone metastasis in this mouse model . Stress-induced sympathetic activation may thus explain , at least in part , the reduced survival rate of breast cancer patients experiencing severe depression . The data also support the use of β-blockers or RANKL blockade as possible adjuvant therapy for women with breast cancer . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"cancer",
"treatment",
"anatomy",
"and",
"physiology",
"oncology",
"bone",
"signaling",
"pathways",
"adrenergic",
"signal",
"transduction",
"musculoskeletal",
"system",
"biology",
"molecular",
"biology",
"autonomic",
"nervous",
"system",
"signal",
"transductio... | 2012 | Stimulation of Host Bone Marrow Stromal Cells by Sympathetic Nerves Promotes Breast Cancer Bone Metastasis in Mice |
Therapeutics with novel modes of action and a low risk of generating resistance are urgently needed to combat drug-resistant Plasmodium falciparum malaria . Here , we report that the peptide vinyl sulfones WLL-vs ( WLL ) and WLW-vs ( WLW ) , highly selective covalent inhibitors of the P . falciparum proteasome , potently eliminate genetically diverse parasites , including K13-mutant , artemisinin-resistant lines , and are particularly active against ring-stage parasites . Selection studies reveal that parasites do not readily acquire resistance to WLL or WLW and that mutations in the β2 , β5 or β6 subunits of the 20S proteasome core particle or in components of the 19S proteasome regulatory particle yield only <five-fold decreases in parasite susceptibility . This result compares favorably against previously published non-covalent inhibitors of the Plasmodium proteasome that can select for resistant parasites with >hundred-fold decreases in susceptibility . We observed no cross-resistance between WLL and WLW . Moreover , most mutations that conferred a modest loss of parasite susceptibility to one inhibitor significantly increased sensitivity to the other . These inhibitors potently synergized multiple chemically diverse classes of antimalarial agents , implicating a shared disruption of proteostasis in their modes of action . These results underscore the potential of targeting the Plasmodium proteasome with covalent small molecule inhibitors as a means of combating multidrug-resistant malaria .
Plasmodium falciparum malaria threatens 40% of the world’s population , resulting in an estimated 220 million cases annually . Of the ~435 , 000 annual malaria deaths worldwide , the majority occur in African children below the age of five [1] . The treatment of P . falciparum malaria is vitally dependent on artemisinin ( ART ) derivatives , exceptionally fast-acting antimalarial endoperoxides that were adopted globally nearly two decades ago as the core components of ART-based combination therapies ( ACTs ) [2 , 3] . The rapid sweep across Asia of parasites that display slow rates of clearance following treatment with the ART derivative artesunate or with ACTs ( referred to herein as ART-resistant or ART-R parasites ) has created a significant need for new treatments that can combat resistance [4–6] . Genomic , clinical epidemiologic , and genetic studies provide compelling evidence that ART resistance is mediated primarily by mutations in the P . falciparum K13 protein [4 , 7–9] . K13 is a member of the BTB-Kelch family that can mediate interactions between certain E3 ubiquitin ligase complexes and substrates targeted for degradation by the ubiquitin-proteasome system ( UPS ) [10 , 11] . In this protein family , the upstream BTB domain typically binds the E3 ligase complex , which transfers ubiquitin moieties to the substrate protein , while the C-terminal six-bladed β-propeller Kelch domain binds the substrate itself , conferring specificity . While the function of P . falciparum K13 is uncharacterized , its Kelch domain harbors single point mutations that are associated with ART resistance , including the C580Y mutation that is predominant in Southeast ( SE ) Asia and the R539T mutation that confers high-level resistance in vitro [7–9 , 12 , 13] . Several experimental studies support a connection between ART , K13 and the UPS . Transcriptional profiling of ART-R SE Asian field isolates earlier revealed an upregulation of components of the UPS in K13 mutant isolates , including several proteasome subunits [14] . Additionally , several genes associated with protein folding and trafficking to or from the ER were upregulated , including subunits of two putative chaperone complexes—the Plasmodium reactive oxidative stress complex ( PROSC ) and the TCP-1 ring complex ( TRiC ) . These results suggest that ART induces widespread protein damage , activating cell stress and proteostasis response pathways , and that ART-R K13 mutant parasites may possess an intrinsic ability to combat drug-induced alkylation via the repair or degradation of damaged proteins and other biomolecules . In support of this hypothesis , a K13 mutant Cambodian P . falciparum isolate ( PL7 ) was reported to show lower amounts of ubiquitinated proteins following exposure to a brief pulse of ART as compared to a K13 wild-type ( WT ) Cambodian isolate ( PL2 ) [15] . However , those data were obtained from non-isogenic field isolates [14 , 15] , and thus the differential responses could be attributable to K13 , or to other genetic differences . The ART metabolite DHA was also recently reported to disrupt P . falciparum proteasome-mediated protein degradation , in addition to generating a backlog of damaged proteins , thereby overwhelming the UPS with substrates . Treating parasites with a translation inhibitor or with an inhibitor of E1 ubiquitin-activating enzymes protected cells against the effects of DHA , which was attributed to the generation of fewer UPS substrates [16] . These observations have highlighted the proteasome as a novel and promising drug target for combatting ART-R P . falciparum infections . This multi-subunit 26S complex consists of a 20S core catalytic subunit capped by 19S regulatory complexes [17] . In eukaryotes , the proteasome contributes to diverse cellular processes ranging from cell cycle progression to apoptosis via its tightly regulated degradation of ubiquitin-tagged substrates [18] . Initial studies of the Plasmodium proteasome revealed that the irreversible inhibitor lactacystin blocked sporozoite development into exoerythrocytic forms and inhibited P . falciparum asexual blood stage cell cycle progression [19] . The proteasome inhibitor epoxomicin was also shown to be potent against transmissible P . falciparum stage V gametocytes and to block oocyst development within the mosquito midgut [20] . Synergy between epoxomicin and DHA was reported in P . falciparum , as was in vivo synergy between the related epoxyketone carfilzomib and DHA in the rodent parasite P . berghei [15] . These previous-generation inhibitors were not viable as antimalarial therapeutics , however , due to high levels of toxicity resulting from inhibition of the host proteasome . Recent studies with a variety of scaffolds have sought to improve selectivity for the P . falciparum proteasome [21–25] . Compounds resulting from these efforts include the covalent peptide vinyl sulfone inhibitors WLL-vs and WLW-vs ( referred to herein as WLL and WLW ) , which are highly selective for the parasite proteasome over the human enzyme [21] . WLL also effectively cleared a Plasmodium chabaudi rodent malaria parasite infection without significant toxicity to the host [21] . These compounds exploit the parasite’s preference for bulky aromatic substrates in various positions of the β2 and β5 subunit active sites of the proteasome [21 , 26] . WLL and WLW potently inhibit the β2 and β5 subunits or the β2 subunit alone , respectively . WLW was shown to have strong activity against the ART-R PL7 isolate and the ART-sensitive ( ART-S ) PL2 isolate , and showed synergy with DHA against PL7 [21] . PL7 was ~two-fold more sensitive than PL2 , suggesting a possible impact of the K13 genotype . Given the threat of multidrug-resistant P . falciparum malaria and the recognized need to delineate the risk for parasite resistance to preclinical antimalarial candidates , we have interrogated mechanisms of resistance in the peptide vinyl sulfone inhibitors WLL and WLW in both ART-S and ART-R parasites . Mutations conferring low-grade resistance were characterized through activity-based profiling of the proteasome beta subunits and molecular modeling based on the known cryo-electron microscopy-based structure of the P . falciparum 20S proteasome [21] . We also screened for antimalarial agents that could overcome existing antimalarial resistance mechanisms when combined with WLL or WLW . Our results , including the identification of a unique stage-specificity profile for these two proteasome inhibitors , highlight the promising features of this class of compounds .
To examine whether the potencies of the Plasmodium-specific proteasome inhibitors WLL and WLW were impacted by mutations in K13 , we tested these two compounds against sets of isogenic P . falciparum lines that express WT or mutant forms of this gene . Assays included the Cam3 . II ( Cambodia ) parental line that expresses the K13 R539T variant and that was culture-adapted in 2010 , as well as the V1/S ( Vietnam ) parental line that is K13 WT and that was culture-adapted in the 1970s , prior to the use of ACTs . These parental lines had previously been edited using zinc-finger nucleases to express the K13 WT or C580Y alleles in Cam3 . II parasites , and K13 WT , R539T or C580Y in V1/S parasites ( S1 Table; [9] ) . These lines are referred to herein as Cam3 . II K13R539T ( the unedited parental line ) , Cam3 . II K13WT , Cam3 . II K13C580Y , V1/S K13WT , V1/S K13R539T , and V1/S K13C580Y . Dose-response 72 hr assays with these six lines , tested as asynchronous cultures , revealed mean half-maximal inhibitory concentrations ( IC50 values ) in the range of 11–13 nM for the β2+β5 inhibitor WLL and 29–59 nM for the β2 inhibitor WLW ( Fig 1A and 1B and S2 Table ) . When compared with isogenic WT K13 parasites , neither the C580Y nor the R539T variants displayed altered susceptibility to either proteasome inhibitor . V1/S parasites yielded slightly lower IC50 values than the Cam3 . II lines in response to both inhibitors . For WLL , these differences were not statistically significant . However , for WLW , Cam3 . IIC580Y showed a modest but nonetheless significant increase in IC50 as compared to V1/SC580Y . Two-way ANOVA showed a statistically significant , albeit small , difference in Cam3 . II and V1/S IC50 values overall for WLW . We also tested WLL and WLW against tightly synchronized early rings ( 0–3 hr post-invasion ) to determine whether K13 mutations might alter ring-stage parasite susceptibility to Plasmodium-selective proteasome inhibitors . These experiments focused on the Cam3 . II K13WT and Cam3 . II K13C580Y isogenic pair . Parasites were exposed to a 3 hr pulse of WLL or WLW across a range of concentrations , after which the inhibitor was removed by repeated rounds of washing ( see Materials and Methods ) . Cultures were continued for an additional 69 hr . These assays revealed IC50 values in the range of 52–54 nM for WLL and 516–531 nM for WLW for early rings , irrespective of their K13 genotype . We also performed 3 hr pulse assays with synchronized Cam3 . II K13WT and Cam3 . II K13C580Y trophozoites , sampled 24 hr later , which yielded substantially higher IC50 values of 422–453 nM for WLL and 1628–1698 nM for WLW . With both rings and trophozoites , we saw no difference between isogenic Cam3 . II lines expressing WT or mutant K13 . Similar to the 72 hr assay , WLL was notably more potent than WLW at both stages ( Fig 1C and 1D and S3 Table ) . We next examined the sensitivity of P . falciparum to the proteasome inhibitors WLL and WLW across the intra-erythrocytic developmental cycle by exposing tightly synchronized parasites to 1 hr drug pulses at specific intervals ( 45–47 , 0–3 , 10–13 , 18–21 and 24–27 hr post-invasion ) . These experiments were conducted on the Cam3 . II K13WT and Cam3 . II K13C580Y isogenic pair . Parasites were exposed to fixed drug concentrations ( 150 nM for WLL and 2000 nM for WLW ) for 1 hr , after which the inhibitor was removed by washout ( see Materials and Methods ) . Cultures were then continued until 72 hr from the start of the experiment . Percent survival was calculated relative to mock ( DMSO ) -treated cultures . As an additional control , we also treated parasites with DHA at 150 nM . In these 1 hr pulse assays , mature schizonts ( treated just prior to or at the time of egress ) and very early post-invasion rings were the most susceptible to proteasome inhibition by WLL and WLW ( Fig 1E and 1F ) . Mid to late rings and trophozoites were less susceptible , exhibiting ~five-fold higher survival rates as compared to schizonts upon exposure to either inhibitor ( Fig 1E and 1F and S4 Table ) . No differences were observed in the sensitivity profiles of Cam3 . II K13WT and Cam3 . II K13C580Y in response to either compound . By comparison , DHA showed a dissimilar stage-specificity profile , with maximal potency against late rings and trophozoites . This profile coincides with the peak period of hemoglobin uptake and degradation , during which Fe2+-heme is liberated and activates ART [27 , 28] . Considerable inhibition was nonetheless observed in other stages , including early rings , which are also thought to undergo some digestion of hemoglobin to activate ART [29] ( Fig 1G ) . Against schizonts , DHA treatment yielded almost twenty-fold higher survival levels than were observed with trophozoites ( S4 Table ) . In these experiments , we observed only a minor increase in survival with K13 C580Y mutant parasites as compared to the isogenic K13 WT in early rings in response to a 1 hr pulse of 150 nM DHA , with no difference observed at later rings and trophozoites . Prior work has established that increasing the concentration ( to 700 nM ) and length of exposure ( to 6 hr ) , in accordance with the ring-stage survival assay ( RSA0-3h ) , leads to a significant gain in the survival rate of the K13 C580Y mutant , specifically in early rings [30–32] . To test the efficacy of our washout protocol , we also exposed uninfected red blood cells ( RBCs ) to the same 1 hr drug pulses described above ( 150 nM WLL , 2000 nM WLW , 150 nM DHA , or DMSO vehicle control ) . Uninfected RBCs were washed as per the 1 hr exposure assays described above . Magnet-purified , synchronized late trophozoites were then added to drug-treated RBCs or control ( DMSO ) -treated RBCs , and parasites were cultured for 48 hr to allow for one cycle of RBC invasion and parasite growth . Parasitemias were measured by flow cytometry and percent growth was calculated relative to DMSO control pre-treated wells . These assays were conducted with the Cam3 . II K13WT and Cam3 . II K13C580Y lines . Results showed that trophozoites added to WLL- or DHA- treated RBCs expanded to levels comparable to those added to DMSO-treated RBCs ( 96 . 0% and 99 . 8% , respectively , when averaged across K13 WT and K13 C580Y lines and across three independent experiments; S5 Table ) . These data show that WLL and DHA were both effectively washed out to sub-inhibitory concentrations ( Fig 1E and 1G ) . By comparison , trophozoites inoculated into WLW-treated RBCs had reduced growth ( 79 . 7% on average across both lines relative to the DMSO control; S5 Table ) , which suggests that some of the inhibition observed in parasites treated with the high-dose of WLW ( Fig 1F ) may be attributable to drug carryover . Nonetheless , the inhibition profile observed for both WLL and WLW across stages was consistent , with both inhibitors showing maximal inhibition in schizonts and early ring stages , contrasting with maximum survival at the mid to late ring stage ( Fig 1E and 1F and S4 Table ) . To evaluate the ability of P . falciparum to generate resistance to the proteasome inhibitors WLL and WLW , we exposed cultured parasites to sub-lethal concentrations of WLL or WLW ( at either three or five times the IC50 level ) for a period of up to 60 days . Selection studies were performed with the two pairs of isogenic K13 WT and C580Y lines: Cam3 . II K13WT and Cam3 . II K13C580Y , and V1/S K13WT and V1/S K13C580Y . Selections were performed in triplicate , each with a large starting inoculum of 2×109 parasites per flask . Parasite clearance was confirmed during the first six days of treatment and subsequent recrudescence of parasitemia was monitored by microscopy two to three times a week . Initial studies using 5×IC50 drug pressure for WLL showed very low levels of recrudescence , with only 2 of 12 selection flasks resulting in detection of parasites by day 60 ( Table 1 ) . Survival in the presence of WLW occurred more readily , with 6 of 12 flasks yielding recrudescent parasites . In the case of WLL , recrudescent parasites did not appear until day 51 , whereas with WLW the mean time to recrudescence was 34 days for Cam3 . II parasites and 27 days for V1/S parasites . A second round of selections performed under 3×IC50 drug pressure resulted in greater levels of recrudescence , with 6 of 12 WLL flasks and 12 of 12 WLW flasks yielding parasites within 24–41 days ( Table 1 ) . These results also suggested a lower propensity for the Cambodian Cam3 . II line to develop resistance compared with V1/S ( 9 of 24 selections yielded parasites compared with 17 of 24 , respectively ) . These data are consistent with earlier findings that V1/S was ~two-fold more mutable than Cam3 . II [33] ( Cam3 . II was referred to therein as PH0306-C ) . Of note , K13 C580Y parasites were twice as likely to have positive selection outcomes when compared with K13 WT parasites ( 17 of 24 selections compared with 9 of 24 separate selections , respectively; Table 1 ) , suggesting that the K13 C580Y mutation might also modestly increase the mutation rate . To identify genetic changes mediating parasite recrudescence following WLL and WLW selection , we performed Illumina-based whole-genome sequencing on the drug-pressured recrudescent lines and their four parental counterparts . In total , ten unique mutations were identified from the 26 recrudescent lines ( Table 2 and S6 Table ) . Four of the eight WLL-selected parasite lines harbored a mutation in the β5 subunit of the 20S proteasome core particle , resulting in an alanine to serine substitution at amino acid position 20 ( A20S ) in the mature protein , while the other four harbored mutations in the 20S proteasome β6 subunit , yielding either an alanine to valine change at position 117 ( A117V ) or a serine to leucine substitution at position 208 ( S208L ) ( Table 2 and S6 Table ) . 14 of 18 ( 78% ) of WLW-selected parasite lines harbored a mutation in the 20S proteasome β2 subunit . These changes included a cysteine to phenylalanine or a cysteine to tyrosine mutation at position 31 ( C31F or C31Y ) , and an alanine to glutamic acid substitution at position 49 ( A49E ) . The four remaining lines selected against WLW harbored mutations in the 19S regulatory particle of the 26S proteasome . These mutations included a premature stop codon in RPT4 ( E380* ) , two non-synonymous mutations in RPT5 ( R295S and G319S ) , and a non-synonymous mutation in RPN6 ( E266K ) ( Table 2 and S6 Table ) . In the 26 sequenced drug-pressured lines , we observed no other mutations at ≥20% allele frequency in the core genome ( which removes sub-telomeric regions and members of multigene family members [34] ) . This observation further supports a primary role for the 19S and 20S proteasome mutations described above in mediating resistance . Copy number variation ( CNV ) analysis of the 26 lines detected amplification of a putative ubiquitin regulatory protein ( PF3D7_0808300 ) in two recrudescent lines . In both cases , the amplification occurred in V1/S K13C580Y parasites selected at 3×IC50 drug pressure with either WLL or WLW ( S6 Table; flask R2 in both cases ) , suggesting that this gene may contribute to low-grade resistance to WLL and WLW . Both drug-pressured lines also harbored single nucleotide polymorphisms in the 20S proteasome β5 ( A20S ) or β2 ( A49E ) subunits , providing evidence that the ability of parasites to withstand drug pressure in the bulk cultures can be multifactorial . As a quality control for our CNV analysis , we confirmed that the V1/S and Cam3 . II genomes differed at the GTP cyclohydrolase locus ( PF3D7_1224000 ) that is amplified in V1/S parasites ( contributing to high-grade pyrimethamine resistance; [35 , 36] ) . We also observed that V1/S parasites harbor a chromosome nine deletion that was previously associated with loss of cytoadhesion [37] . For these lines , the mean genome coverage was 54-fold ( range 16–85; S6 Table ) . From the set of WLL- or WLW-selected lines , we selected nine for 72 hr dose-response assays . Lines that contained a given proteasome mutation at ≥ 90% allele frequency were assayed directly without cloning , whereas lines that displayed mixed parasite populations were first cloned by limiting dilution . These lines are referred to herein by their unique proteasome mutations and are grouped by their respective parent: RPT4 E380* and RPN6 E266K ( selected in the Cam3 . II K13WT parental line ) ; β5 A20S , β2 C31Y and RPT5 G319S ( Cam3 . II K13C580Y ) ; β6 A117V and β2 C31F ( V1/S K13WT ) ; and β6 S208L and β2 A49E ( V1/S K13C580Y; S7 Table ) . In parallel , we also assayed the four parental lines: Cam3 . II K13WT , Cam3 . II K13C580Y , V1/S K13WT , and V1/S K13C580Y . Results are shown in Fig 2 , in which K13 WT and K13 C580Y lines are colored blue and red , respectively . WLL-selected and WLW-selected lines are shown with thin and thick hatching , respectively . WLL-selected lines harboring β5 A20S or β6 A117V mutations were observed to confer small ( ~two-fold ) but nonetheless statistically significant increases in WLL IC50 levels compared with their parental lines ( Fig 2A and S8 Table ) , and slightly higher increases ( up to three-fold ) in the WLL 90% inhibitory concentration ( IC90 value; S9 Table ) . The WLL-selected β6 S208L line showed only very modest ( <two-fold ) and not statistically significant increases in WLL IC50 and IC90 values ( Fig 2A and S8 Table and S9 Table ) . By comparison , WLW-selected lines harboring a β2 subunit mutation ( C31Y , C31F and A49E ) revealed slightly larger shifts ( ~three- to five-fold ) in their WLW IC50 values compared to their parental lines ( Fig 2B and S8 Table ) . WLW-selected lines harboring mutations in the 19S proteasome regulatory particle ( RPT4 E380* , RPN6 E266K , and RPT5 G319S ) displayed small ( ~two-fold ) but significant increases in IC50 values compared with their parental lines . These trends were maintained at the IC90 level ( S9 Table ) . We next evaluated the ability of mutations identified in WLW-selected lines to confer resistance to WLL and vice versa . Surprisingly , we observed that the β2 C31F and β2 C31Y lines ( both WLW-selected ) were hypersensitive to inhibition by WLL , despite the fact that WLL inhibits both the β2 and β5 subunits of the parasite proteasome . The third β2 mutant line , β2 A49E , showed no shift in WLL IC50 or IC90 as compared to the parental line . Similarly , none of the 19S mutant lines displayed any cross-resistance to WLL ( Fig 2A and S8 Table and S9 Table ) . The three mutations identified in WLL-pressured lines affected sensitivity to WLW in distinct ways: whereas the β5 A20S line showed significant hypersensitization to WLW , the β6 A117V line showed a minor ( 2 . 5-fold ) increase in WLW IC50 as compared to the parental line . Finally , the β6 S208L mutation did not result in any significant shift in WLW IC50 or IC90 values ( Fig 2B and S8 Table and S9 Table ) . To simulate the effects of mutations in the β2 , β5 and β6 subunits on WLL and WLW binding in silico , we performed structural analyses with both inhibitors using the high resolution cryo-EM based structure of the P . falciparum 20S proteasome ( PDB accession code 5FMG ) . For these studies , we docked WLL into the β5 active site of the cryo-EM-based atomic model of the P . falciparum 20S proteasome ( Fig 3A and 3B ) and for WLW used the previously solved structure of the inhibitor bound to the β2 active site ( Fig 3F and 3G ) [21] . As expected , docking studies revealed that WLL was well-accommodated within the β5 active site ( Fig 3B ) . To examine the effect of the WLL-selected mutations β5 A20S and β6 A117V and S208L , we used molecular dynamics simulations to individually evaluate the effects of these mutations on the β5 active site ( Fig 3C–3E ) . For the β5 A20S mutation , the introduction of a serine side chain was predicted to directly impose steric constraints on the S3 binding pocket of the β5 active site . These constraints would not favor large groups at the P3 position of the ligand , such as the P3 tryptophan of WLL . The β6 A117V mutation was predicted to destabilize a cluster formed by three tyrosine residues at positions 150 , 152 and 158 of β6 , inducing a slight displacement of the β strands of the β6 subunit towards the β5 binding pocket . This conformational change would produce steric constraints on the β5 active site binding pocket , particularly for access of the WLL P3 group , as recently suggested for the similar β6 A117D substitution [23] . For the β6 S208L mutation , introduction of the leucine side chain in β6 was predicted to cause a significant clash with the adjacent α helix in the β3 subunit . This substitution induced conformational rearrangements in the model that were able to propagate as far as the S3 site of the β5 binding pocket . These changes would again impose steric constraints on WLL binding ( Fig 3C–3E ) . The rearrangements induced by the WLL-selected mutations were all characterized by smaller , sterically-constrained β5 binding pockets , particularly at the S3 position , as compared with the WT Plasmodium proteasome β5 active site . We also modeled the WLW-selected β2 mutations , namely C31Y , C31F and A49E , using the known cryo-EM based P . falciparum 20S proteasome structure ( 5FMG; [21] ) . The three mutated residues are located near the S1 binding pocket of the β2 active site . The introduction of bulky tyrosine or phenylalanine residues in place of the cysteine at position 31 was predicted to cause a steric clash with the large P1 tryptophan of WLW , likely preventing binding of WLW to the β2 subunit ( Fig 3H and 3I ) . Similarly , the β2 A49E mutation was predicted to produce steric constraints near the S1 binding pocket caused by the introduction of the bulky glutamate side chain ( Fig 3J ) . Structural modeling suggested that these β2 mutants should still retain sensitivity to WLL , which has a smaller P1 side chain compared to WLW . These predictions are consistent with the lack of cross-resistance between the WLW-selected mutants and WLL; in fact , the β2 C31Y and C31F mutant lines selected under WLW pressure showed hypersensitivity to WLL ( Fig 2 ) . We next examined whether mutations in the 20S proteasome subunits confer resistance to WLL or WLW by directly precluding binding of the inhibitors to the active sites of the proteasome . These experiments involved activity-based probe ( ABP ) labeling of the three catalytic subunits β1 , β2 and β5 using the proteasome active-site fluorogenic probe BMV037 ( [26 , 38 , 39]; see Materials and Methods ) . This probe competes for binding with proteasome-specific inhibitors including WLL and WLW , which allows for direct assessment of inhibitor binding to each of the active beta subunits of the proteasome through the quantification of residual protein labeling after inhibitor treatment . Competition assays were performed for the three WLL-selected lines harboring β5 or β6 mutations ( β5 A20S , β6 A117V , and β6 S208L ) and the three WLW-selected lines harboring mutations in β2 ( C31Y , C31F and A49E ) . P . falciparum schizont lysates from the six test lines and their respective parents were pre-incubated for 1 hr with WLL or WLW at concentrations ranging from 0 . 5 to 50 μM , or mock-treated , then incubated for 2 hr with the BMV037 probe . In the absence of either inhibitor , the probe showed similar labeling regardless of whether lines carried mutations in the proteasome subunits or not ( Fig 4A ) . In the presence of increasing concentrations of WLL or WLW , probe labeling of the β2 and β5 subunits was reduced by the binding of these inhibitors . To quantify this effect , we calculated the concentration at which WLL or WLW achieved half-maximal inhibition of probe labeling to each subunit ( shown as IC50 values in Fig 4B–4D and S10 Table; see Materials and Methods ) . For the WLL-selected lines , the β5 A20S mutation had a small effect on WLL binding to β5 , manifesting as a slight increase in the WLL β5 IC50 ( Fig 4A and 4C ) . This result is consistent with our structural analysis . This mutation , however , did not impact inhibition of the β2 subunit by either WLL or WLW ( Fig 4A , 4B and 4D ) . The WLL-selected β6 mutation A117V slightly reduced the potency of WLL in blocking the labeling of the β5 subunit by BMV037 , whereas the β6 S208L mutation appeared to have a minor effect on WLL binding to β2 . These two β6 mutations did not substantially affect binding of WLW to the β2 active site ( Fig 4A–4D ) . For the WLW-selected lines , the β2 C31Y and C31F mutations both reduced WLW binding to the β2 active site , but did not prevent WLL binding to β2 ( Fig 4A–4D ) . This finding agrees with our structural analysis , which shows a preference against mutant site occupancy by the large tryptophan P1 group of WLW . This also explains why the WLW-selected mutants were not resistant to WLL , which has a leucine in the P1 position . The third WLW-selected β2 mutation , A49E , also prevented WLW binding to β2 , but not as strongly as C31Y or C31F . This mutation resulted in a reduction of WLL binding to β2 , which is consistent with the proximity of this mutation to the entrance of the β2 binding pocket [40]; Fig 4A–4D ) . To test for interactions between WLL or WLW and other classes of antimalarials , we performed isobologram assays with the isogenic ART-S Cam3 . II K13WT and ART-R Cam3 . II K13C580Y lines . These assays included DHA , the related endoperoxide-containing compound OZ439 that is suspected to have a similar mode of action [41 , 42] , methylene blue ( MB ) that disrupts redox homeostasis , and two compounds implicated in pathways related to the UPS , namely b-AP15 that inhibits proteasome-associated deubiquitinases , and eeyarestatin I ( ESI ) that inhibits the ER-associated degradation ( ERAD ) pathway ( S11 Table ) . Assays were initiated with either asynchronous parasites that were exposed to drugs for 72 hr , or with synchronized early rings ( 0–3 hr post-invasion ) or trophozoites ( tested 24 hr later ) that were exposed to drugs for 3 hr followed by three rounds of drug washout and a further 69 hr of incubation in drug-free medium prior to measuring parasitemias . Drug combinations were tested at fixed ratios ( 1:0 , 4:1 , 2:1 , 1:1 , 1:2 , 1:4 , 0:1 ) across a range of concentrations ( see Materials and Methods ) . From these data , we derived fractional IC50 ( FIC50 ) values for the two compounds at each of the ratios tested , and plotted these values on isobologram graphs . The shapes of the resulting curves were then compared against a hypothetical isobole line illustrating a perfectly additive interaction ( dashed line in Fig 5 ) . With these graphs , a concave curve with points lying substantially below the isobole line is evidence of synergy , whereas points near the isobole line indicate additivity , and a convex curve with points lying substantially above the isobole line indicates antagonism . Our drug combination studies provided clear evidence of synergy between WLW and DHA , OZ439 , MB and b-AP15 , for both the ART-S and ART-R isogenic lines ( Fig 5 ) . These results were obtained with asynchronous parasites , as well as synchronized early post-invasion rings and trophozoites . Rings showed the clearest evidence of synergy , as evidenced by the most concave curves . This finding was particularly significant as rings are generally the least susceptible to antimalarial drugs including ART derivatives ( Fig 1G; [43] ) , with the notable exception of proteasome inhibitors that our data show are the most potent against this stage . Mild synergy was observed between WLW and ESI at the early ring stage , however the interaction between these two compounds was additive on asynchronous parasites and on trophozoites ( Fig 5 ) . Synergy was also evidenced with DHA , OZ439 and b-AP15 in combination with WLL on early rings and trophozoites , though to a reduced degree as compared with WLW ( S1 Fig ) . For MB , synergy with WLL was limited to the early ring stage , and for ESI , the interaction with WLL was largely additive ( S1 Fig ) . We extended our isobologram studies to the mitochondrial inhibitor atovaquone ( ATQ ) , the licensed ACT partner drugs and suspected heme-interacting agents lumefantrine ( LMF ) and piperaquine ( PPQ ) , and the former first-line antimalarial chloroquine ( CQ; S11 Table ) . Additive to antagonistic profiles were observed between the proteasome inhibitors and these four compounds . Results for these compounds and those tested above are shown as the mean of the sums of the fractional IC50 values of the combinations ( mean ΣFIC50 ) and are represented as heat maps ( Fig 6; values tabulated in S12 Table ) . A mean value less than or equal to 0 . 5 indicates that the interaction between the two compounds was potently synergistic ( blue ) , a mean close to 1 . 0 is indicative of an additive interaction ( white ) , and a mean greater than or equal to 1 . 5 indicates potent antagonism ( red ) . These thresholds are rarely met with P . falciparum , in part because such interactions manifest the most clearly only at certain combination ratios [21 , 44–46] . Mean ΣFIC50 values lying between these cutoffs , i . e . greater than 0 . 5 but less than 1 . 0 , or greater than 1 . 0 but less than 1 . 5 , suggest mild synergy or mild antagonism , respectively . For LMF , mild antagonism was observed on asynchronous parasites with WLL or WLW , while these interactions were largely additive on early rings and trophozoites . For PPQ and ATQ , the antagonism observed in asynchronous parasites and early rings was attenuated in trophozoites . For CQ , moderate to potent antagonism was observed across all stages ( Fig 6A–6C ) . These heat maps visually illustrate the potent synergy observed with DHA , OZ439 , MB and b-AP15 ( Fig 5 ) . We also tested a panel of experimental compounds with diverse targets and modes of action , including AN3661 , an inhibitor of the P . falciparum cleavage and polyadenylation specificity factor subunit 3; ACT-451840 , suspected to inhibit PfMDR1; cycloheximide , an inhibitor of tRNA translocation and protein synthesis; DDD107498 , an inhibitor of P . falciparum translation elongation factor 2; DSM265 , a dihydroorotate dehydrogenase inhibitor; halofuginone , a prolyl tRNA synthetase inhibitor; and NITD609 , an inhibitor of P . falciparum P-type Na+ ATPase 4 ( S11 Table ) [47–53] . Each compound showed antagonism with WLL and WLW . No substantial differences were observed in the responses of K13 WT versus K13 C580Y parasites to any of these compounds ( S2 Fig and S13 Table ) .
The evolution of drug-resistant P . falciparum parasites has severely compromised prior-generation first-line antimalarials such as CQ , with devastating consequences . Resistance is also increasingly undermining the efficacy of ACTs in SE Asia . The search for new antimalarials has uncovered a spectrum of novel scaffolds active against specific parasite targets , including mitochondrial factors ( DHODH , cytochrome B ) , the Na+-ATPase PfATP4 , the cis-Golgi protein PfCARL , several cytosolic tRNA synthetases , and the eukaryotic elongation factor PfeEF2 [54 , 55] . Nonetheless , despite their nanomolar potency , many of these inhibitors can readily select for resistance via mutations in their respective target proteins , sometimes with as few as 106 or 107 parasites . Also , depending on the target and inhibitor , the IC50 increases observed in resistant parasites can vary from several fold up to two thousand-fold [54] . These findings can have direct implications for treatment outcomes in vivo , as recently evidenced in a clinical trial where two of 24 P . falciparum-infected patients treated with a low single dose of DSM265 recrudesced with parasites that had acquired DHODH mutations previously observed in DSM265 resistance selections in vitro [56] . Here , we report that the Plasmodium proteasome-specific peptide vinyl sulfone inhibitors WLL and WLW have low nanomolar potency against genetically diverse parasites , are equally effective against parasites expressing mutant or WT forms of the ART resistance determinant K13 , and display a unique stage-specificity profile . Stage-specificity assays reveal WLL and WLW to be most potent against schizonts and early ring stages . This finding is particularly promising as the majority of currently-employed antimalarials are most active against the trophozoite stage , including ART derivatives ( which also inhibit rings ) [43] . Importantly , we also observe a minimal resistance liability with WLL and WLW . Inocula of 2 × 109 parasites exposed to low drug levels yielded recrudescent parasites in only half the selections , with IC50 increases averaging <three-fold ( Table 1 and S8 Table ) . WLL- or WLW-resistant lines retained full sensitivity to the alternate inhibitor , and in several cases mutations that conferred resistance to one inhibitor sensitized parasites to the other . This finding suggests that resistance is highly compound-specific and that selective pressures exerted by different inhibitors even within the same series can act in opposing directions . The different selectivity of WLL and WLW was confirmed by activity-based profiling of the proteasome catalytic sites in parasites harboring mutations in β2 , β5 or β6 , which revealed that WLL-selected mutations interfered specifically with binding of WLL and not WLW , and vice versa . We also leveraged the existing high-resolution cryo-EM structure of the P . falciparum 20S proteasome [40] to perform in silico modeling , which predicted that mutations in the β2 , β5 and β6 subunits could specifically reconfigure proteasome active sites in the 20S core particle to confer compound-selective resistance . Gene-editing studies , which would provide an additional layer of confirmation , have not yet been undertaken . Of note , a recent study of noncovalent , reversible asparagine ethylenediamine ( AsnEDA ) 20S proteasome inhibitors reported parasites that were cross-resistant to several AsnEDA inhibitors but were hypersensitized to the pan-active proteasome inhibitors bortezomib and carfilzomib , as well as to WLW [23] . This finding adds to the evidence that proteasome inhibitor resistance can be class- , and , in the case of the vinyl sulfones , even compound-specific . Resistance selections with the AsnEDA compound PKS21004 yielded parasites with high levels of resistance ( >130-fold increases over the sensitive parent ) , and yielded mutations in the same residue as one of our WLL-pressure lines ( β6 A117D for PKS21004 versus β6 A117V for WLL ) . These differing levels of resistance may be attributable to distinct modes of inhibition for the AsnEDA compounds versus the vinyl sulfone inhibitors , notably the ability of the latter to form permanent covalent linkages with the active site threonine of the proteasome beta subunits . Vinyl sulfone-mediated inhibition is controlled by an initial reversible binding event and the subsequent formation of a covalent adduct with the proteasome active site . Mutations identified in our resistant lines may alter the initial binding of the inhibitor , which would in turn limit the rate of covalent modification of the active site . However , the potency of these inhibitors would only be nominally compromised because in the mutant enzyme covalent adducts would still accumulate over time . For reversible binding compounds such as the AsnEDA series , these same mutations are predicted to lead to reduced steady-state levels of inhibitor-bound active sites , causing higher resistance levels . Our data suggest that compounds with covalent modes of inhibition may thus be preferable over reversible binding inhibitors in helping to reduce the risk of high-grade resistance . We also sought to explore potential partner agents for combination therapies including proteasome inhibitors . Results from isobologram studies revealed potent synergy between WLL or WLW and five distinct antimalarial agents: DHA , OZ439 , methylene blue , b-AP15 , and ESI . Synergy was the most pronounced with early rings , highlighting the value of assessing drug-drug interactions with synchronized cultures . Our data suggest that these structurally diverse compounds might share a common feature of generating damaged or misfolded proteins that accumulate as UPS substrates . Given that the UPS is a major regulator of the cell stress response , we propose that inhibition of the P . falciparum proteasome precludes parasites from resolving protein damage caused by these compounds , thereby creating synergy . DHA and OZ439 , for instance , are endoperoxide-containing drugs that generate carbon-centered radical species , which non-specifically alkylate intracellular heme and other biomolecules in blood-stage parasites [41 , 42 , 57 , 58] . This activity is suspected to increase the burden of misfolded and damaged proteins . We confirmed earlier observations of proteasome inhibitor synergy with DHA [15 , 21–25] . Leveraging the availability of isogenic K13 WT and mutant parasites we also showed that synergy was unaffected by K13 sequence . Our studies also revealed potent synergy between WLL or WLW and OZ439 , an ozonide compound related to ART that is now in human clinical trials [2 , 59 , 60] . MB is also of interest for potential combination therapies , as this redox-perturbing drug is very potent against Plasmodium asexual blood stages and gametocytes and has proven gametocytocidal efficacy in P . falciparum-infected patients [61–63] . Its activity has been attributed in part to disruption of glutathione ( GSH ) redox cycling [64] . GSH is an antioxidant that neutralizes cellular damage by reactive oxygen species ( ROS ) . The major source of ROS in P . falciparum is parasite-mediated hemoglobin proteolysis and the oxidation of liberated , reactive heme . Destabilization of the parasite’s antioxidant defense mechanisms results in widespread damage to cell membranes , proteins and other molecules [65] . Similar to DHA and OZ439 , treatment with MB might increase the parasite’s reliance on the UPS for clearance of damaged biomolecules , creating synergy with proteasome inhibitors . Prior experiments have also shown synergy between MB and ART derivatives , further indicating that both drugs may activate similar cellular defense pathways in response to protein damage , including the UPS [66 , 67] . Of the two experimental compounds that we tested , one , the anti-cancer agent b-AP15 , acts directly on the UPS , inhibiting the activity of proteasome-associated deubiquitinases ( DUBs ) that catalyze the deubiquitination of proteasome-targeted substrates prior to their translocation into the 20S proteolytic machinery [68 , 69] . Interference with ubiquitin deconjugation from target substrates prevents polypeptide translocation into the 20S proteasome catalytic core and abrogates proteasomal degradation [17] . The antimalarial activity of b-AP15 has been attributed to inhibition of PfUSP14 and/or PfUCH54 , two 19S-subunit associated DUBs [68–70] . The second compound , ESI , inhibits the UPS-related ER-associated degradation ( ERAD ) pathway that mediates the disposal of misfolded ER-resident or trafficked proteins [71 , 72] . In ERAD , misfolded proteins are shuttled out of the ER through a retrotranslocation machinery into the cytoplasm where they are ubiquitinated and subsequently degraded by the proteasome [72] . In human cancer cell lines , ESI specifically inhibits p97 , an AAA-ATPase that is an essential component of the ER retrotranslocon [73 , 74] , suggesting this as a potential target in Plasmodium . Synergies observed between our proteasome inhibitors and both ESI and b-AP15 can likely be attributed to the disruption of two targets in the UPS pathway . Both sets of interactions suggest promising avenues for developing novel combination therapies . Our data also identified several antimalarial classes that were antagonistic with proteasome inhibitors . These included inhibitors of hemoglobin metabolism and heme detoxification ( CQ , PPQ and putatively LMF ) , mitochondrial function ( ATQ and DSM265 ) , protein synthesis ( AN3661 , CHX , DDD107498 and HFG ) , sodium homeostasis ( NITD609 ) , and digestive vacuole transport processes ( ACT-451840 ) [54 , 55] . These drugs may antagonize proteasome inhibitors by reducing the protein degradative burden on the UPS . As an example , attenuating protein synthesis through translation inhibitors ( such as DDD10798 or CHX ) could reduce the parasite’s reliance on the proteasome to eliminate defective nascent proteins and thereby diminish the impact of proteasome inhibition [16 , 75] . For the 4-aminoquinolines CQ and PPQ , antagonism of WLL and WLW might be attributable to their inhibition of hemoglobin proteolysis , which occurs at similar concentrations to their inhibition of hemozoin formation [76] . This may lead to translation attenuation due to a lack of available amino acid precursors stemming from liberated and digested globin , thereby reducing dependency on the proteasome and antagonizing its inhibitors . This postulate could be addressed by studying hemoglobinase inhibitors , such as ALLN and E-64 [45 , 77] . Further studies are required to determine whether inhibition of mitochondrial functions ( specifically pyrimidine biosynthesis and maintenance of the electron transport chain , inhibited by DSM265 and atovaquone respectively ) would also antagonize proteasomal inhibitors by attenuating protein synthesis . Our findings that WLL and WLW share a low propensity for selecting for parasite resistance and favorable stage-specificity and synergy profiles provide a compelling case for the continued development of Plasmodium-selective proteasome inhibitors as antimalarial therapeutics . Ongoing efforts are focused on improving selectivity and pharmacological properties of the lead vinyl sulfone inhibitors . Recently , we reported a new set of optimized peptide vinyl sulfone inhibitors , chemically related to WLL , which retained their potency and synergy with DHA and displayed over three orders of magnitude selectivity for the P . falciparum enzyme . These compounds had improved solubility , metabolic stability , and oral bioavailability , and were active in a P . berghei rodent malaria model [24] . The data presented herein reveal multiple chemical classes that display synergistic interactions with peptide vinyl sulfones , and highlight covalent proteasome inhibitors as promising new agents for use in resistance-refractory combination therapies to treat multidrug-resistant malaria .
The P . falciparum Cam3 . II and V1/S lines were previously engineered to express WT K13 or the C580Y or R539T variants [9] . Parasite lines were maintained in RBCs obtained from Interstate Blood Bank ( Memphis , TN ) at 3% hematocrit , in RPMI 1640 medium supplemented with gentamicin , hypoxanthine , and Albumax II . Cultures were maintained at 37°C in modular incubator chambers gassed with 5% CO2 , 5% O2 and 90% N2 . To obtain highly synchronized parasites , predominantly ring-stage cultures were exposed to 5% D-Sorbitol ( Sigma-Aldrich ) for 15 min at 37°C to remove mature parasites . After 36 hr of subsequent culture , multinucleated schizonts were purified over a 75% Percoll ( Sigma-Aldrich ) gradient or a magnetic-activated cell sorting ( MACS ) column ( Miltenyi Biotec ) . Purified schizonts were allowed to invade fresh RBCs for 3 hr , and early rings ( 0–3 hr post-invasion ) were treated with 5% D-Sorbitol to remove any remaining schizonts . These synchronized rings were then used for stage-specific assays . Synchronized trophozoites were harvested 24 hr later . IC50 values were determined by testing parasites against two-fold serial dilutions of antimalarial compounds [48] . Compounds were tested in duplicate in 96-well plates , with the final volume per well equal to 200 μL . Parasites were seeded at 0 . 2% parasitemia and 1% hematocrit . Parasites were either continuously exposed to drugs for 72 hr , or pulsed with drug for 3 hr followed by three rounds of washing to remove drug and a further 69 hr of culture in drug-free medium . Washes were performed by centrifuging 96-well plates at 800×g for 2 minutes to pellet cells , removing drug-containing medium , and resuspending in an equal volume of fresh , drug-free medium . On the third wash , cultures were transferred to a new 96-well plate . Removal of media and resuspension were performed on a Freedom Evo 100 liquid-handling instrument ( Tecan ) . After 72 hr , parasites were stained with 1×SYBR Green and 100 nM Mitotracker Deep Red ( ThermoFisher ) and parasitemias were measured on a BD Accuri C6 Plus Flow Cytometer with a HyperCyt attachment sampling 10 , 0000–20 , 000 events per well [78] . Data were analyzed using FlowJo and IC50 values were derived using nonlinear regression analysis ( GraphPad Prism ) . Highly-synchronized parasites were exposed to WLL ( 150 nM ) , WLW ( 2000 nM ) , DHA ( 150 nM ) , or DMSO vehicle control for 1 hr followed by drug washout and further culture . All stage-specificity tests were concluded 72 hr from the start of the experiment . Assays were performed in duplicate . Parasites were prepared by synchronizing ring stages with 5% D-Sorbitol and later isolating the schizonts over a 75% Percoll gradient . These schizonts were allowed to reinvade fresh RBCs at 2% hematocrit . Parasites were tested as purified schizonts , early rings , mid rings , late rings or trophozoites ( ~45–47 , 0–3 , 10–13 , 18–21 and 24–27 hr post-invasion , respectively ) , and exposed to drug or DMSO vehicle control in 1 mL volumes . Parasites were seeded at 0 . 5% parasitemia and 1% hematocrit in a 48-well plate . Drug-treated parasites were washed as previously described for the RSA0-3h [9] . Briefly , 1 mL cultures were transferred to 15 mL conical tubes , centrifuged at 800×g for 5 minutes to pellet cells , and the culture medium containing drug was carefully removed . Cells were subsequently resuspended in 10 mL drug-free medium , centrifuged , and the wash medium removed . Finally , cells were resuspended in 1 mL of fresh drug-free medium , transferred to a new well , and returned to standard culture conditions for the duration of the assay . Parasitemias were measured by flow cytometry and the survival of drug-treated parasites was calculated as a percent of DMSO-treated control cultures . Uninfected RBCs were exposed to WLL ( 150 nM ) , WLW ( 2000 nM ) , DHA ( 150 nM ) or DMSO vehicle control for 1 hr , followed by drug washout exactly as above . For each treatment , 20 μL of packed RBCs were exposed to drug , equivalent to 1 mL of a 2% hematocrit complete media and blood mixture . Synchronized late trophozoites were purified over a MACS column ( Miltenyi Biotec ) following initial synchronization with 5% D-Sorbitol , and were exposed to drug-pretreated RBCs at 0 . 5% parasitemia . Experiments were performed in duplicate . Parasitemias were measured by flow cytometry 48 hr later and percent growth was calculated relative to DMSO-treated control cultures . To select for WLL- or WLW-resistant parasites , triplicate flasks of 2×109 parasites , each with a starting parasitemia below 2% , were exposed to these compounds at concentrations of 3× or 5× their IC50 values . Selections were performed on Cam3 . II K13WT , Cam3 . II K13C580Y , V1/S K13WT , and V1/S K13C580Y . Drug-containing media was refreshed every day for the first six days , then every 2–4 days . RBCs were replenished every seven days and volumes were reduced by half every week starting on day 14 . Cultures were monitored by Giemsa staining and microscopy every day until parasites cleared , then monitored two to three times per week to detect recrudescence . Selections were maintained for 60 days or until recrudescent parasites were observed microscopically . gDNA was prepared from 0 . 05% saponin-lysed cultures using a DNeasy Blood and Tissue Kit ( Qiagen ) . To prepare the sequencing libraries , gDNA was fragmented and amplified with the Nextera XT kit ( Cat . No FC-131–1024 , Illumina ) using the standard dual index protocol . Libraries were sequenced on an Illumina HiSeq 2500 using the RapidRun mode [34] . Sequence reads ( 2×100 bp ) were aligned to the P . falciparum 3D7 reference genome ( PlasmoDB v . 13 . 0 ) using the Platypus pipeline and the Genome Analysis Toolkit’s ( GATK ) HaplotypeCaller was used to call single nucleotide variants ( SNVs ) , copy number variants ( CNVs ) , or insertion/deletions ( INDELs ) [79 , 80] . SNVs were filtered out if they met the following criteria: ReadPosRankSum >8 . 0 or <−8 . 0 , QUAL <500 , Quality by Depth ( QD ) <2 . 0 , Mapping Quality Rank Sum <−12 . 5 , or filtered depth ( DP ) <7 . INDELS were filtered out if they met the following criteria: ReadPosRankSum <−20 , QUAL <500 , QD <2 , or DP <7 . We also removed mutations where the read coverage was <5 . Variants were annotated using SnpEff [81] . In instances where drug-selected lines showed sequence heterogeneity , clones were generated by limiting dilution and the candidate SNVs were confirmed by targeted gene sequencing . Proteasome activity was profiled using the BMV037 active-site probe [26 , 38] . Synchronized late schizonts were harvested and lysates were prepared by adding equal volumes of hypotonic lysis buffer ( 50 mM Tris pH 7 . 4 , 5 mM MgCl2 , 1 mM DTT ) to parasite pellets . Lysates were incubated on ice for 1 hr , with occasional vortexing , then spun at 13 , 000×rpm for 15 min to recover supernatants . Protein concentrations were determined using a Bradford assay ( Pierce ) . Lysates ( 10 μg ) were pre-incubated with each inhibitor for 1 hr at 37°C prior to adding 10 μM BMV037 for a further 2 hr at 37°C . Samples were denatured by adding SDS sample buffer , then briefly boiled and electrophoresed on a 12% SDS-PAGE gel . Gels were scanned on the Cy5 channel on a Typhoon Scanner ( GE Healthcare ) . IC50 values were calculated for each inhibitor for β2 and β5 ( WLL ) or β2 only ( WLW ) by quantifying labeled subunits using ImageJ and normalizing to mock-treated DMSO controls . For these labeling studies with parasite lysate , inhibitor concentrations ( 0 . 5 to 50 μM ) were selected based on an earlier study with BMV037 [21] . The cryo-EM derived atomic structure of the P . falciparum 20S proteasome ( PDB accession code 5FMG ) was used to model the following mutations in silico: A20S in β5 , A117V in β6 , S208L in β6 , C31Y in β2 , C31F β2 , and A49E in β2 . These models were used to dock WLL or WLW into the mutant β5 and β2 subunits , respectively . For the β5 and β6 mutations , the structural consequences of these mutations were examined using molecular dynamics simulations on models containing only the active site subunits without the WLL ligand . Standard procedures in the AMBER 14 software package ( PME MD protocol ) were used , namely generation of topology files for proteins and preparation of AMBER input data using the LeaP module , development of the hydration model in a periodic water box ( TIP3P model of water for explicit solvent ) , thermodynamic equilibration of the system and the solvent , and molecular dynamics calculations at 37°C using the AMBER ff99bsc0 force field with a simulation time of 10 ns [82] . The most frequent states for each simulation were used to infer the effects of mutations on the binding of WLL . In contrast , the β2 mutations were superposed onto the P . falciparum 20S β2β3 dimer structure ( PDB accession code 5FMG ) . These mutated models were prepared using the Protein Preparation Wizard in Maestro ( Schrödinger ) . Structures were minimized with a harmonic constraint on all heavy atoms ( maximum RMSD of 0 . 3 Å ) . The binding poses of WLW in the 20S β2 mutants obtained by superposition were then refined by local minimization with Prime ( Schrödinger ) . This local minimization included WLW as well as all protein residues within 5 Å of the ligand , and used the variable-dielectric Generalized Born solvation model . Graphic representations of all resulting models were prepared using the PyMOL Molecular Graphics System ( Schrödinger ) . Assays were performed with fixed ratios of drug combinations ( 1:0 , 4:1 , 2:1 , 1:1 , 1:2 , 1:4 , and 0:1 ) , tested in duplicate [45] . Combinations were prepared from 16×IC50 drug stocks and tested across a range of two-fold dilutions . Assays were conducted with asynchronous cultures exposed for 72 hr or with tightly synchronized rings or trophozoites ( tested at 0–3 hr or 24–27 hr post-invasion , respectively ) exposed for 3 hr . Post-pulse drug washouts were conducted as described above for in vitro determination of IC50 levels . IC50 values were derived for each compound tested alone , and fractional IC50 ( FIC50 ) values were determined for each compound tested in combination ( FIC50 = IC50 of the drug alone/IC50 of the drug in combination ) and plotted for each drug combination . For each combination , the mean of the sums of FIC50 values at each combination ( mean ΣFIC50 ) was also calculated and the results illustrated using heat maps . Human RBCs used in this study were purchased from the Interstate Blood Bank ( Memphis , TN ) as blood from anonymized donors . Approval to use this material for P . falciparum in vitro culture has been granted by the Columbia University Medical Center Institutional Review Board , which has classified this work as not being human subjects research . | The spread of artemisinin-resistant Plasmodium falciparum malaria across Southeast Asia creates an imperative to develop new treatment options with compounds that are not susceptible to existing mechanisms of antimalarial drug resistance . Recent work has identified the P . falciparum proteasome as a promising drug target . Here , we report potent antimalarial activity of highly selective vinyl sulfone-conjugated peptide proteasome inhibitors , including against artemisinin-resistant P . falciparum early ring-stage parasites that are traditionally difficult to treat . Unlike many advanced antimalarial candidates , these covalent proteasome inhibitors do not readily select for resistance . Selection studies with cultured parasites reveal infrequent and minor decreases in susceptibility resulting from point mutations in components of the 26S proteasome , which we model using cryo-electron microscopy-based structural data . No parasites were observed to be cross-resistant to both compounds; in fact , partial resistance to one compound often created hypersensitivity to the other . We also document potent synergy between these covalent proteasome inhibitors and multiple classes of antimalarial agents , including dihydroartemisinin , the clinical candidate OZ439 , and the parasite transmission-blocking agent methylene blue . Proteasome inhibitors have significant promise as components of novel combination therapies to treat multidrug-resistant malaria . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"trophozoites",
"parasite",
"groups",
"medicine",
"and",
"health",
"sciences",
"chemical",
"compounds",
"plasmodium",
"methylenes",
"cancer",
"treatment",
"drugs",
"microbiology",
"parasitology",
"antimalarials",
"oncology",
"mutation",
"apicomplexa",
"methylene",
"blue",
... | 2019 | Covalent Plasmodium falciparum-selective proteasome inhibitors exhibit a low propensity for generating resistance in vitro and synergize with multiple antimalarial agents |
The DNA uptake of naturally competent bacteria has been attributed to the action of DNA uptake machineries resembling type IV pilus complexes . However , the protein ( s ) for pulling the DNA across the outer membrane of Gram-negative bacteria remain speculative . Here we show that the competence protein ComEA binds incoming DNA in the periplasm of naturally competent Vibrio cholerae cells thereby promoting DNA uptake , possibly through ratcheting and entropic forces associated with ComEA binding . Using comparative modeling and molecular simulations , we projected the 3D structure and DNA-binding site of ComEA . These in silico predictions , combined with in vivo and in vitro validations of wild-type and site-directed modified variants of ComEA , suggested that ComEA is not solely a DNA receptor protein but plays a direct role in the DNA uptake process . Furthermore , we uncovered that ComEA homologs of other bacteria ( both Gram-positive and Gram-negative ) efficiently compensated for the absence of ComEA in V . cholerae , suggesting that the contribution of ComEA in the DNA uptake process might be conserved among naturally competent bacteria .
Recombination between the bacterial chromosome and DNA fragments that enter the cell through horizontal gene transfer ( HGT ) either replace damaged or mutated alleles with the original alleles , thereby repairing the gene , or transfer mutated alleles or new genes to naïve strains . Thus , HGT plays a key role in transferring genetic information from one bacterium to another and maintaining the balance between genome maintenance and evolution . Natural competence for transformation is one of three modes of HGT in bacteria and promotes the uptake of free DNA from the environment ( for recent reviews see [1]–[6] ) . Many naturally transformable bacteria have been described [7] , including the pathogenic bacterium Vibrio cholerae [6] , [8] . The physiological state of natural competence of this Gram-negative bacterium is associated with its primary niche , the aquatic environment . Within this habitat , V . cholerae attaches to the exoskeleton of zooplankton or zooplankton molts [9] . Those exoskeletons comprise the polymer chitin , which is the natural inducer of competence in V . cholerae [6] , [8] , [10] . Whereas the regulatory network driving competence has been well investigated ( reviewed by Seitz and Blokesch [6] ) , so far very little is known about the DNA uptake complex of V . cholerae [11] . With respect to the DNA uptake machinery of naturally transformable bacteria it has been suggested that a ( pseudo- ) pilus [1] , [2] , similar to type IV pili ( Tfp ) [12] , represents a core element of the DNA import machinery . However , it is still unclear how the proteins interact to pull the transforming DNA through the cell envelope [3] . A proposed mechanism for DNA uptake involves repeating cycles of pilus extension and retraction [1] , [2] , [4] , [13] although recent review articles suggested that other competence proteins , such as ComEA , might be involved in pulling the DNA into the cell [4] , [14] ( though without experimental evidence ) . The present study reinforces those ideas and shows that ComEA is a prerequisite for DNA uptake in naturally competent V . cholerae . Furthermore , based on an earlier study on DNA ejection from bacteriophages [15] we propose a model suggesting that the DNA translocation across the outer membrane is possibly accomplished by ratcheting and entropic forces associated with the binding of ComEA to the incoming DNA . Currently , the majority of studies on the cellular localization of competence proteins were performed on Gram-positive bacteria [16]–[19] , whereas far less is known about competence protein localization in Gram-negative bacteria . We recently identified the minimal competence gene set of V . cholerae and provided first insight into the DNA uptake machinery of this organism [11] . Notably , through the analysis of knockout strains lacking specific components of the DNA uptake complex we demonstrated that natural transformation still occurred in the absence of the proteins involved in the Tfp structure and biosynthesis though at very low frequencies . Such rare transformants were never detectable for comEA− strains [11] , suggesting that ComEA plays an important role in the DNA uptake process , the focus of this work . In studies on B . subtilis and S . pneumoniae it was reported that binding of transforming DNA to those Gram-positive cells is at least partially mediated by ComEA and that ComEA is “absolutely required” for DNA uptake and transformation [20]–[22] . Likewise , ComE ( ComEA homolog ) -negative strains of Neisseria gonorrhoeae [23] and V . cholerae [8] , [24] were severely or completely impaired for natural transformability , indicating that ComEA might also play an important role in Gram-negative bacteria . A recent study by Lo Scrudato and Blokesch indicated that comEA and the gene encoding the inner membrane transporter comEC were differentially regulated from the Tfp-like components of the DNA uptake machinery [25] , [26] , which , together with our study on the DNA uptake machinery , suggest that DNA transport might be a multi-step process in V . cholerae ( as previously proposed for Helicobacter pylori [14] , [27] , which does not contain a bona fide Tfp-based DNA uptake machinery ) . Here , we show that the Tfp-like elements of the DNA uptake machinery of V . cholerae are not sufficient to translocate DNA across the outer membrane and that the competence protein ComEA plays an essential role in this process .
In a previous study by Chen and Gotschlich the authors predicted a 19-residues signal sequence for sec-dependent transport of the ComEA-homolog of Neisseria gonorrhoeae ( ComE ) into the periplasm [23] . Such a predictable signal sequence ( amino acid residues 1–25 ) is also present in ComEA of V . cholerae . To experimentally address the localization of the ComEA protein we aimed at visualizing it in vivo by constructing a functional translational fusion between ComEA and mCherry . Using this construct we observed a uniform localization pattern of ComEA ( Fig . 1A ) , which is consistent with the presence of such an N-terminal signal sequence and the transport of ComEA to the periplasm . To validate this microscopical observation , we generated a translational fusion between comEA and the gene encoding beta-lactamase ( bla; without the region encoding the signal sequence ) , which replaced the wild-type comEA allele on the V . cholerae large chromosome . The resulting strain retained natural transformability at a frequency of 2 . 5×10−5±3 . 0×10−5 compared with 7 . 9×10−5±2 . 5×10−5 for the parental wild-type strain ( average of four biological replicates ± SD ) indicating the functionality of the fusion construct . Most importantly , the construct conferred full resistance to ampicillin , which provides further evidence for the periplasmic localization of ComEA-bla as beta-lactamase can only exert activity against beta-lactam antibiotics in the periplasm of Gram-negative bacteria ( Fig . S1 ) . Next , we aimed to investigate whether the ComEA protein is motile within the periplasm . To this extent we used a fluorescence loss in photobleaching ( FLIP; Fig . 1B ) approach because photobleaching can reveal protein dynamics in live cells [28] . In contrast to fluorescence recovery after photobleaching ( FRAP ) , where fluorescent proteins within a small area of the cell are bleached and the back-diffusion of the surrounding non-bleached proteins into this region is recorded , FLIP consist of repetitive bleaching of the same region ( e . g . region of interest 1 in Fig . 1B ) , thereby preventing fluorescence recovery in that region . Moreover , any mobile protein from elsewhere in the same compartment ( e . g . region of interest 2 in Fig . 1B ) will also enter this continuously photo-bleached area , eventually resulting in a complete loss of fluorescence in the compartment . In contrast , any not connected compartment will be spared from bleaching ( e . g . region of interest 3 in Fig . 1B ) . Therefore , FLIP is often used to reveal the mobility of proteins within certain compartments of the cell [29] , which is what we were aiming for . Indeed , our FLIP experiments indicated that ComEA was highly motile within the periplasm ( Fig . 1B ) . Likewise , a translational fusion between the signal sequence of ComEA ( amino acid residues 1–25; ss[ComEA] ) alone and mCherry resulted in a similar localization ( Fig . 1A ) and mobility pattern ( Fig . S2 ) . This uniform localization pattern differed from that obtained from previous studies on B . subtilis , where Hahn et al . used immunofluorescence microscopy to show that ComEA localizes in a non-uniform punctate manner [16] . Kaufenstein et al . confirmed those data and concluded that the distinct assemblies of ComEA were mobile [19] . Studies using purified tagged ComEA/ComE homologs demonstrated that the protein binds DNA in vitro; thus , ComEA was considered as a DNA receptor protein [21] , [23] , [30] , [31] . DNA binding could be attributed to a conserved helix-hairpin-helix ( HhH ) motif [32] . Notably and in contrast to helix-turn-helix or helix-loop-helix motifs , which are widespread in proteins that interact with DNA in a sequence-dependent manner , HhH motifs bind DNA in a non-sequence-specific manner . Such binding is based on hydrogen bonding between the protein and the DNA phosphate groups [32] and HhH motifs have been described in various protein classes , including DNA polymerases , DNA ligases or DNA glycosylases [32] , [33] . However , the in vivo binding of DNA through ComEA has never been demonstrated . We genetically engineered a fusion protein between ComEA and GFP , which was transported across the inner membrane via the Tat-transport machinery in a folded state ( as GFP is improperly folded when translocated to the periplasm in a sec pathway dependent manner [34] ) . Interestingly , the protein failed to translocate in Escherichia coli; instead , ComEA was tightly bound to the bacterial chromosome , which appeared as a highly compacted structure ( Fig . 2A ) . The increased protein expression levels resulted in cell death , indicating that the strong binding of ComEA to the DNA in vivo interfered with cellular processes . Due to this lack of translocation of ComEA-GFP into the periplasm and the in vivo binding to the chromosome we conducted further experiments using the ComEA-mCherry fusion despite the lower signal intensity of mCherry compared with GFP . To investigate the function of ComEA in vivo , we excluded artifacts caused by artificial ( over- ) expression as those have been recognized as having detrimental effects on subcellular localization [35] . Thus , all V . cholerae strains used in these experiments were generated through the substitution of chromosomal comEA with diverse comEA-mCherry alleles . In these strains , the expression of comEA-mCherry was driven through its native promoter and consequently co-regulated with other competence genes . The functionality of the chromosomally encoded ComEA fusion protein was confirmed using a transformation assay , and the chromosomally-encoded fusion protein was uniformly localized within the periplasm ( Fig . 2B and Fig . S3 ) . Importantly , the addition of external transforming DNA ( tDNA ) led to the formation of distinctive ComEA-mCherry foci ( Fig . 2B ) . The size and numbers of these protein aggregates was dependent on the length of the supplemented tDNA . Periplasmic mCherry alone did not aggregate ( ss[ComEA]-mCherry; Fig . 2B ) . A similar relocalization pattern after the addition of external DNA was also observed when the cells were grown on chitin surfaces mimicking the natural reservoir V . cholerae ( Fig . S4 ) . This observation suggested that ComEA binds transforming DNA in the periplasm thereby potentially contributing to DNA translocation across the outer membrane . To test this hypothesis , we repeated the experiments using YoYo-1-labeled DNA . Indeed , a perfect colocalization pattern was observed when the fluorescent signals of ComEA-mCherry and DNA were compared ( Fig . 2C ) . Foci formation through ComEA and colocalization with YoYo-1-labeled DNA were absent in a strain lacking the outer membrane pore PilQ [2] , [8] , [11] , whereas the absence of the inner membrane transporter ComEC did not interfere with ComEA-DNA colocalization ( Fig . 2C ) . Similar foci formation of YoYo-1-labeled DNA was also observed in a strain carrying wild-type ComEA , excluding a translational artifact resulting from the mCherry-fusion ( Fig . 2C ) . Notably , YoYo-1 foci were absent in a comEA-negative strain , which was also the case for a strain lacking the major Tfp subunit PilA ( Fig . S5A ) . Using a whole-cell duplex PCR-based DNA uptake assay [24] , [11] that aims at detecting DNA strands , which have either entered the periplasm or have already reached the cytoplasm of the competent bacteria ( thereby becoming resistant against externally applied DNase ) , we confirmed that tDNA ( both unlabeled or YoYo-1-labeled ) was undetectable in comEA-negative strains even though it was readily detectable in the wild-type strain and in comEC negative derivatives ( Fig . 2D and Fig . S5 ) . Whereas the absence of YoYo-1 labeled DNA foci and PCR-amplifiable DNA in comEA negative strains is indicative of a failure to transport tDNA across the outer membrane , such results would also be consistent with ComEA's main function being to protect and stabilize incoming tDNA against potential nucleases . Indeed , two nucleases have been described for V . cholerae , Dns and Xds , which are solely responsible for extracellular nuclease activity in this organism [36] . Interestingly , Focareta and Manning demonstrated that even though Dns can be recovered from culture supernatants , it was also detectable in the periplasmic space of V . cholerae [37] . We recently confirmed the extracellular localization of Dns [38] but also its at least partial association with the bacterial cells ( through western blot analysis; [25] ) . Moreover , Blokesch and Schoolnik showed that expression of dns has to be silenced in V . cholerae to allow natural transformation to occur at high cell density [26] , [38] . Thus , to rule out the possibility that ComEA might protect incoming tDNA against either of those two nucleases we tested dns , xds , and comEA single , double , and triple mutants for natural transformation and the recovery of DNase resistant tDNA in whole cells ( Fig . S6 ) . Notably , the absence of dns resulted in higher transformability ( Fig . S6A ) , consistent with an early study [38] , and in the detection of increased amounts of DNase-resistant tDNA within the bacteria ( Fig . S6B ) . However , no transformants or translocated tDNA were detectable if comEA was concomitantly absent ( Fig . S6 ) . We therefore conclude that ComEA's main role is not to protect incoming tDNA against degradation by the nucleases Xds or Dns , though we cannot exclude the presence of any other hitherto unidentified nuclease in the periplasm of V . cholerae . Instead , we suggest that translocation of tDNA across the outer membrane is not solely driven through Tfp-like elements of the DNA uptake machinery but also requires ComEA . To gain insights into the molecular mechanism through which ComEA binds dsDNA , we predicted the structure of ComEA and characterized the interactions of this protein with the transforming DNA . First , we used comparative modeling to create a 3D structure of ComEA using the X-ray structure of the ComEA-related protein HB8 from Thermus thermophilus ( PDB ID: 2DUY , unpublished ) as a template ( Fig . 3 , movie S1 ) . Based on structural similarity with structures from the HhH family [39] , we identified K62 and K63 as candidate residues for DNA binding interactions and could model the putative ComEA-DNA adduct ( Fig . 3B ) . The electrostatic potential of the ComEA model is consistent with the identified DNA-binding region , showing positively charged regions corresponding to the lysine pair ( Fig . 3C ) . To validate this model , we used site-directed mutagenesis to create ComEA variants with single or double amino acid substitutions . All comEA-mCherry alleles were inserted into the chromosome , thereby replacing the wild-type comEA copy . The ComEA-mCherry variants were tested for expression and periplasmic localization , foci formation upon provision of tDNA , for their ability to induce DNA translocation into a DNase resistant state ( using the DNA uptake assay ) and to restore natural transformation ( Fig . 4 and Fig . S7 ) . Consistent with the in silico predictions , K63 was of major importance . ComEAK63A was severely impaired for natural transformation ( ∼250-fold reduction; Fig . 4C ) , resulting in DNA uptake levels below the limit of detection ( Fig . 4B ) . The substitution of K63 with a negatively charged residue ( ComEAK63E ) or the concomitant exchange of K62 ( ComEAK62/63A ) completely abolished natural transformation ( Fig . 4C ) . The ComEA-DNA model also explains why K63 has the major role in DNA binding: while K62 is engaged with a single backbone phosphate moiety , K63 is inserted into the DNA minor groove , chelating the backbone of both strands ( Fig . 3B , inset ) . Moreover , a substitution of the nearby glycine residue at position 60 by alanine had no effect on DNA binding and transformation , whereas strains producing ComEAG60V and ComEAG60E were impaired in DNA uptake and were non-transformable ( Fig . 4 ) . We suggest that the combined effect of impairing the interactions of K62 and K63 with the dsDNA ( as in the case of ComEAG60E ) and perturbing the HhH1 GIG hairpin motif ( Fig . 3A ) has a major impact on the ability of ComEA to bind DNA . To further investigate whether the lysine pair is indeed involved in DNA binding we heterologously expressed those variants as tat-ComEA-GFP fusions in E . coli ( Fig . S8; as for wild-type ComEA in Fig . 2A ) . Using this approach we showed that the ComEAK63E and ComEAK62/63A variants behaved differently from WT ComEA in that they localized evenly within the cytoplasm . In addition , most of the E . coli cells did not show any compaction of the chromosome ( and if so the variant did not co-localize with the compacted chromosome ) . The same phenotype was observable for variants that lacked either of the two HhH motifs ( Fig . S8 ) , suggesting that those variants had lost their ability to bind DNA . In contrast , a K63A variant showed an intermediate phenotype ( Fig . S8 ) consistent with the ∼250-fold decreased transformation frequency observed for the ComEAK63A-mCherry variant in V . cholerae ( Fig . 4 ) . Apart from this patch at HhH1 , the only other amino acid important for the in vivo functionality of ComEA was the conserved arginine residue at position 71 ( Fig . 3A ) . The DNA uptake ability of ComEAR71A was slightly reduced , and less DNA-protein foci were observed for this variant ( Fig . S7 ) . However , the strain containing ComEAR71A remained naturally transformable , a feature that was completely abolished for the ComEAR71D variant . The latter mutant protein was also unable to bind DNA within the periplasmic space and did not foster the uptake of transforming DNA ( Fig . S7 ) . Based on our ComEA model structure , R71 is located in a position not particularly favorable for DNA binding ( Fig . 3B ) ; therefore , it is likely that R71 might be important for the structural stability of ComEA . To unambiguously show that the lysine residues are required for DNA binding we purified a tagged ( Strep-tag II ) version of ComEA , ComEAK62/63A , ComEA-mCherry , and ComEAK62/63A-mCherry ( Fig . S9 ) . The purified ComEA protein showed an unexpected UV-Vis spectrum , which was consistent with bound DNA ( due to an absorption peak around 260 nm; Fig . S9A ) . Interestingly , if we compared purified ComEA-mCherry with the ComEAK62/63A-mCherry , we observed that the peak at 260 nm was absent in this variant , indicating that the protein was indeed no longer able to bind DNA . To remove any pre-bound DNA from the ComEA protein we included a DNase treatment step prior to the elution of the protein from the affinity column ( see Material and Methods; Fig . S9C and D ) . All four proteins were tested for in vitro binding to DNA using an electrophoretic mobility shift assay ( EMSA ) . Notably , ComEA-mCherry and ComEA bound to DNA in a concentration dependent manner as visualized by the retarded migration of the DNA probe ( Fig . 5A and Fig . S10A ) and the likewise changed migration of the protein ( visualized by the fluorescence of mCherry; Fig . 5A ) . Notably , the K62/63A variants of ComEA did not change the migration behavior of the DNA probe ( Fig . 5B and Fig . S10B ) , again confirming that the protein had lost the ability for DNA binding . It should be noted that the shifted DNA signal was detectable at DNA to protein ratios as low as 1∶10 and the probe seemed completely shifted at a ratio of 1∶25–30 ( Fig . 5A and Fig . S10A ) , which was significantly lower than what has been described for the B . subtilis ComEA homolog ( 98% of the DNA probe was shifted when 5 . 5×10−11 M of DNA was incubated with 1 . 6 µM of purified protein; [21] ) or for the neisserial ComE ortholog [23] . A possible explanation for this difference could be that the ComEA/ComE proteins investigated in those earlier studies were pre-occupied by DNA as we observed for ComEA of V . cholerae in the absence of DNase treatment . Provvedi and Dubnau suggested that the in vitro DNA binding behavior of the ComEA protein of B . subtilis was indicative of cooperative binding [21] . To test whether any cooperative binding was observable for ComEA of V . cholerae we used Atomic Force Microscopy ( AFM ) . AFM allows investigating the extent of ComEA-mCherry binding to a DNA fragment and to also determine where on the DNA the protein is bound ( e . g . fractional occupancies at any specific site , binding to the ends , or to nonspecific sites ) . To minimize overestimation of the binding affinity that can occur in the case when coverage of protein on the surface is too high , such that the protein coincidently lands on DNA , we kept the DNA-protein molecular ratio low by not exceeding a ratio of 1∶10 ( DNA to protein ) . Prior to AFM imaging , we pre-incubated the ComEA-mCherry protein with a random PCR fragment ( 809 bp ) at a molecular ratio of 1∶2 . 5 or 1∶10 . As illustrated in Fig . 5 we observed a mixture of bare DNA molecules , free protein molecules , and protein/DNA complexes . To identify the ComEA-mCherry protein in topographic AFM images we used height and width criteria ( height >2 nm , width from 10 to 20 nm ) . Using an approach reported by Yang et al . [40] we found that the probability of protein molecules located on DNA was 5 times higher than it would be for stochastically binding of the protein to the mica surface . Moreover , in the case of a DNA to protein ratio of 1∶10 we observed 2 . 5-fold higher affinity of the protein to the free ends of DNA than to random sites on the DNA strand . These AFM data indicate that , at least at the measured concentrations , no cooperative binding of the ComEA protein to DNA occurred and again contradicts the hypothesis that binding of ComEA might primarily protect the tDNA from degradation . Such protective effect has been demonstrated for the competence protein DprA of Streptococcus pneumoniae [41] , which binds the single-stranded tDNA after its translocation into the cytoplasm . Indeed , Mortier-Barrière et al . , described in their study that DprA binding to DNA appeared to be cooperative since fully covered protein-DNA complexes were observed next to free ssDNA molecules at a protein to nucleotide ratio of 1∶20 . We never observed such scenario for ComEA's binding to dsDNA using AFM ( though we used a ∼4-fold lower protein to nucleotide ratio ) . Interestingly , a passive DNA uptake mechanism has recently been proposed for single-stranded T-DNA translocation into plant cells involving the VirE2 protein of Agrobacterium tumefaciens [42] . We reasoned that if a similar mechanism is responsible for DNA uptake in competent V . cholerae cells , although dsDNA is involved and ComEA shows no similarity to VirE2 , then the aggregation of ComEA should occur at one distinct DNA entry point ( most likely next to the PilQ secretin ) . To test this hypothesis , we performed time-lapse microscopy experiments using ComEA-mCherry-expressing V . cholerae strains in the presence of external DNA ( Fig . 6 ) . We consistently observed the accumulation of ComEA as one large focus before smaller subclusters separated from the main ComEA focus and spread throughout the periplasm until the uniform localization of ComEA was restored ( Fig . 6 , movies S2 , S3 , S4 ) . Based on the data presented above we hypothesize that ComEA might play a direct role in the translocation of DNA across the outer membrane solely based on its ability to bind to DNA . If this were the case then ComEA homologs of other naturally competent bacteria should be able to replace ComEA of V . cholerae . And indeed , ComEA of B . subtilis was able to efficiently compensate for the absence of ComEA of V . cholerae ( Fig . 7 ) . Moreover , even the C-terminal ( HhH ) 2 motif of ComEA of B . subtilis alone , which was shown to bind DNA in vitro [21] , was sufficient to restore natural transformation of a comEA negative V . cholerae strain as were the ComEA homologs from N . gonorrhoeae , Haemophilus influenzae , and Pasteurella multocida ( Fig . 7 ) . It should be noted that Sinha et al . suggested that H . influenzae might contain an additional but so far unidentified paralog of comE1 due to the modest effect observed for a comE1 minus strain [43] . It is tempting to speculate that ComEA might fulfill a similar role in Gram-positive bacteria . Indeed , the localization of ComEA has been previously described for B . subtilis [16] , [17] , [19] but those studies were either based on immunofluorescence microscopy [16] , which does not allow following protein localization over time , or were done in the absence of tDNA [17] , [19] . Therefore , it was concluded by Kaufenstein et al . that ComEA localizes to many sites of the cell membrane and only occasionally co-localizes with the polar DNA uptake machinery , which was mainly achieved by changing the artifical inducer concentration [19] . However if the cell wall would be considered as a similar barrier in Gram-positive bacteria as the outer membrane is in Gram-negatives , creating a kind of periplasmic space between the cell wall and the ( inner ) membrane as suggested by Matias and Beveridge [44] , then the binding of ComEA could also participate in the transport of DNA across the cell wall layer . However , in contrast to ComEA of Gram-negative bacteria , ComEA of Gram-positives is anchored to the membrane and therefore accumulation of ComEA can only occur in two dimensions , which might still be sufficient to prevent backward diffusion of the tDNA and contribute to DNA translocation across the cell wall . Notably , while this article was under revision Bergé et al . published a study on the nuclease EndA of naturally competent Streptococcus pneumoniae [45] . The authors demonstrated that EndA aggregates at midcell in this Gram-positive bacterium and that this recruitment is dependent on “the dsDNA receptor” ComEA [45] . Interestingly , ComEA also localized to the midcell and the authors speculated “a direct interaction of EndA and ComEA , an hypothesis which received indirect support” [45] . Our findings suggest that the ability of ComEA proteins to bind to dsDNA emerging from the PilQ pore can potentially prevent the retrograde movement of the substrate , and ComEA binding might contribute to pull DNA into the periplasm ( Fig . 8 ) . It has been suggested that ratcheting produced through binding proteins can significantly accelerate translocation events [46] , [47] , as for the case of phage DNA injection into bacterial cells [15] . Based on our data , a similar mechanism can be envisioned for the ComEA-mediated transfer of DNA into the periplasm , with the rate of uptake depending on the specific binding kinetics and concentration of ComEA [15] . We hypothesize that ComEA-mediated DNA internalization might start occurring once short stretches of tDNA would enter the periplasm ( most likely through the outer membrane secretin PilQ and potentially after a single Tfp retraction event ) . The ratio between the periplasmic ComEA protein and the incoming tDNA should be high at that stage thereby leading to an increased ComEA effective binding density , which , potentially together with the higher affinity of ComEA for DNA ends as observed by AFM ( Fig . 5 ) , would promote efficient DNA internalization . The absence of cooperative ComEA-DNA binding revealed by our AFM data ( Fig . 5 ) is not an obstacle to a ComEA-mediated ratchet mechanism of internalization , as cooperativity would only contribute to increase the relative speed of the process [15] , [46] , [47] . The binding of proteins has undeniably been recognized as a driving force , both in the translocation of proteins as well as of DNA [48] . To this extent , Salman et al . investigated the translocation of double-stranded ( ds ) DNA through the nuclear pore complex using a combination of epifluorescence microscopy and single-molecule manipulation techniques [49] . They presented evidence that the DNA uptake process in their reconstituted system was based on a passive ratchet , directed by the retention of the already translocated segment of the DNA [49] . We suggest that ComEA might play a similar role in the DNA uptake process in naturally competent V . cholerae cells . In summary , we used a cell biological approach to better understand DNA uptake in naturally competent V . cholerae cells . We visualized the competence protein ComEA and observed the in vivo binding of this protein to dsDNA in real time . Structural modeling and AFM experiments suggested that the binding of ComEA to DNA is primarily responsible for DNA translocation across the outer membrane . Consistent with this suggestion , ComEA variants unable to bind to DNA in vivo were also defective in promoting DNA uptake and natural transformation . We hypothesize that ComEA encounters incoming DNA immediately after short stretches of DNA have crossed the outer membrane ( through the PilQ secretin or in exceptional cases also in a Tfp-independent manner [11] ) and that ComEA subsequently promotes DNA translocation across the outer membrane without the need for any external energy source ( Fig . 8 ) . ComEA might therefore be more than a DNA receptor protein , but rather a crucial player for mediating DNA uptake in V . cholerae and potentially also other naturally competent bacteria .
Vibrio cholerae strains and plasmids used in this study are listed in Table S1 . Escherichia coli strain DH5α [50] was used as host for cloning purposes and for heterologous expression of ComEA and its variants for protein purification . Genomic DNA ( gDNA ) extracted from E . coli BL21 ( DE3 ) [51] was utilized to test DNA uptake by PCR as described [24] . E . coli S17-1λpir [52] served as donor strain for bacterial mating with V . cholerae . All V . cholerae and E . coli strains were grown aerobically in Luria-Bertani ( LB ) medium at 30°C and 37°C , respectively . Solid LB plates contained 1 . 5% agar . For tfoX expression and induction of other constructs under control of the PBAD promoter the LB medium was supplemented with 0 . 02% L-arabinose ( L-ara ) . For expression of tat-gfp , tat-comEA-gfp , and its derivatives in E . coli DH5α ( Fig . 2A and Fig . S8 ) L-ara concentrations were lowered to 0 . 002% . Thiosulfate Citrate Bile Salts Sucrose ( TCBS ) agar plates were prepared following the manufacturer's instructions ( Fluka ) and used to counterselect E . coli after bacterial mating . For sucrose-based counterselection , NaCl-free LB medium containing 6% sucrose was used . LB medium and LB agar plates were supplemented with antibiotics when required . Final concentrations of antibiotics were 50 µg/ml , 75 µg/ml and 100 µg/ml for gentamicin , kanamycin , and ampicillin , respectively . The ampicillin concentration was lowered to 50 µg/ml for V . cholerae strains induced for competence . Standard molecular biology-based methods were used for DNA manipulations . Restriction enzymes and DNA modifying enzymes were obtained from New England Biolabs , Taq DNA polymerase ( GoTaq ) was obtained from Promega and used for colony PCR , and Pwo DNA Polymerase ( Roche ) was used for high-fidelity PCR amplifications . Modified DNA sequences were verified using Sanger sequencing ( Microsynth , CH ) . All plasmid constructs were based on pBAD/Myc-HisA ( Invitrogen ) , which contains the araBAD ( PBAD ) promoter followed by a multiple cloning site ( MCS ) for dose-dependent protein expression . A derivative of pBAD/Myc-HisA , pBAD ( kan ) , was created through substitution of the ampicillin resistance cassette ( bla ) with a kanamycin resistance cassette ( aph ) . The genes and translational fusion constructs were PCR amplified and cloned into the MCS of pBAD/Myc-HisA or pBAD ( kan ) . For the amplification of V . cholerae genes , the gDNA of strain A1552 [53] served as a template . The accuracy of the plasmids was verified through sequencing . Genes were deleted from the parental strain A1552 , using either a gene disruption method based on the counter-selectable plasmid pGP704-Sac28 [54] , or natural transformation and FLP recombination , as recently described ( TransFLP method [55]–[57] ) . Strains containing comEA-mCherry or site-directed variants thereof were constructed using the TransFLP method [55]–[57] . For the construction of ComEA site-directed variants , a silent ‘watermark’ restriction site was inserted close to or including the changed nucleotide sequence . This watermark simplified screening purposes after homologous recombination . The comEAB . s . gene ( or parts thereof ) was amplified from gDNA derived from B . subtilis strain 168 . The DNA fragment containing comE1 from Neisseria gonorrhoeae ( N . g . ; Neisseria gonorrhoeae strain FA 1090 , NCBI Reference Sequence: NC_002946 . 2; locus YP_208252 ) , Haemophilus influenzae ( H . i . ; Haemophilus influenzae strain R2846 , NCBI Reference Sequence: NC_017452 . 1; locus YP_005829750 ) , and Pasteurella multocida ( P . m . ; Pasteurella multocida subsp . multocida str . Pm70 , NCBI Reference Sequence: NC_002663 . 1; locus NP_246604 , hypothetical protein PM1665 ) was synthesized using the GeneArt® Strings™ technology ( Life technologies/Invitrogen ) and served as PCR template for the TransFLP strain construction method [55]–[57] . The beta-lactamase gene ( bla ) was amplified from plasmid pBR-flp [55]–[57] . All strains were verified through colony PCR ( in part followed by restriction enzyme digestion according to inserted watermarks ) and confirmed through PCR amplification and sequencing . Microscopy images were obtained using a Zeiss Axio Imager M2 epifluorescence microscope . Details about the instrumentation and configurations are provided elsewhere [25] . All bacterial samples were mounted on 2% agarose/PBS pads . Image processing and annotation was done using ImageJ and Adobe Illustrator . Strains carrying fluorescent fusion constructs were grown aerobically for ∼5 h in LB supplemented with the respective antibiotics and 0 . 02% L-arabinose ( 0 . 002% L-ara for E . coli experiments; Fig . 2 and Fig . S8 ) . The strains carrying chromosomally encoded fluorescent fusion proteins were grown aerobically and at 30°C in LB supplemented with 0 . 02% L-ara for ∼7 h ( OD600 2 . 5; [11] ) . The samples were washed once in PBS and immediately imaged . The staining of chromosomal DNA was performed through the addition of 4′ , 6-diamidino-2-phenylindole ( DAPI; final concentration 5 µg/ml ) to the bacterial cultures for at least 5 min . To characterize the ComEA-mCherry localization dynamics during DNA uptake , comEA-mCherry-expressing strains were grown as described above . A total of 50 µl of washed culture was mixed with 1 µg of either gDNA derived from V . cholerae strain A1552-lacZ-Kan [58] , commercially available phage lambda DNA ( Roche ) or a 10 . 3 kb fragment amplified through PCR . After 5 min of incubation with the DNA the bacteria were mounted on agarose pads and imaged . To visualize the DNA during the relocalization of ComEA-mCherry , phage lambda DNA ( Roche ) was pre-stained with 10 µM YoYo-1 ( Molecular Probes/Invitrogen ) at 4°C corresponding to a base pair to dye ratio of 15∶1 . The bacterial culture was mixed with the pre-stained DNA and incubated for 20 min . The cells were washed in PBS , mounted on agarose pads and imaged . For time-lapse microscopy , the samples were prepared as described above , but immediately imaged after the addition of DNA . The images were taken every 3 or 120 sec as indicated in the figure and movie legends . For time-lapse imaging , the agarose pads were sealed using a mixture of Vaseline , lanolin and paraffin ( VALAP ) . Fluorescence loss in photobleaching ( FLIP ) experiments were performed on a Zeiss LSM710 microscope equipped with a 561 nm solid-state laser ( 20 mW ) . A Plan-Apochromat 63×/1 . 40 Oil objective was used . The microscope was controlled with the Zen 2009 software suite ( Zeiss ) . Time intervals ranged from 104 to 120 ms/frame for live cells to max . 160 ms/frame for fixed cells . The maximum ( 100% ) laser power was used for bleaching . V . cholerae strains ΔcomEA-TntfoX harboring pBAD ( kan ) -comEA-mCherry or pBAD ( kan ) -ss[ComEA]-mCherry were grown aerobically for 5 h in LB supplemented with 0 . 02% arabinose and 75 µg/ml of kanamycin . After the cells were mounted , the slides were sealed and the bacteria were immediately imaged ( live samples; Fig . 1 and Fig . S2A ) . Alternatively , the cells were fixed for 30 min ( 4% paraformaldehyde/150 mM phosphate buffer ) before imaging ( fixed samples; Fig . S2B ) . For FLIP data acquisition a circular bleaching region of ∼440 nm width was defined at one cell pole ( region-of-interest ( ROI ) 1; labeled as 1 in Fig . 1 ) . A circular ROI of the same size was defined at the opposite cell pole of the same bacterium ( labeled as 2 in Fig . 1 ) and in an adjacent cell ( labeled as 3 in Fig . 1 ) . The average fluorescence intensity of all regions was recorded . Bleaching of ROI 1 was initiated after a lag of 20 frames and repeated after each frame . The acquired data were exported and processed in ‘R’ [59] . The recorded fluorescence intensities were normalized to the average fluorescence intensity of the first 10 frames . Moving averages were calculated using the SMA ( x , n = 5 ) function from the ‘TTR’ package [59] . Transformation assays were performed as previously described [25] with gDNA of strain A1552-lacZ-Kan [58] as transforming material . Transformation frequencies were calculated as the number of transformants divided by the total number of colony forming units ( CFU ) . Differences in transformation frequencies were considered significant for P-values below 0 . 05 ( * ) or 0 . 01 ( ** ) as determined by Student's t-test on log-transformed data . DNA uptake was verified using a whole-cell duplex PCR assay as described [24] with slight modifications . Briefly , competence-induced bacteria were grown aerobically until an OD600 of 1 . 0–1 . 5 before genomic DNA ( gDNA ) ( 2 µg/ml ) of E . coli strain BL21 ( DE3 ) was added for 2 h . For the uptake of YoYo-1-labeled DNA gDNA of E . coli strain BL21 ( DE3 ) was pre-labeled as described for the microscopy experiments and YoYo-1 was maintained in the solution throughout the 2 h incubation period . Next , cells were harvested and treated with DNase I ( Roche ) for 15 min at 37°C . Excess nuclease was removed by washing and cells were resuspended in 100 µl PBS . ∼3×106 bacteria were used as template in a whole-cell duplex PCR . Primer pairs were specific for the donor DNA derived from E . coli BL21 ( DE3 ) and for gDNA of the V . cholerae acceptor strain ( at a 10-fold lower concentration ) . The latter reaction served as control for the total number of acceptor bacteria [24] . A 3D model structure was produced for ComEA ( truncating the first 37 residues including the 25 residue-containing signal peptide ) using comparative modeling ( MODELLER package [60] ) on the Thermus thermophilus HB8 ( PDB ID: 2DUY ) template ( with 43% sequence identity ) ( Fig . 3 ) . The ComEA-DNA complex was modeled , to identify structurally similar DNA-binding proteins using the DALI server [39] . The DNA polymerase , PolC , from Geobacillus kaustophilus ( PDB ID: 3F2D ) [61] was selected as the best match , with 24% sequence identity and a root mean square deviation ( RMSD ) of 2 . 4 Å compared with the modeled ComEA of V . cholerae . The PolC X-ray structure complexed with DNA was used to identify potential DNA poses on the V . cholerae ComEA model using the Chimera MatchMaker tool [62] . This assessment led to the production of a DNA-ComEA model ( Fig . 3B , movie S1 ) , which was further refined and equilibrated using the minimization and molecular simulations detailed below . The estimated binding energy for the ComEA-DNA association is in the order of 29±8 kcal/mol , based on MM/PBSA calculation on the MD trajectory . Molecular dynamics simulation was used to relax and study the dynamics and energetics of ComEA and the ComEA-DNA complex for 55 and 50 ns , respectively . The MD simulations were run using the NAMD simulation package [63] with Amber force field ( with Barcelona modification for nucleic acids [64] and the TIP3P water model [65] . The systems were first energy minimized using constrained C-alpha atoms , followed by analysis without any constraint for 2000 steps . To equilibrate the system , the temperature was gradually increased up to 300 K in the NVT ensemble and maintained at 300 K for 100 ps with a 1 fs time step . Finally , an NPT simulation was run at 300 K for 500 ps with a 2 fs time step to complete the equilibration procedure . The equilibrated structure was used as starting point for production simulations . All production MD simulations were run at 1 bar with a time step of 2 fs , using SHAKE algorithm [66] on all bonds and PME [67] for treating electrostatic interactions . To control the temperature and the pressure , Langevin dynamics and the Nose-Hoover Langevin piston , respectively , were used [68] , [69] . The trajectories were saved every 500 steps in the production simulations . To characterize the binding affinity of different systems , the free binding energies were calculated using the MMPBSA . py package [70] . 100 frames were sampled from the trajectories for analysis using MMPBSA . py . The entropy portion of the free energy was not considered in the calculation . In addition , the PME module in VMD was used to estimate the electrostatics potential of the modeled ComEA monomers ( Fig . 3C ) . ComEA , ComEAK62/63A , ComEA-mCherry , and ComEAK62/63A-mCherry ( all containing the eight amino acid Strep-tag II sequence at the C-terminus ) were purified as previously described [71] with minor modification . Briefly , E . coli cells containing the respective plasmids ( Table S1 ) were grown aerobically at 37°C until an OD600 of 1 . 0 . At that time expression was induced by the addition of 0 . 2% arabinose to the culture medium and the cells were further incubated for 2 hours before their harvest at 4°C and storage of the cell pellet at −80°C . The cells were lysed by sonication ( Vibra-cell; 10 min . in total with 30 sec on and 30 sec off intervals and an amplitude of 80% ) and the lysate was further processed as described [71] . Notably , after realization that the protein was pre-occupied by DNA ( see results section ) , we included a on-column DNase treatment step ( 10 µg/ml of DNase I ( Roche ) in 100 mM Tris/HCl pH 8 . 0 buffer containing 20 mM MgCl2 and 0 . 2 mM CaCl2; 30 min . at 30°C ) after the soluble protein fraction was loaded onto the streptactin resin and washed with 5 column volumes of washing buffer . The DNase I treatment step was followed by extensive washing of the column ( 10 to 30 volumes ) before the respective protein was eluted as described [71] . The eluted proteins were concentrated using Amicon Ultra spin columns ( with a MWCO of 3 kDa or 10 kDa; Millipore ) . For the AFM experiment , the protein was dialyzed against AFM buffer ( 5 mM Tris/HCl pH = 8 . 0 and 10 mM MgCl2 ) . The protein concentration was determined according to Bradford [72] . Electrophoretic Mobility Shift Assays were basically performed as previously explained [71] . However , as preliminary experiments indicated that neither the absence of DTT nor the storage of the protein in the absence of glycerol and at 4°C did change the results of the experiments , the protocol was changed accordingly . The 200 bp DNA fragment was PCR-amplified using gDNA of strain A1552 as template and represented the upstream region of the comEA gene . Other DNA fragments ( e . g . the aphA promoter region as previously tested [71] ) were similarly shifted ( data not shown ) . The protein/DNA mixture was incubated for 5 min at room temperature before electrophoretic separation on an 8% polyacrylamide gel . DNA was visualized by ethidium bromide staining [71] whereas the fusion proteins ( ComEA-mCherry and ComEAK62/63A-mCherry were detected using a Typhoon scanner ( GE Healthcare; excitation at 532 nm ( green ) and emission detected with a 610 BP30 ( red ) filter ) . To prepare the protein/DNA complex we mixed 0 . 85 ng/µl of a PCR-amplified DNA fragment ( 809 bp ) with the protein in the molecular ratios of 1∶2 . 5 and 1∶10 ( DNA∶protein ) in buffer containing 5 mM Tris/HCl pH 8 . 0 and 10 mM MgCl2 . After incubation for 10 min at 37°C , 15 µl of the mixture was deposited on freshly cleaved mica and rinsed thoroughly with ddH2O for two minutes . Preparation of the sample with bare DNA was done under the same conditions but in the absence of the protein . The AFM images were acquired in air and in tapping mode using an Asylum Research Cypher microscope . We used Olympus silicon cantilevers ( Olympus OMCL-AC240TS-R3 ) with a spring constant of 1 . 7 N/m and a resonant frequency of 70 kHz . The typical scan rate was 2 . 0 Hz . | Horizontal gene transfer ( HGT ) plays a key role in transferring genetic information from one organism to another . Natural competence for transformation is one of three modes of HGT used by bacteria to promote the uptake of free DNA from the surrounding . The human pathogen Vibrio cholerae enters such a competence state upon growth on chitinous surfaces , which represent its natural niche in the aquatic environment . Whereas we have gained a reasonable understanding on how the competence phenotype is regulated in V . cholerae we are only at the beginning of deciphering the mechanistic aspects of the DNA uptake process . In this study , we characterize the competence protein ComEA . We show that ComEA is transported into the periplasm of V . cholerae and that it is required for the uptake of DNA across the outer membrane . We demonstrate that ComEA aggregates around incoming DNA in vivo and that the binding of DNA is dependent on specific residues within a conserved helix-hairpin-helix motif . We propose a model indicating that the DNA uptake process across the outer membrane might be driven through ratcheting and entropic forces associated with ComEA binding . | [
"Abstract",
"Introduction",
"Results",
"and",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bacteriology",
"microbial",
"evolution",
"protein",
"structure",
"microbial",
"pathogens",
"microbiology",
"biology",
"computational",
"biology",
"macromolecular",
"structure",
"analysis"
] | 2014 | ComEA Is Essential for the Transfer of External DNA into the Periplasm in Naturally Transformable Vibrio cholerae Cells |
Cognitive and behavioral disorders are thought to be a result of neuronal dysfunction , but the underlying molecular defects remain largely unknown . An important signaling pathway involved in the regulation of neuronal function is the cyclic AMP/Protein kinase A pathway . We here show an essential role for coronin 1 , which is encoded in a genomic region associated with neurobehavioral dysfunction , in the modulation of cyclic AMP/PKA signaling . We found that coronin 1 is specifically expressed in excitatory but not inhibitory neurons and that coronin 1 deficiency results in loss of excitatory synapses and severe neurobehavioral disabilities , including reduced anxiety , social deficits , increased aggression , and learning defects . Electrophysiological analysis of excitatory synaptic transmission in amygdala revealed that coronin 1 was essential for cyclic–AMP–protein kinase A–dependent presynaptic plasticity . We further show that upon cell surface stimulation , coronin 1 interacted with the G protein subtype Gαs to stimulate the cAMP/PKA pathway . The absence of coronin 1 or expression of coronin 1 mutants unable to interact with Gαs resulted in a marked reduction in cAMP signaling . Strikingly , synaptic plasticity and behavioral defects of coronin 1–deficient mice were restored by in vivo infusion of a membrane-permeable cAMP analogue . Together these results identify coronin 1 as being important for cognition and behavior through its activity in promoting cAMP/PKA-dependent synaptic plasticity and may open novel avenues for the dissection of signal transduction pathways involved in neurobehavioral processes .
Behavioral and cognitive deficits comprise a heterogeneous collection of pathologies . Copy number variants and several single gene alterations predisposing to neurobehavioral and cognitive diseases have been identified and are believed to act either independently or in a combinatorial fashion [1] , [2] . The molecular functions of the candidate genes that are associated with cognitive and behavioral impairment are beginning to be elucidated [1]; several of these molecules were shown to be located at synapses , suggesting that synaptic dysfunction is involved in neurobehavioral disorders [3]–[6] . However , for many of the candidate genes a direct link with neurobehavioral disorders as well an understanding of their molecular function remains unknown [7] , [8] . An important neuronal signaling cascade involved in synaptic plasticity and learning occurs downstream of G protein–coupled receptors , resulting in the activation of adenylate cyclase that produces cAMP through stimulation with the Gαs subunit of trimeric G proteins [9]–[12] . cAMP in turn activates protein kinase A ( PKA ) , which drives long-term changes in synaptic efficacy through direct effects on the pre- or postsynapse and through CREB-dependent regulation of gene transcription [13]–[15] . However , the mechanisms regulating cAMP production remain incompletely understood . In this article we describe a crucial role for the conserved WD repeat protein coronin 1 in cognition and behavior through the activity of coronin 1 in modulating the cAMP/PKA pathway . Coronin 1 is encoded in a genomic region on chromosome 16 in human and the corresponding region of mouse chromosome 7 whose copy number variations are associated with varying degrees of cognitive impairment [16]–[19] . Coronin 1 is a member of the WD repeat containing protein family , and is expressed in immune cells as well as in nervous tissue [20]–[22] . In immune cells , coronin 1 is required for the transduction of cell surface signals to intracellular signaling cascades , thereby regulating a number of different processes , ranging from pathogen destruction to the survival of T cells [22]–[26] . A function for coronin 1 in neuronal cells or tissues has , however , not been described . We here show that mice lacking coronin 1 displayed increased aggression , social deficits , increased repetitive behavior , reduced fear/anxiety , and a severe defect in learning and memory . We found that coronin 1 was specifically expressed in excitatory synapses and required for cAMP/PKA-dependent synaptic plasticity . We further show that upon cellular stimulation , coronin 1 interacts with the G protein subunit Gαs to stimulate the cAMP/PKA pathway . Importantly , infusion of the cAMP analogue 8-Br-cAMP reversed the learning and memory deficits in coronin 1–deficient mice . These results identify coronin 1 as being important for cognition and behavior through its activity in modulating cAMP/PKA-dependent synaptic plasticity .
In the course of analyzing mice harboring a targeted deletion of the coronin 1 gene [23] , [26] we consistently observed behavioral abnormalities . In particular , coronin 1–deficient mice showed a significantly enhanced aggressive behavior , shorter attack latency , and social dominance compared to wild-type mice ( Figure 1A , B and Table S1 ) . Also , upon transfer to a dark versus light chamber , mice lacking coronin 1 showed an increased duration of stay in the light compartment relative to wild-type as well as an increased number of transitions between the dark and light chambers and a reduced latency to enter the light compartment ( Figure 1C and Figure S1A ) . Analysis using the elevated plus maze test revealed that coronin 1–deficient mice spent a significantly longer duration of time in the open arm of the maze ( Figure 1D and Figure S1B ) . Furthermore , analysis of vocalization [27] revealed a reduction in calls in coronin 1–deficient mice compared to wild-type animals ( Figure 1E ) . Finally , we analyzed self-grooming , which in mice is assumed to be an equivalent to stereotypic behavior observed in models of behavioral abnormalities [28] , [29] . As shown in Figure 1F , coronin 1–deficient mice showed a drastically increased self-grooming . Together these results suggest that the absence of coronin 1 results in reduced anxiety and increased aggression . Analysis of classical cued fear conditioning [30] showed that coronin 1–deficient mice exhibited markedly reduced freezing levels as compared to wild-type animals ( Figure 1G and Figure S1D , E ) . Likewise , contextual fear conditioning , induced by exposing mice to a single unconditioned stimulus ( US ) in a neutral environment , was strongly impaired in the absence of coronin 1 ( Figure 1H ) , although no defects in locomotion or pain thresholds nor perception were observed as assessed by Y-maze , grip test , rotarod , beam walking , open field , formalin pain sensitivity , hot plate , and tail flick analysis ( Figure S1C , E , F and Table S1 ) , suggesting that the behavioral defects are not the result of low exploratory activity and/or impaired sensory abilities . Also , analysis of novel object recognition did not reveal any difference between wild-type and coronin 1–deficient mice ( Figure S1G ) . Next , we monitored the degree of socialization of both wild-type and coronin 1–deficient mice using the three-chamber socialization test [27] , [28] . Wild-type and coronin 1–deficient mice showed similar olfactory abilities ( Figure S2A ) and did not display any side preferences or differences in the number of entries into the chambers in the absence of a companion mouse ( Figure S2B–E ) . However , coronin 1–deficient mice showed a significantly reduced sociability as well as reduced preference for social novelty ( Figure 1I , J ) . A similar social defect was observed using a modified Paylor's partition analysis ( Figure S2F ) , which , together with the three-chamber analysis , suggests defective social interaction upon coronin 1 deletion . To analyze the consequences of coronin 1 depletion in human , we evaluated an individual from a consanguineous family presenting with immunodeficiency due to a homozygous missense mutation in the coronin 1 coding region , causing a valine to methionine change at the conserved amino acid position 134 ( Figure S3A ) that resulted in coronin 1 deficiency [31] . To rule out gene duplication or deletion of the 16p11 . 2 fragment , which is known to be associated with cognitive and behavioral dysfunction , we performed conventional karyotyping and high-resolution array CGH analysis that did not reveal genetic alterations ( see Materials and Methods ) . Furthermore whole exome sequencing did not show any homozygous mutation in another gene that may be linked to neurocognitive impairment . Bioinformatic structural predictions revealed that this valine is located at a critical position in the interface between two blades [32] , [33] , tightly packed against Val106 , His130 , and Ser149 . As a consequence the coronin 1V134M mutant was rapidly degraded ( Figure S3B , C ) . The neurological examination of the patient was normal , presenting no microcephaly , hypertonia , spasticity , or anomaly of cranial nerves and no facial dysmorphic features . Broad biological , metabolic , and medical screenings in order to rule out other causes for developmental delay were performed , with no abnormalities being diagnosed ( see Text S1 and Table S2 ) . However , neuropsychological examination revealed that although the patient acquired gross motor milestones timely ( crawling at 8 months and free walking at 12 months , respectively ) , he presented a severe delay in language acquisition as well as significant behavioral anomalies including increased levels of aggression ( scratching and biting ) , attention/concentration deficit , hyperactivity , impulsivity , and sleep disturbances ( Figure S3D , E ) . Furthermore , in intelligence and achievement tests , the patient presented a significantly lower performance than his age-matched controls ( Figure S3F ) . Together these observations indicate that loss of coronin 1 results in severe cognitive and behavioral dysfunction . The above results suggest a function for coronin 1 in cognition and behavior . Analysis of neuronal distribution of coronin 1 by immunohistochemistry and immunoblotting revealed expression throughout different brain regions , including expression in cortex , basolateral amygdala , as well as olfactory bulb , hippocampus , cerebellar molecular layer , with minimal expression in the thalamus ( Figure 2A–D and Figure S4 ) . Notably , coronin 1 expression was found to be mutually exclusive with the expression of GAD67 , a marker for GABAergic inhibitory neurons [34] , and was predominantly found in neurons expressing the glutamate transporter vGLUT1 specific to excitatory neurons ( Figure 2E ) [35] . Consistent with the expression pattern found in vivo , in hippocampal neuronal cultures , coronin 1 colocalized with vGLUT1 but not with vGAT , markers of excitatory and inhibitory presynapses , respectively ( Figure 2F ) . Coronin 1 was highly enriched in synapses , as judged by the extensive colocalization with synapsin 1 ( Figure 2G ) [36] . Together , these results suggest that coronin 1 is enriched at excitatory synapses . Cognitive and behavioral dysfunction has been linked to an imbalance of excitatory and inhibitory synapses [6] , [37] . To analyze the consequences of coronin 1 deletion for the E/I synapse ratio , excitatory and inhibitory synapses in wild-type and coronin 1–deficient hippocampi were examined by serial block face scanning electron microscopy ( see Materials and Methods ) . Strikingly , although the number of inhibitory synapses was not affected by coronin 1 deletion , the number of excitatory synapses was significantly reduced ( Figure 3A , B ) , similar to the reduction of vGLUT1 but not vGAT synapses from in vitro cultured neurons ( Figure 3C ) . In addition , electrophysiological analysis showed a significant reduction of the E/I miniature events frequency ratio in the absence of coronin 1 ( Figure S5 ) . To assess the functional role of coronin 1 at excitatory synapses in anxiety-related brain regions , whole-cell current and voltage-clamp recordings were performed from lateral amygdala ( LA ) projection neurons , while stimulating excitatory pathways from cortex and thalamus , the two main sensory inputs onto LA principal neurons ( Figure 3D ) . Baseline synaptic transmission as well as postsynaptic long-term potentiation ( LTP ) at thalamo-LA synapses [38] was found to be largely normal in the absence of coronin 1 ( Figure 3E and Figure S6 ) . However , PKA-dependent presynaptic cortico-LA LTP [39] was completely absent in coronin 1–deficient animals ( Figure 3F ) . Next , to bypass upstream induction mechanisms , we applied forskolin , which directly increases the presynaptic probability of release at cortico-LA synapses [39] . Although in wild-type mice application of a brief pulse of forskolin persistently increased excitatory postsynaptic potential ( EPSC ) amplitude and decreased the paired pulsed ratio ( PPR ) , forskolin had no effect on synaptic transmission in coronin 1–deficient mice ( Figure 3G ) . We therefore conclude that the absence of cortico-LA LTP in coronin 1–deficient amygdala results from defective cAMP/PKA signaling . In neurons , as in other cells , PKA activation leads to the phosphorylation of the cAMP response element binding protein ( CREB ) at serine 133 , thereby activating CREB , which is required for CREB-mediated gene transcription that is involved in the regulation of cognitive functioning , memory consolidation , and the late phase of postsynaptic LTP [14] , [15] , [40] . Staining of wild-type and coronin 1–deficient hippocampal neurons with antibodies directed against cAMP , a PKA consensus site , or phospho-CREB revealed strongly reduced labeling in coronin 1–deficient neurons ( Figure 4A ) . Furthermore , analysis of brain lysates revealed significantly increased CREB phosphorylation at serine-133 only in the presence of coronin 1 ( Figure 4B ) . Similarly , in brain sections , P-CREB labeling was severely reduced in neurons from coronin 1–deficient animals compared to wild-type ( Figure 4C ) . Consistent with reduced cAMP levels in coronin 1–deficient mouse brain ( Figure 4D ) , direct measurement of PKA activity in amygdala lysates showed a significantly reduced activity in the absence of coronin 1 compared to wild-type animals ( Figure 4E ) . Inclusion of cAMP during the assay resulted in an elevation of the PKA activity in coronin 1–deficient brain sections comparable to that in wild-type sections , indicating that PKA is fully functional in the absence of coronin 1 . A decrease in neuronal cAMP levels was recently found to be associated with increased ventricular size [41] . Interestingly , when analyzing brains from wild-type and coronin 1–deficient mice using in vivo magnetic resonance imaging ( MRI ) , we observed that deletion of coronin 1 resulted in a significant enlargement of the lateral ventricles in adult but not in young mice ( Figure 4F , G and Figure S7A , B ) , corroborating a role for neuronal coronin 1 in the modulation of cAMP signaling . Ventricle enlargement was associated with a reduction in the cell numbers in the hippocampus , whereas basic histopathology of brain sections did not reveal any other obvious differences ( Figure S7A–C and unpublished data ) . Together these results suggest an essential role for coronin 1 in activating cAMP-dependent signaling , thereby impacting memory and behavior via modulating synaptic plasticity . To address the molecular mechanism underlying the coronin 1–dependent activation of cAMP/PKA signaling , we analyzed stimulus-dependent cAMP production in a coronin 1–negative cell line expressing or lacking coronin 1 ( Mel JuSo cells , see [20] ) . When these cells were left untreated or stimulated with isoproterenol , cAMP production , as analyzed using a competitive antibody assay , was drastically increased by the expression of coronin 1 ( Figure 5A ) . Expression or deletion of coronin 1 did not influence β2-adrenergic receptor levels at the cell surface , nor did it modulate the expression of other molecules involved in cAMP production , including adenylate cyclases , β-adrenergic receptor , Gαs , and Gαi ( see Figure S8 ) . To analyze coronin 1–dependent cAMP production in real time , we used a live cell cAMP reporter based on the activity of the exchange protein activated by cAMP 1 ( Epac1 ) , a guanine exchange factor for the small GTPase Rap1 ( ICUE3 ) [42] , [43] . Upon cAMP binding , the Epac1 fusion protein undergoes a conformational change that results in a decrease in fluorescence resonance energy transfer ( FRET ) efficiency , visualized as an increase in the CFP∶YFP ratio . Stimulation of coronin 1–expressing cells resulted in a ratiometric increase ( CFP∶YFP ) that originated at the plasma membrane , rapidly increasing and spreading over the entire cell body ( Figure 5B , C and Movie S1 ) . In contrast , in coronin 1–negative cells , a severe reduction in the ratiometric increase was observed ( Figure 5B , C and Movie S2 ) . Together these results corroborate the role for coronin 1 in the activation of cAMP signaling . Coronin 1 colocalizes with the F-actin cytoskeleton in vitro as well as within cells . To directly test whether the coronin 1–mediated cAMP production was dependent on the F-actin cytoskeleton , cells expressing or lacking coronin 1 were treated with the F-actin depolymerizing drug latrunculin B [44] for 30 min , followed by stimulation with isoproterenol and analyzed by FRET . Although latrunculin B fully depolymerized the actin cytoskeleton , as evidenced by the absence of any F-actin staining ( Figure 5E ) , the presence of latrunculin B did not affect coronin 1–dependent cAMP production as analyzed by FRET ( Figure 5D , F and Movie S3 and Movie S4 ) . Together these results suggest that coronin 1 promotes cAMP production in an F-actin–independent manner . The above results strongly suggest that upon stimulation with isoproterenol , coronin 1 functions in enhancing the production of cAMP that results from β2-adrenergic receptor stimulation . To analyze whether the coronin 1–dependent cAMP production resulted from a physical interaction between coronin 1 and Gαs , cells expressing coronin 1 were activated with isoproterenol , lysed , and Gα complexes immunoprecipitated using specific antibodies . Protein complexes were separated by SDS-PAGE and immunoblotted using either anti-Gα or anti–coronin 1 antibodies . Although only low to undetectable amounts of coronin 1 were observed in anti-Gα immunoprecipitates from unstimulated cells , upon isoproterenol stimulation the coronin 1 signal drastically increased ( Figure 6A , B ) . In accordance with the capacity of coronin 1 to promote cAMP production in an F-actin–independent manner , depolymerization of F-actin by latrunculin B did not affect the stimulus-dependent interaction of coronin 1 with Gα ( Figure 6C ) . Similarly , coronin 1 was co-immunoprecipitated with Gα in cell lysates from stimulated N1E-115 neuronal cells , whereas low to undetectable amounts of coronin 1 were co-immunoprecipitated with Gα in unstimulated cells ( Figure S9 ) . To analyze whether co-immunoprecipitated Gα molecules represented Gαs , as suggested by the lack of inhibition by pertussis toxin ( Figure S10 ) , coronin 1 was immunoprecipitated from untreated or isoproterenol-stimulated N1E-115 cells , and immunoblotted for either Gαs or Gαi . As shown in Figure 6D , although Gαs was readily detected in coronin 1 immunoprecipitates following isoproterenol stimulation , Gαi could not be detected . Furthermore , addition of cholera toxin , which permanently ADP ribosylates and thereby constitutively activates Gαs , induced an equivalent cAMP production in both coronin 1–expressing and coronin 1–deficient cells ( Figure 6E ) . Next , we investigated the possibility that the previously shown results could arise from a direct interaction between coronin 1 and Gα . Because the crystal structure of coronin 1 is known [33] , we generated a series of coronin 1 mutant molecules in which residues at the surface of coronin 1 ( and therefore potentially involved in protein–protein interactions within a complex ) were mutated to alanine ( see Table S3 ) . Coronin 1 mutant molecules were expressed in Mel JuSo cells , and the capacity to stimulate cAMP production following stimulation of the β2-adrenergic receptor with isoproterenol was analyzed by cAMP ELISA . As shown in Figure 7A , mutation of Lys-20 , Arg-69 , Glu-102 alone or in combination with Lys-355 strongly decreased cAMP production . Importantly , these coronin 1 mutants , which were stably expressed and correctly localized ( Figure 7B lower panels and Figure S11 ) , were unable to associate with Gα ( Figure 7B ) . In addition , the role of the mutated residues in the interaction with Gαs was assessed by surface plasmon resonance experiments between purified immobilized Gαs and either purified wild-type coronin 1 or the purified coronin 1 alanine quadruple mutant ( coronin 1K20A/R69A/K355A/E102A ) . Strikingly , the interaction of the coronin 1 quadruple mutant with Gαs was drastically decreased as compared to the wild-type coronin 1 ( Figure 7C ) . Together these data suggest that coronin 1 interacts with Gαs to promote cAMP production in a stimulus-dependent manner . Together the data thus far reveal a role for coronin 1 in learning and memory via promoting cAMP-dependent plasticity through the association with Gαs . We next wanted to directly test whether the defective cAMP/PKA signaling associated with coronin 1 deficiency was responsible for the lack of fear consolidation as revealed by the low freezing levels observed during the recall tests ( Figure 1G , H ) . To do so , we tested whether some of the behavioral deficits observed in coronin 1–deficient mice could be rescued by timely applications of the membrane-permeable nonhydrolysable cAMP analogue 8-Br-cAMP [45] within the amygdala ( Figure 8 ) . To that end , 8-Br-cAMP was bilaterally infused in the amygdala of wild-type and coronin 1–deficient animals via chronically implanted cannula ( Figure 8A , B ) . To correct presynaptic plasticity during CS/US association , cAMP injections were performed 30∼45 min prior to fear acquisition in coronin 1–deficient animals . When tested in vitro , such 8-Br-cAMP pre-incubations partially rescued the absence of plasticity at cortico-LA synapses in coronin 1–deficient slices ( Figure S12 ) . Infusion of 8-Br-cAMP in amygdala failed to normalize fear reactions upon contextual re-exposure in coronin 1–deficient mice ( Figure 8C “single injection” ) . As such a defect might be due to a lack of CREB activation in the hours following neuronal activation [46] , we performed an additional cAMP infusion 2 . 5∼3 h after the fear acquisition . Importantly , this second drug application was able to restore the contextual fear recall defect in coronin 1–deficient animals , which became indistinguishable from similarly treated wild-type animals ( Figure 8C “double injection” ) . Together with the rescued LTP in coronin 1–deficient slices by application of 8-Br-cAMP , these pharmacological results obtained in vivo strongly support a role for coronin 1 in learning and memory through the activation of cAMP-dependent synaptic plasticity .
Cognitive and behavioral disorders can arise as a result of defective synaptic plasticity , but the underlying molecular mechanisms remain incompletely understood . The work presented here suggests an important role for neuronal coronin 1 in behavioral and cognitive function via regulation of cAMP/PKA-dependent signal transduction and synaptic plasticity . We found that coronin 1 deletion or expression of an unstable coronin 1 mutant is associated with severe neurobehavioral dysfunction in both mice and humans . Mice lacking coronin 1 displayed defective socialization , enhanced grooming , defective vocalization , as well as lowered anxiety and enhanced aggression . Interestingly , the normal object recognition response together with the altered fear conditioning suggests that the defective learning and memory in the absence of coronin 1 may be a result of altered anxiety [47] . One should also note that the lower level of freezing could be a consequence of altered exploratory behavior and/or locomotor traits; However , the Y-maze analysis along with the open field and beam walking suggests that coronin 1 deficiency per se does not result in lowered exploratory activity , locomotor defects , and/or novelty seeking that are potential confounds in these behavioural tests [48] . Furthermore , coronin 1 deletion was associated with a reduction in the number of excitatory synapses as well as a virtual absence of PKA-dependent LTP at cortico-lateral amygdalary synapses , structures that are important for learning and memory [39] , [49] . Together these results suggest an important function for coronin 1 in activating PKA-dependent signaling at excitatory synapses . The specific role for coronin 1 in promoting anxiety-related synaptic plasticity via activation of cAMP/PKA signaling is illustrated by several of the findings presented here . First , the increase in ventricular size in mice lacking coronin 1 phenocopies mice in which neuronal cAMP levels are constitutively decreased [41] . Second , the membrane-permeable cAMP analogue 8-Br-cAMP was able to rescue cortico-LA long-term plasticity . Third , intra-amygdala infusion of 8-Br-cAMP resulted in a virtually complete restoration of the memory defect associated with coronin 1 deletion . Although application of 8-Br-cAMP might activate LTP in all the cells of the injection area , regardless of whether they were triggered or not during the fear conditioning , the results shown suggest that in the presence of extraneous 8-Br-cAMP the circuit specific for this fear conditioning is strengthened upon exposure to the context , thereby bypassing the cAMP/PKA defect because of coronin 1 deficiency . Furthermore , the behavioral defects are unlikely to be a consequence of developmental defects or neurodegeneration , as they occur well before the observed ventricular enlargement and neuronal loss , although more subtle morphological changes that are not detected by the methods used here cannot be ruled out . Finally , the coronin 1 neurobehavioral defects phenocopy cAMP signaling defects [41] , [50]–[52] . Together these results strongly argue for a specific role for coronin 1 in the activation of cAMP signaling . We found that the mechanism via which coronin 1 promotes cAMP production occurs via the association of coronin 1 with Gαs molecules: first , inclusion of pertussis toxin , which inactivates Gαi via ADP ribosylation , did not prevent coronin 1–dependent cAMP production . Second , coronin 1 was readily co-immunoprecipitated with Gαs but not with Gαi . Third , addition of cholera toxin , which locks Gαs proteins in their active state via ADP ribosylation , resulted in a level of cAMP production in coronin 1–deficient cells that was equivalent to the cAMP production in coronin 1–expressing cells . Therefore , neuronal coronin 1 may have evolved to specifically activate the adenylate cyclase pathway , as these enzymes are the predominant targets of Gαs proteins [53] . Possibly , exon shuffling during evolution might have endowed coronin 1 with sequence motifs that allow its binding to Gαs in order to promote adenylate cyclase activation , which , given the tissue-specific expression of coronin 1 , occurs in a cell-type-specific manner . The exact molecular mechanism by which coronin 1 modulates Gαs activity remains to be elucidated . Both coronin 1 and the Gβ subunit of the G protein heterotrimer have a seven-bladed propeller fold , sharing sequence and structural homology [33] . Also , the residues of coronin 1 that we have found to be necessary for Gαs interaction are located in a region that is homologous to the Gβ–Gαs interacting surface [54] . Thus , one possibility is that coronin 1 binds to Gαs in a similar manner as Gβ . Interestingly , several lines of evidence suggest that following receptor activation , the components of the G protein heterotrimer rearrange rather than dissociate and are maintained in a complex with the βγ subunit [55]–[57] . In this scenario , in coronin 1–expressing cells , upon stimulation coronin 1 might displace the Gβ subunit from the trimeric complex in order to modulate adenylate cyclase activity , similarly to the activation of downstream effectors by heterotrimeric G protein complexes [58]–[60] . Alternatively , coronin 1 may bind to Gαs in a manner that is distinct from Gβ , thereby promoting cAMP production ( see also Figure S13 ) . In vitro reconstitution of coronin 1–dependent adenylate cyclase activation might help to further delineate the molecular details of the role for coronin 1 in stimulating cAMP production . Interestingly , the here described activation of cAMP/PKA signaling occurred independent of an intact F-actin cytoskeleton . These findings are consistent with earlier work using lymphoma cells , where disruption of the F-actin cytoskeleton by cytochalasin B also failed to modulate cAMP production within the time frame analyzed here [61] , [62] . Although relatively little is known about the regulation of G protein–mediated adenylate cyclase activation by the F-actin cytoskeleton , these data reinforce the conclusion that , in vivo , coronin 1 is unlikely to mediate its function in regulating signal transduction via inducing F-actin rearrangement as shown in leukocytes [63] . However , whether or not coronin 1–mediated IP3/calcium signaling in naïve T cells and macrophages [23] , [26] , [64] occurs through the here described activation of adenylate cyclase remains to be established; although there are indications that both T cell activation as well as macrophage receptor triggering may involve G protein–coupled signaling [65] , [66] , cAMP mediated signaling in leukocytes is still relatively poorly understood . Also , recent work showed that upon deletion of the coronin 1 homologue in the lower eukaryote Dictyostelium , coronin A , the resulting developmental defect can be fully restored upon direct activation of PKA via 8-Br-cAMP incubation ( Vinet et al . , in press ) . Given the importance of regulated cAMP signaling for appropriate neuronal functioning [13] , [14] , [67] , [68] , the here described role for coronin 1 in the activation of PKA/cAMP-dependent synaptic plasticity is fully consistent with neuronal coronin 1 being involved in the activation of cAMP production . Furthermore , the restricted expression of coronin 1 within the brain as well as to excitatory neurons may allow a subset of neuronal structures to depend for the activation of cAMP/PKA-mediated synaptic plasticity on coronin 1 , while maintaining coronin 1–independent regulation of neuronal activity in other regions . Interestingly , the coronin 1 gene is located in a genomic region on chromosome 16 in human and 7 in mouse , whose copy number variations are associated with a wide range of neuropsychiatric conditions including behavioral dysfunction and intellectual disability [16] , [19] , [29] . Both the mouse model as well as the patient described here exclusively lack coronin 1 expression , instead of having a deletion across a larger genomic region , which suggests a strong correlation of coronin 1 deletion and/or mutation with severe neurobehavioral abnormalities such as social deficits , stereotypic behavior , aggression , and cognitive impairment . A delay in language acquisition and behavioral as well as cognitive impairment is also described for an individual that paired a de novo deletion of 16p11 . 2 on the maternal allele with a 2 bp deletion in the paternal coronin 1 allele [69] . In the patient described here , the hypomorphic nature of the coronin 1 mutation may explain the differences in the neurobehavioral phenotype when compared with the mouse knock-out . With additional individuals being identified harboring coronin 1 mutations associated with cognitive deficits , it may be feasible to dissect the molecular details of the coronin 1–dependent signal transduction pathway ( s ) responsible for behavior and cognition . The development of neurobehavioral and cognitive disorders is believed to result from an imbalance between inhibitory and excitatory synaptic transmission [6] , [37] . Interestingly , in several models of behavioral abnormalities the E/I synapse ratio was found to be altered [70] , [71] consistent with the selective expression of neuronal coronin 1 at excitatory synapses and the disturbed E/I synapse ratio in the absence of coronin 1 . In conclusion , the results presented here define coronin 1 as a regulator of behavioral processes via promoting PKA-dependent synaptic plasticity and may open novel avenues for the dissection of signal transduction pathways involved in cognitive processes .
Coronin 1–deficient and wild-type mice were described before [23] , [26] and were used from backcross 8 . All animal experimentation was approved by the veterinary office of the Canton of Basel-Stadt ( approved license number 1893 and 2336 ) and performed according to local guidelines ( Tierschutz-Verordnung , Basel-Stadt ) and the Swiss animal protection law ( Tierschutz-Gesetz ) . Hippocampal neurons were prepared as described [72] and used between day 7 and day 11 . Mel JuSo cells and Mel JuSo cells transfected with coronin 1 cDNA have been described [20] , [73] . H19/7 ( ATCC: CRL-2526 ) and N1E-115 ( ATCC-CRL-2263 ) cells were from ATCC . H19/7 cells were cultured in DMEM supplemented with 10% fetal bovine serum , 0 . 2 mg/ml G418 , and 0 . 001 mg/ml puromycin in a 34°C incubator with 5% CO2 . N1E-115 cells were cultured in DMEM supplemented with 10% fetal bovine serum at 37°C and 5% CO2 . Isoproterenol and cholera toxin were from Sigma and roscovitine from Cayman or Sigma chemicals . Antibodies used were from the following sources: rabbit anti-coronin 1 has been described before [20] , [32] , mouse anti-coronin 1 was from Abnova , mouse anti-neurofilament medium chain was from Sigma , mouse anti-GAD67 was from Chemicon , mouse anti–synapsin 1 was from BD Sciences , anti-tubulin ( E7 ) was developed by M . Klymkowsky and obtained from the Developmental Studies Hybridoma Bank at the University of Iowa , anti-cAMP from Santa-Cruz Biotechnologies , anti-PKA substrate antibody from Cell Signaling , anti-P-CREB antibodies were from Abcam , and mouse anti-VGAT and mouse anti-Vglut1 antibodies were from Synaptic Systems . Mouse anti-histone antibody was from Santa Cruz Biotechnology and DRAQ5 was from Biostatus . Forskolin was from Tocris , and Rolipram was from Sigma . Neurotrace red and Dapi ( brain stain kit ) were from Life Technologies . Rolipram was from Sigma . The ICUE3 plasmids were kindly obtained from Dr . Jing Zhang ( Johns Hopkins University ) . For cAMP ELISA , cells were seeded at a density of 2×106 cells per well on a six-well plate and allowed to adhere overnight and in the case of Mel JuSo cells transfected with WT Cor 1 HA pCB6 or control plasmid ( pMax GFP or pCB6 ) and used after 48 h . For live cell imaging , coronin 1–expressing or wild-type Mel JuSo cells [20] were seeded at a density of 20 , 000 cells per well on an eight-well chambered coverslip ( 0 . 11 mm ) and cultured overnight . For mutagenesis and cell transfection , see Text S1 “Materials and Methods . ” Data shown in Figure S1C , F , G , were performed according to established and standardized protocols at the Institut Clinique de la Souris , Strasbourg , France . Twelve wild-type and 12 coronin 1–deficient male mice aged 15–20 wk were used for this study . Mice were allowed to acclimatize for 2 wk prior to analysis using the following tests: For Y-maze spontaneous alternation , each mouse ( males , 17–18 wk ) was placed at the end of one arm of the Y-maze , the head directed to the walls , and allowed to explore freely the apparatus for 5 min , with the experimenter out of the animal's sight . Alternations are operationally defined as successive entries into each of the three arms as on overlapping triplet sets . Percent spontaneous alternation performance was defined as the ratio of actual ( total alternations ) to possible alternations ( total arm entries , 2 ) ×100 . Total arm entries and the latency to exit the starting arm were also scored as an index of ambulatory activity and emotionality in the Y-maze , respectively ( n = 12 per group ) . For dark versus light preference , a rectangular , black Plexiglas box of dimension 80 cm×30 cm divided into two compartments with one compartment well lit and the other kept dark [5] , [74] was used ( see Text S1 “Materials and Methods” for analysis ) . For grooming analysis , a rectangular , black Plexiglas box of dimension 80 cm×30 cm was used that was divided into two equal compartments using a transparent Plexiglas partition placed in the middle [27] , [28] . Both the compartments were covered with a red Plexiglas lid to avoid any external interference . For analysis , see Text S1 “Materials and Methods . ” Resident-intruder test was performed as described [75] , [76] and further detailed in Text S1 “Materials and Methods . ” For analysis of social dominance using the tube displacement test , mice ( males , 14–26 wk ) were tested using a plastic tube [30 cm long , 3 . 5 cm Ø and cleft at the top ( 1 . 5 cm ) ] with their openings closed with a carton plug and connected to side chambers [10 cm×6 cm×10 cm ( LxBxH ) ] [76]–[78] . Analysis was carried out as described in Text S1 “Materials and Methods . ” Regarding the elevated plus maze test , this is a conflict test based on a natural tendency of mice to actively explore a new environment , versus the aversive properties of an elevated open runway [27] , [79] ( see Text S1 “Materials and Methods” ) . For the olfactory habituation-dishabituation analysis to assess olfactory function [5] , mice ( wild-type or coronin 1–deficient , males , 13–20 wk ) were shifted and acclimatized to the test room at least 30 min prior to the analysis ( see Text S1 “Materials and Methods” ) . The three chamber social interaction assay conducted as described [79] using an apparatus consisting of a polycarbonate box with removable partitions separating the box into three chambers ( 40 . 5×60 . 0×22 . 0 cm; Noldus Information Technology , the Netherlands ) . The partitions had openings that allowed the animal to move freely from one chamber to another . Both the side chambers contained one stranger cage ( steel wire mesh cage , 11 cm diameter , individual wire mesh separated by 1 . 1 . cm , height 20 cm ) with a Plexi glass lid . For details on the analysis , see Text S1 “Materials and Methods . ” Vocalization was analyzed using a rectangular , black Plexiglas box of dimension 80 cm×30 cm divided into two equal compartments using a transparent perforated Plexiglas partition placed in the middle ( multiple perforations , diameter 1 cm ) ( see Text S1 “Materials and Methods” ) . For fear conditioning , acquisition and retrieval of cued fear conditioning took place in two different contexts ( context A and B ) , and analysis was performed as described in Text S1 “Materials and Methods . ” Microdissected tissue from wild-type and coronin 1–deficient mice ( age 6–8 wk ) were lysed in Triton-X 100 buffer ( 20 mM HEPES pH7 . 4 , 100 mM NaCl , 5 mM MgCl2 , 1% Triton X-100 ) with 0 . 2% SDS containing protease and phosphatase inhibitors ( Roche ) at 4°C . After protein determination ( BCA , Pierce ) , equal amounts of protein ( 15 µg per lane ) were separated by 12 . 5% SDS-PAGE , transferred onto nitrocellulose , and probed using the mentioned antibody followed by HRP-labeled secondary antibodies and developed using an enhanced chemi-luminescence imager ( Fuji ) as described before [73] . In brief , hippocampal neurons ( 7 d cultures ) were washed with PBS followed by fixation with PFA ( 4% in Dulbecco's PBS ) immediately at room temperature for 20 min , followed by washing and saponin permeabilization ( 15 min 0 . 5% ) at room temperature . After 30 min blocking with 2% BSA in PBS , the cells were incubated with primary antibodies ( 1∶1 , 000 , 1 h RT ) of rabbit anti–coronin 1 and mouse anti–synapsin 1 or mouse anti–coronin 1 followed by Alexa Fluor 488 or −568 labeled secondary antibody incubation at room temperature for 1 h , respectively . Preparations were analyzed on a Zeiss LSM 510 Meta Confocal Laser Scanning microscope . Wild-type and coronin 1–deficient mice ( age 6–8 wk ) were anesthetized in a 4% isoflurane chamber , decapitated , and the brains dissected out immediately and transferred stepwise into 2-methylbutane solution supercooled to −40°C . Once frozen , the sections were transferred to −80°C until sectioning . Prior to sectioning , the brains were warmed to −20°C for 30 min and mounted on a cryostat ( Microm HM 560 or Leica CM1950 ) . Thin brain sections ( 12 micron ) were transferred onto superfrost glass slides ( Thermo Scientific ) , air dried , fixed with 4% paraformaldehyde for 10 min at room temperature followed by a cycle of dehydration with 70% , 95% , and 100% ethanol and rehydration in the reverse order . Alternatively , brain sections from C57BL/6 or CD1 mice ( Zyagen ) were used for staining . The samples were then subjected to antigen retrieval in 10 mM citrate phosphate buffer ( 10 mM Citric acid pH6 , 0 . 05% Tween 20 ) overnight at 60°C followed by blocking with 5% FBS for 2 h . For incubation with primary antibodies , rabbit anti–coronin 1 primary antibody ( serum 1002 , 1∶1 , 000 dilution ) or rat anti–coronin 1 serum ( 1∶100 ) along with mouse anti-neurofilament medium chain or chicken anti-neurofilament heavy chain antibody , rabbit anti-pCREB , and mouse anti-histone antibody in PBS with 2% fetal bovine serum were layered over the samples for 4 to 16 h . After three washes with PBS containing 2% fetal bovine serum , anti-rabbit Alexa Fluor 488 ( 1∶200 ) or anti-rat Alexa Fluor 488 ( 1∶200 ) and anti-mouse-Alexa Fluor 546 or anti-chicken Alexa Fluor 563 tagged secondary antibodies ( 1∶200 ) were added . One hour postincubation , the slides were washed three times and the slides sealed with a coverslip in the presence of antifade ( BioRad ) and viewed under a confocal microscope ( Carl Zeiss LSM 510Meta or LSM 700 ) . Quantitation was carried out from captured images of brain sections and analyzed using ImageJ ( NIH ) software . Hippocampal tissue was processed according to Knott et al . ( 2002 ) [80] . In brief , wild-type or coronin 1–deficient mice ( n = 3 ) were transcardially perfused with 2% paraformaldehyde and 2% glutaraldehyde . Then the brain was removed and cut with a vibratome ( Leica ) . Sections containing the CA1 region of hippocampus were chosen , stained in 1% osmium , and 1% uranyl acetate before being dehydrated in a successively increased concentration of ethanol . Sections were then flat-embedded in the Epon ( Fluka ) resin between two glass slides . The material was subsequently processed for ultrastructural analysis using the 3view microtome ( GATAN , UK ) inserted in a QUANTA 200 VP-FEG scanning electron microscope ( FEI , Netherlands ) as described [81] . In brief , volumes of 500 cubic microns located in the CA1 area for each animal were imaged , and synapses were classified as excitatory or inhibitory according to established criteria [80] and based on their asymmetric ( excitatory ) or symmetric ( inhibitory ) aspects . Wild-type or coronin 1–deficient pups ( P0 to P2 ) were sacrificed and their brain hippocampal regions microdissected and rapidly processed further by trypsinization and tituration for single cell preparation . The cells were initially plated in plating medium ( DMEM supplemented with fetal bovine serum 10% , glutamax 1% , penicillin 100 units/ml , streptomycin 100 µg/ml , and glucose 1 . 25% ) for 2–6 h and subsequently in maintenance medium ( Neurobasal supplemented with B27 2% , glutamax 1% , penicillin 100 units/ml , and streptomycin 100 µg/ml ) . The cultures were treated with cytarabine ( 10 µM , Sigma ) from day 4 of the culture and used between day 7 to 14 for cAMP determination . For cAMP analysis , the cells were cultured in plain neurobasal medium for 2 h in the presence of the cAMP-specific phosphodiesterase inhibitor rolipram ( 100 µM , Sigma ) . Subsequently , cells were either left unstimulated or stimulated as indicated in the presence of rolipram and lysed with the provided cAMP ELISA lysis buffer ( cAMP parameter kit , R&D Systems ) . The lysate was centrifuged at 9400×g for 10 min/4°C , and the supernatant assayed for total protein amounts by BCA assay . Subsequently equal protein amounts of the lysate were analyzed for cAMP levels as per the manufacturer's protocol ( cAMP parameter kit , R&D Systems ) . Wild-type or coronin 1–deficient mice ( 4–5 wk old ) were sacrificed and their brains were immediately dissected out and rapidly processed further at 4°C . The amygdalary regions were microdissected , homogenized , and lysed with the provided cAMP ELISA lysis buffer . The lysate was centrifuged at 9400×g for 10 minutes/4°C and the supernatant assayed for total protein amounts by BCA assay . Subsequently equal protein amounts of the lysate were analyzed for cAMP levels as per the manufacturer's protocol ( cAMP Parameter Kit , R&D Systems ) . PKA activity was measured from 6–8-wk-old wild-type or coronin 1–deficient mice using the Peptag nonradioactive cAMP-dependent protein kinase assay ( V5340; Promega , three animals per group ) . Manually dissected cerebral structures were snap-frozen in liquid nitrogen . All samples were treated together , in duplicates , and corrected for protein concentration ( per milligram of total protein ) . Samples were loaded in the same electrophoretic gels . Protein density was controlled under UV light and simultaneously measured using the Genetools 1 . 0 software ( SYNGENE , Frederick , MD ) . The basal PKA activity represents the difference between the ratios of phospho-/nonphospho forms . A series of experiments was done in the presence of 1 µM cAMP to estimate the maximal PKA activity . All MRI was performed with a 7-Tesla Bruker MRI imaging spectrometer ( Bruker Biospec 70/21 ) using a mouse brain surface coil ( Rapid Biomedicine GmbH , Germany ) . For in vivo MRI scanning , mice were anesthetized with 1 . 5%–2% isoflurane and secured using a head holder to reduce motion artifacts . We obtained 12 ( Figures 4F , G and S7B ) or 18 ( Figure S7A ) coronal T2-weighed images through the entire mice brains using a 2-dimensional multislice spin echo sequence with the following parameters: for the data shown in Figures 4F , G and S7B , 12 coronal T2-weighed images were obtained ( Paravision software PV 3 . 0 . 2 ) using a 2-dimensional multislice spin echo sequence with the following parameters: field of view , 20 mm , acquisition matrix , 256×256; slice thickness , 0 . 6 mm; time of repetition , 3130 . 9 ms; Echo , 1; TE effective 1 , 80 ms; number of averages , 8 . The ventricular areas of all coronal MRI images from each mouse were quantified using the ImageJ program and converted to total pixels per mouse . The data shown are the average ventricular area represented in pixels with SEM ( n = 6 per group ) . For the data in Figure S7A , 18 coronal T2-weighed images were obtained using Paravision software PV 4 . 0 , with the following deviations: acquisition matrix , 512×280; time of repetition , 4 , 691 . 8 ms; TE effective 1 , 76 . 6 ms , with the other parameters being as described above . The ventricular areas of the coronal MRI images were quantified using ImageJ and converted to total pixels per mouse . Data shown represent the average ventricular area in pixels with SEM ( n = 4–6 mice per genotype ) . Cells were subjected to 2 h of serum starvation and either left unstimulated or stimulated with isoproterenol ( 5 or 10 µM for 4 min or as stated ) in the presence of rolipram ( 100 µM , where indicated ) followed by a wash in ice-cold PBS and processed for cAMP analysis ( R&D Systems ) according to the manufacturer's protocol . For the cholera toxin stimulation experiments , coronin 1–expressing Mel JuSo cells or control cells were seeded on a six-well plate and allowed to grow to ∼80% confluence , starved in Opti-MEM for 1 h ( with 100 µM Rolipram ) , followed by stimulation with cholera toxin ( 1 µg/mL ) to the starving cells and further incubation for another 1 h at 37°C . Cells were left untreated or stimulated with isoproterenol ( 10 µM ) for 4 min before cAMP production was measured . Values were normalized to total protein amounts ( BCA analysis ) and converted to pmol cAMP/mg protein . Data shown are representative of three independent experiments . Cells were transfected with the ICUE3 plasmid ( from Dr . Jing Zhang , Johns Hopkins University ) using Fugene 6 following the manufacturer's protocol . After 24 h or 48 h of transfection , the cell culture medium was changed to phenol red-free opti-MEM ( Invitrogen ) and live cell imaging was performed using a Zeiss Cellobserver ( 63× , Oil ) . FRET measurement and analysis was carried out as described below . Cells were excited using a band pass filter ( 445/25 nm , for CFP ) , and the emission signals from both the CFP ( band pass filter 480/22 nm ) and the YFP ( long pass filter 530 nm ) channel were collected by two different EMCCD cameras ( Evolve ) simultaneously . Isoproterenol ( 5 µM ) was added at the time point of 25 s . After acquisition , the background of both CFP and YFP channels was corrected , and the images of both CFP and YFP channels were subjected to ratiometric analysis . The ratiometric imaging of the CFP/YFP channels was calculated pixel by pixel . The ratiometric time course was calculated by the average fluorescence intensity of the entire cell in both CFP and YFP channels . Following cellular stimulation , cells were homogenized in buffer A [20 mM HEPES-NaOH pH 7 . 8 , 30 mM NaCl , 0 . 5 mM DTT , 0 . 2 mM PMSF , 1 mM EDTA , protease and phosphatase inhibitor cocktail ( Thermo Scientific ) ] , cooled on ice for 10 min , followed by centrifugation ( 5 min at 1 , 700×g ) . The pellet was resuspended in buffer A containing 1% sodium-beta-D-maltoside , incubated on ice for 20 min , and centrifuged at 14 , 000×g for 5 min and pooled with the first supernatant . Immunoprecipitation was carried out using antibodies coupled to Dynabeads Protein G ( Invitrogen ) using DMP crosslinking ( Abcam ) according to the manufacturer's protocol . Immunoprecipitated samples were washed four times with buffer A followed by elution in Laemmli sample buffer and analyzed by SDS PAGE ( 10% or 12% ) and immunoblotting . To analyze surface expression of β2-adrenergic receptor levels in coronin 1–expressing and nonexpressing cells , Mel JuSo cells ( lacking coronin 1 ) were labeled with Celltrace Violet ( Invitrogen ) following the manufacturer's protocol and mixed with an equal number of coronin 1–expressing Mel JuSo cells [20] and fixed with 4% paraformaldehyde for 20 min at room temperature . Following this , the cells were washed , blocked with 2% Fetal Bovine Serum containing PBS for 30 min , and incubated with goat polyclonal β2-adrenergic receptor antibody ( 1∶500 , Abcam ) for 1 h . Following washes , the cells were incubated with donkey anti-goat Alexa fluor 488 antibody ( 1∶1 , 000 , Invitrogen ) for 45 min and washed again before being taken up for FACS analysis using BD FACS Canto II with UV laser line 405 nm ( for celltrace violet ) and laser line 488 ( for β2-adrenergic receptor ) . The profiles were analyzed and processed using Flowjo software . The data shown are representative of three independent experiments . RT-PCR analysis for components of the GPCR/cAMP pathway was performed in Mel JuSo cells , as the coronin 1–dependent cAMP production in Mel JuSo cells equals that in neurons or macrophages and cultured cell systems are less prone to variation compared to mouse organ preparations . Total RNA was isolated from 5×106 cells , expressing or lacking coronin 1 ( Mel JuSo ) , using a Qiagen RNA isolation kit in accordance with the manufacturer's protocol . RNA integrity was analyzed employing the RNA 6000 Nano Assay Kit ( Agilent Technologies ) . Reverse transcription ( RT ) reactions were set up according to the manufacturer's protocol with 0 . 65 µg total RNA , Super RT enzyme ( HT Biotechnology ) , and oligo-dT primers ( Invitrogen ) . Polymerase chain reactions using various specific primers ( Table S4 ) and cDNA templates from the RT reactions consisted of 40 cycles of 96°C for 30 s , 56°C for 30 s , and 72°C for 60 s using a RT-PCR Power SYBR Green cocktail ( Applied Biosystems ) using Rotor gene 6000A ( Corbett Research ) . As a control , PCR reactions were performed concurrently with primers for human transferrin receptor along with appropriate RT minus controls . PCR efficiency was determined by performing an eight fold dilution series of four steps in duplicates ( 100% for TFRC ) . Threshold cycles ( crossing point ) were determined using Rotor-Gene software version 6 . 1 . Expression levels were normalized to TFRC [86] . Fold differences were calculated using the delta-delta Ct method [87] . Statistical significance was calculated using Student's t test . Three independent experiments were performed . Results shown are from quadruplicate biological samples per cell type and quadruplicate RT-PCR runs per sample ( 4×4 = 16 runs per gene ) normalized against TFRC gene . Note that for ADCY2 , ADCY4 , ADCY5 , ADCY8 , ADCY9 , ADRB1 and ADRB3 no signal was obtained , suggesting lack of expression . Coronin 1 purification was performed from macrophages essentially as described before [32]; prior to purification , cells were treated with 5 µM Roscovitine for 24 h . Ten confluent 15 cm tissue culture plates of macrophages were harvested and washed using ice-cold PBS 3 times and lysed in 0 . 5% Triton X-100 buffer ( 50 mM Tris-HCl , 137 mM NaCl , 10 mM EGTA , 0 . 5% Triton X-100 ) , containing protease inhibitor ( Roche , Mini complete protease inhibitor cocktail tablets and 1 mM PMSF added briefly before use ) on ice for 30 min . Afterwards , the nucleus and unlysed cells were sedimented ( 14 , 000×g , 15 min at 4°C ) and the supernatant used for coronin 1 purification . After the preparation of the column , the cell lysate was loaded twice on the anti–coronin 1 affinity column [32] , washed using 15 ml washing buffer ( 50 mM Tris-HCl , 150 mM NaCl , pH 8 ) , and eluted by 5 ml elution buffer ( 0 . 1 M glycine , pH 2 . 5 ) . Then , the eluate containing purified coronin 1 was immediately neutralized by 1 M Tris pH 10 . Following purification , the buffer was changed to kinase assay buffer ( 60 mM HEPES-NaOH , 3 mM MgCl2 , 3 mM MnCl2 , 3 µm Na-orthovanadate , 1 . 2 mM DTT , 50 µg/ml PEG 20 , 000 , pH 7 . 5 ) and the eluate concentrated to 0 . 5 mL using centrifugal filter units ( Millipore , Amicon Ultra Centrifugal Filters ) according to the manufacturer's protocol . The purity of the coronin 1 was analyzed by SDS-PAGE ( 10% ) followed by Coomassie staining . For the surface plasmon resonance experiments ( see below ) , coronin 1 or the coronin 1 mutant in which Arg-69 , Lys-20 , Lys-355 , and Glu-102 had been mutated to alanine were expressed as HA-tagged molecules in Mel JuSo cells using the lipofectamine LTX ( Invitrogen/Life Technologies ) protocol according to the manufacturer's instructions . Cells were harvested after 48 h and washed with ice-cold PBS 3 times followed by lysis on ice in cell lysis buffer ( 50 mM HEPES-NaOH pH 7 . 5 , 150 mM NaCl , 1% Triton-X 100 , 0 . 5% NaDOC , 2 mM MgCl2 , 1 mM CaCl2 , 0 . 5 mM EDTA , 0 . 5 mM PMSF , 1 tablet of Roche complete EDTA free protease inhibitor cocktail ) for 30 min . The lysate was centrifuged at 21 , 000×g at 4°C for 20 min . Supernatants were collected , filtered ( 0 . 2 µm ) , and subjected to protein purification . For purification , an anti-HA bead column ( Pierce ) was prepared and washed with 10 ml 100 mM Tris-HCl pH 7 . 5 , followed by 10 ml of buffer A ( 50 mM HEPES-NaOH pH 7 . 5 , 150 mM NaCl , 2 mM MgCl2 , 1 mM CaCl2 , 0 . 5 mM EDTA ) followed by 10 ml cell lysis buffer . Afterwards , the cell lysate was loaded on the anti-HA column and the column was sealed and rotated at 4°C for 2 h . Subsequently , the lysate was collected and reloaded on the anti-HA column for another 3 times . After the lysate loading , the column was washed with 10 ml cell lysis buffer followed by 10 ml of buffer A . The HA-tagged coronin 1 proteins were eluted with 2 ml HA peptide ( 0 . 5 mg/ml , Pierce ) in a buffer containing 150 mM NaCl , 2 mM MgCl2 , 1 mM CaCl2 , 50 mM HEPES-NaOH pH 7 . 5 . Fractions were collected and analyzed by silver staining as well as Western blotting as described above . For Gαs purification , the Gαs proteins , both short and long form , were expressed using in-house constructed expression vectors [88] from the T7 promoter . The N terminus was fused to a 6×His-tagged E . coli thioredoxin A . In addition , a short form construct with a C-terminal 6×His tag was prepared . The constructs were verified by DNA sequencing ( GATC , Germany ) . The proteins were produced in E . coli NiCo21 ( DE3 ) cells ( New England Biolabs ) in ZYM-5052 auto-induction medium [89] overnight at 20°C . Subsequently , they were purified by Ni-NTA-chromatography and gel filtration in 100 mM Tris-Cl pH 8 . 0 . Total yields were about 2 mg pure protein from 6l ZYM5052 culture . Purified Gαs was immobilized on a Series S Sensor Chip NTA ( GE Healthcare , BR-1005-32 ) as follows: The Series S Sensor Chip NTA sensor chip was precleaned by washing with 0 . 35 M EDTA in water followed by activation of the chip with 0 . 5 mM NiCl2 . The mixture ( 1∶1 ) of 1-ethyl-3-[3-dimethylaminopropyl]-carbodiimide hydrochloride ( EDC , 0 . 4 M ) and N-hydroxysuccinimide ( NHS , 0 . 1 M ) solution was injected into the NTA chip ( 480 s , flow rate 5 µl/min ) to initiate the reaction . Afterwards , the purified Gαs ( 10 µg diluted in running buffer , 150 mM NaCl , 2 mM MgCl2 , 1 mM CaCl2 , 50 mM HEPES-NaOH pH 7 . 5 , 1 mg/ml BSA , 0 . 05% NP40 ) was injected into the chip . Finally , the reaction was quenched by injecting ethanolamine solution ( 1 M in water ) for 7 min ( flow rate 5 ml/min ) . The unbound protein was further removed by washing with 0 . 35 M EDTA in water for 90 s . The chip was equilibrated by using the running buffer for 40 min before use . Purified coronin 1 protein or coronin 1 mutant protein was diluted in running buffer at the concentrations indicated and injected into the Gαs immobilized SPR chip with the following parameters: contacting time , 180 s; dissociation time , 360 s; flow rate , 30 ml/min . A regeneration step was included after each cycle by injecting for 60 s a regeneration buffer ( 1 . 5 M NaCl in running buffer ) , followed by a 360 s stabilization step in running buffer . The binding level and binding kinetics were collected and analyzed by the software provided by the supplier . Under continuous anesthesia with isoflurane , mice were positioned in a stereotaxic apparatus ( David Kopf Instruments , Tujunga , CA ) . Stainless steel guide cannula ( 26 gauge; Plastics-One , Roanoke , VA ) were bilaterally implanted above amygdala ( from Bregma position , anteroposterior [AP] −1 . 6 mm , mediolateral [ML] ±3 . 3 mm , dorsoventral [DV] −3 . 2 mm ) . Cannulas were anchored to the skull with dental cement ( Super-Bond , Sun Medical Co . Lt ) . In the end , the mice waked up on a 35°C warm pad , and dummy cannula were inserted into the guide to reduce the risk of infection . To keep mice not stressed during the injection , the mice were trained with a dummy cannula remove and insertion protocol , and they could move freely in their cage . An infusion cannula ( 33 gauge; connected to a 1 µL Hamilton syringe via polyethylene tubing ) projected out of the guide with 1 mm to target the amygdala . The 8-Br-cAMP sodium salt ( Sigma; diluted in saline with 1 . 5 µg in 300 nL ) was infused bilaterally at a rate of 0 . 1 µL per min in a volume of 300∼350 nL per side [45] , [90] , which was controlled by an automatic pump ( Legato 100 , Kd Scientific Inc . , Hilliston , MA ) . The single injection group was injected 30∼45 min before acquisition , and the second injection group was performed both 30∼45 min before and 2 . 5∼3 h after acquisition . To allow penetration of drug , the injector was maintained for an additional 3 min . Then the mice were transferred back to the cages for additional rest . In the end , to analyze the location and extent of the injection area , brains were infused with a fluorophore BODIPY TMR-X ( Invitrogen; 5 mM in PBS 0 . 1 M , DMSO 40% ) . Then slices ( 60 µm ) were imaged using a 5× epifluorescence microscope ( Leica DM5000 ) . Mice were preserved in the analysis only if one side of bilateral injections was precisely targeted or if both saturated areas covered more than 25% of the amygdala . Statistical analysis was performed using GraphPad Prism ( version 4 . 0 ) and SPSS ( version 20 ) . Significance was considered at p<0 . 05 if not otherwise specified . *p<0 . 05 , **p<0 . 01 , ***p<0 . 001 . | Memory and behavior depend on the proper transduction of signals in the brain , but the underlying molecular mechanisms remain largely unknown . Coronin 1 is a member of a highly conserved family of proteins , and although its gene lies in a chromosome region associated with neurobehavioral dysfunction in mice and men , it has never been directly ascribed a specific function in the brain . Here we show that coronin 1 plays an important role in cognition and behavior by regulating the cyclic AMP ( cAMP ) signaling pathway . We find that when cell surface receptors are activated , coronin 1 stimulates cAMP production and activation of protein kinase A . Coronin 1 deficiency resulted in severe functional defects at excitatory synapses . Furthermore , in both mice and humans , deletion or mutation of coronin 1 causes severe neurobehavioral defects , including social deficits , increased aggression , and learning disabilities . Strikingly , treatment with a membrane-permeable analogue of cAMP restored synaptic plasticity and behavioral defects in mice lacking coronin 1 . Together this work not only shows a critical role for coronin 1 in neurobehavior but also defines a role for the coronin family in regulating the transmission of signals within cells . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"signaling",
"signal",
"transduction",
"molecular",
"neuroscience",
"cell",
"biology",
"biology",
"and",
"life",
"sciences",
"neuroscience"
] | 2014 | Coronin 1 Regulates Cognition and Behavior through Modulation of cAMP/Protein Kinase A Signaling |
DNA repair mechanisms in mitotically proliferating cells avoid generating crossovers , which can contribute to genome instability . Most models for the production of crossovers involve an intermediate with one or more four-stranded Holliday junctions ( HJs ) , which are resolved into duplex molecules through cleavage by specialized endonucleases . In vitro studies have implicated three nuclear enzymes in HJ resolution: MUS81–EME1/Mms4 , GEN1/Yen1 , and SLX4–SLX1 . The Bloom syndrome helicase , BLM , plays key roles in preventing mitotic crossover , either by blocking the formation of HJ intermediates or by removing HJs without cleavage . Saccharomyces cerevisiae mutants that lack Sgs1 ( the BLM ortholog ) and either Mus81–Mms4 or Slx4–Slx1 are inviable , but mutants that lack Sgs1 and Yen1 are viable . The current view is that Yen1 serves primarily as a backup to Mus81–Mms4 . Previous studies with Drosophila melanogaster showed that , as in yeast , loss of both DmBLM and MUS81 or MUS312 ( the ortholog of SLX4 ) is lethal . We have now recovered and analyzed mutations in Drosophila Gen . As in yeast , there is some redundancy between Gen and mus81; however , in contrast to the case in yeast , GEN plays a more predominant role in responding to DNA damage than MUS81–MMS4 . Furthermore , loss of DmBLM and GEN leads to lethality early in development . We present a comparison of phenotypes occurring in double mutants that lack DmBLM and either MUS81 , GEN , or MUS312 , including chromosome instability and deficiencies in cell proliferation . Our studies of synthetic lethality provide insights into the multiple functions of DmBLM and how various endonucleases may function when DmBLM is absent .
Crossover repair of DNA damage is associated with detrimental side effects , including loss of heterozygosity and formation of chromosome rearrangements . This genomic instability is highly deleterious , being linked to loss of cell cycle regulation and cell death; consequently , crossover ( CO ) formation is strongly suppressed in normal mitotic cells . One source of COs is the recombinational repair of DNA double-strand breaks ( DSBs ) . The most widely cited model for formation of COs during DSB repair involves formation of an intermediate with two four-stranded Holliday junctions ( HJs; see Figure S1 ) [1] . Apparent double-Holliday junction ( dHJ ) intermediates have been isolated as precursors of meiotic COs in Saccharomyces cerevisiae [2] . Similar structures are also formed during DSB repair in vegetative S . cerevisiae cells , though at a much lower frequency [3] . The BLM helicase has been identified as a key anti-CO factor . Mutations in the human BLM gene lead to Bloom syndrome , which is characterized by reduced size , fertility defects , immunodeficiency , and highly increased risk for a broad spectrum of cancers [4] . On the cellular level , BLM mutation increases COs between sister chromatids and homologous chromosomes , and increases the frequency of deletions and genome rearrangements [5] . This function is widely conserved , as the S . cerevisiae ortholog , Sgs1 , also prevents COs during DSB repair [6] and the Drosophila ortholog , DmBLM , prevents both spontaneous and induced mitotic COs [7] . At least two models to explain the anti-CO activity of BLM have been proposed ( Figure S1 ) . First , DmBLM has been shown to promote the synthesis-dependent strand annealing ( SDSA ) pathway for DSB repair ( Figure S1 ) [8] . It has been suggested that BLM's function in this pathway is to dissociate D-loops generated by strand exchange and repair synthesis [9] . Second , BLM has been proposed to catalyze convergent branch migration of the two HJs in the dHJ intermediate and facilitate subsequent decatenation by TOP3α [10] . These hypotheses are supported by in vitro demonstration of D-loop disruption , HJ branch migration , and , together with TOP3α and other proteins , dHJ dissolution activities [11]–[13] . In the absence of BLM , it is thought that COs may be generated through cleavage of a dHJ or similar structure by a HJ resolvase ( dHJ resolution ) . The identity of the hypothesized HJ resolvase remains unknown , but may be one or more of the three structure-selective nuclear endonucleases that have been reported to cleave HJs in vitro: Mus81–Eme1/Mms4 , GEN1/Yen1 , and SLX4–SLX1 . The first of these to be implicated in HJ resolution was Mus81-Eme1/Mms4 [14] . Mus81–Eme1 is required for most meiotic COs in S . pombe and a subset of meiotic COs in several other organisms [14]–[18] , and appears to be involved in generating many spontaneous mitotic COs in S . cerevisiae [19] . However , while Mus81–Eme1/Mms4 can cut fully-ligated HJs in vitro , it has more robust activity on other structures , including nicked HJs , 3′ flaps , and structures that mimic replication forks [20]–[24] . Genetic studies implicate this enzyme in replication-associated repair [25] . Mammalian cells mutant for MUS81 or EME1 are hypersensitive to agents that generate DNA damage that blocks replication forks , such as the interstrand crosslinking agent cisplatin [26]; yeast mus81 mutants are hypersensitive to the alkylating agent methyl methanesulfonate ( MMS ) and UV radiation [27] . Curiously , Drosophila mus81 mutants do not display the same strong hypersensitivities or have defects in generating meiotic COs [28] . The second eukaryotic nuclease found to resolve HJs was initially purified from yeast ( Yen1 ) and human cells ( GEN1 ) by its HJ-resolvase activity . GEN1 and Yen1 both cut HJs in a symmetrical , re-ligatable manner; they also cut 5′ flap and replication fork-like structures , though not as well as they cut HJs [29] . In vivo functions of Yen1/GEN1 are poorly understood . S . cerevisiae yen1 mutants are not hypersensitive to DNA damaging agents and grow normally , but mus81 yen1 double mutants exhibit slow growth [30] . These double mutants also are more hypersensitive to MMS , HN2 , camptothecin , and hydroxyurea and have fewer spontaneous mitotic COs than mus81 single mutants [19] . S . pombe lacks a Yen1 ortholog , but expression of human GEN1 rescues the meiotic CO and mutagen sensitivity defects of mus81 mutants [31] . Together , these observations suggest that Yen1/GEN1 primarily plays a backup role to Mus81–Mms4/Eme1 . The most recent enzyme reported to cut HJs in vitro is human BTBD12/SLX4–SLX1 [32]-[34] . Like GEN1 , SLX4–SLX1 also cuts 5′ flaps and structures that mimic replication forks . Functions of this enzyme in vivo are also poorly understood . The Drosophila ortholog of SLX4 is MUS312 [35]; mus312 mutants are hypersensitive to agents that induce DNA interstrand crosslinks ( ICLs ) , suggesting that MUS312 is required to repair ICLs [36] . Vertebrate SLX4–SLX1 has also been implicated in ICL repair , based on hypersensitivity to crosslinking agents of cells in which either subunit is knocked down by RNAi [32]–[35] . In support of this conclusion , recent studies have identified mutations in SLX4 in some patients with Fanconi anemia , a disorder associated with an aberrant response to crosslinking agents [37] , [38] . Slx1 , the catalytic subunit , appears to function solely when dimerized with Slx4 , but Slx4 has other nuclease partners [39] . One of these is Rad1–Rad10 , an endonuclease that functions in nucleotide excision repair [40] . These dual interactions are conserved in the Drosophila and vertebrate orthologs [32]–[35] . The Drosophila ortholog of Rad1–Rad10 , MEI-9–ERCC1 , is required for most meiotic COs . Interaction with MUS312 is essential for this function; it has been proposed that MUS312–MEI-9–ERCC1 generates COs by resolving HJs [41] , [42] , but in vitro analysis of this enzyme has not been published . Mouse SLX4 and the C . elegans ortholog HIM-18 are also involved in generating meiotic COs , though the extent to which different interacting nucleases are involved in these organisms is not yet clear [43] , [44] . In fungi , simultaneous loss of both the BLM helicase ortholog , Sgs1 , and either Mus81–Eme1/Mms4 or Slx4–Slx1 is lethal [45] , [46] , [47] . Studies of these synthetic lethal phenotypes have provided additional insights into functions of Sgs1/BLM and these putative HJ resolvases . Likewise , previous studies revealed that mutations in Drosophila mus309 , the gene that encodes DmBLM , are synthetically lethal with mutations in mus81 or mus312 [28] , [35] , [48] . Genetic interactions have also been observed in Bloom syndrome cells in which these nucleases were knocked down singly or in combinations , ranging from modest decreases in sister chromatid exchange to chromosome fragmentation and decreased cell viability [49] . We have now obtained mutations in Drosophila slx1 and Gen . We find that slx1 mus309 mutants are inviable , with phenotypes similar to those of mus312 mus309 mutants . As in yeast , GEN and MUS81 have some overlapping or compensatory functions; however , Gen mutants have more severe hypersensitivities than mus81 mutants , suggesting that GEN plays a more critical role in DNA repair in Drosophila . In support of this conclusion , Gen mus309 mutants are inviable , and die much earlier in development than either mus81; mus309 or mu312 mus309 double mutants . Therefore , three putative HJ resolvases – MUS81–MMS4 , GEN , and MUS312–SLX1 are essential in the absence of DmBLM . Each of the double mutants has defects in cell proliferation and features of chromosome instability , though the severities vary from genotype to genotype . The effects of blocking recombination by mutating the gene encoding the ortholog of the strand exchange protein Rad51 ( SPN-A in Drosophila ) also vary , from nearly complete suppression of defects to selective suppression of a subset of phenotypes . We also analyzed the effects of a mus309 mutation that abolishes a subset of DmBLM functions . Together , our results suggest models for functions of DmBLM in responding to spontaneous replication fork problems , and how MUS81–MMS4 , GEN , and MUS312–SLX1 might function in alternative , DmBLM-independent pathways .
We previously reported synthetic lethalities between mutations in mus309 , which encodes the Drosophila ortholog of the BLM helicase , and mutations in mus81 or mus312 [28] , [35] , which encode subunits of putative HJ resolvases [14] , [32] , [33] , [34] . MUS81 is the catalytic subunit of a heterodimeric endonuclease ( MUS81–MMS4 ) . MUS312 is a non-catalytic subunit that interacts with at least two nucleases , SLX1 and MEI-9–ERCC1 [35] . It seemed likely that the mus312 mus309 lethality is due to loss of the MUS312–SLX1 nuclease , since S . cerevisiae slx1 sgs1 double mutants are inviable and have phenotypes similar to those of slx4 sgs1 double mutants [47] . To determine the contributions of SLX1 and MEI-9–ERCC1 to mus312 mus309 lethality , we generated double mutants between each of them and mus309 . We found that mei-9; mus309 double mutants are viable , consistent with the previous finding that MEI-9–ERCC1 interacts with MUS312 in meiotic recombination , but not in somatic DNA repair [42] . Mutations in Drosophila slx1 ( CG18271 ) have not been reported previously . A complication in generating a mutation in this gene is that the first ( non-coding ) exon overlaps the first exon of MED31 , which is thought to be an essential gene [50] . We therefore generated a synthetic deletion by combining a 30-kb chromosomal deficiency with a transgene that spans the region but on which we disrupted slx1 ( Materials and Methods , Figure S2 ) . As an alternative approach , we ordered Targeted Induction of Local Lesions in Genes ( TILLING ) from the Seattle Drosophila TILLING Project [51] . This effort identified 24 missense mutations in slx1 , the most promising being one that changes a conserved phenylalanine residue to isoleucine ( F93I; Figure S2 ) . Both slx1F93I and the synthetic deletion are lethal when combined with mus309 mutations . Double mutants die early in the pupal stage , larvae lack imaginal discs , and larval neuroblasts are frequently polyploid . These phenotypes are similar to those of mus312 mus309 double mutants [35 , see below] . We conclude that the inviability of mus312 mus309 double mutants is indeed due to simultaneous loss of DmBLM and MUS312–SLX1 . In experiments described below , we used mus312 mus309 mutants to further characterize defects caused by loss of DmBLM and MUS312–SLX1 . Orthologs of MUS81–MMS4 and MUS312–SLX1 have been implicated in HJ resolution . We therefore wanted to determine whether GEN , which is orthologous to the HJ resolvases Yen1 and GEN1 , is also essential when DmBLM is absent . Drosophila GEN was initially identified as a novel RAD2/XPG family nuclease [52] . GEN was found to cut flaps and the lagging strand of replication fork-like structures , as well as to have a weak exonuclease activity on nicked substrates [53] . However , no stocks carrying Gen mutations have been reported . To identify such mutations , we screened through a collection of mutagen-sensitive ( mus ) lines for which the causative mutations had not been mapped [54] . We discovered that the two available mus324 stocks both have mutations in Gen . Although the two alleles , mus324Z4325 and mus324Z5997 , were assumed to be independent , they have identical mutations ( deletion of ATATAC and insertion of a single G , creating a frameshift at codons 374-5 , which is within the conserved nuclease domain ) , suggesting that they are two isolates of the same mutational event . mus324 mutants are hypersensitive to the crosslinking agent HN2 and to the alkylating agent MMS [54]; we found that these hypersensitivities are uncovered by Df ( 3L ) Exel6103 , a deletion that removes Gen and 16 other genes ( data not shown ) . We conclude that mus324 is Gen and hereafter refer to these alleles as GenZ4325 and GenZ5997 . In contrast to the situation in S . cerevisiae , Drosophila Gen mutants are more hypersensitive to MMS and HN2 than mus81 mutants [28 , 54 , S . Bellendir and JS , unpublished data] . As in S . cerevisiae , however , there appears to be overlapping functions for these two nucleases . In the absence of exogenous DNA damage , mus81; Gen double mutants have wild-type survival rates relative to heterozygous siblings , but the eyes of double mutants are reduced in size and exhibit mild roughening ( data not shown ) . This phenotype often results from cell cycle defects and/or increased apoptosis disrupting the highly ordered ommatidia . We quantified apoptosis in larval imaginal discs , which consist of proliferative diploid cells that give rise to adult epidermal structures such as eyes , wings , and legs . Imaginal wing discs from mus81 and Gen single mutants have the same levels of apoptosis as wing discs from wild-type larvae , but double mutants have significantly increased levels ( Figure 1 ) . The high level of apoptosis in the double mutant imaginal discs suggests that MUS81 and GEN have shared functions that contribute to cell survival or proliferation even in the absence of exogenously-induced DNA damage . Knowing that two putative HJ resolvases ( MUS81–MMS4 and MUS312–SLX1 ) are required when DmBLM is absent , we wanted to determine whether GEN is also essential when DmBLM is absent . We generated Gen mus309 double mutants and found that they die early in larval development , reaching only the first instar stage ( Table 1; Figure 2 ) . This is earlier in development than mus81; mus309 double mutants , which die as pharate adults ( adult structures such as wings and eyes are visible within the pupal case , but no adults eclose ) , and also earlier than mus312 mus309 double mutants , which die at an early pupal stage . As a consequence of the genetic crosses we used ( see Materials and Methods ) , the mus81; mus309 double mutants we analyzed have no maternal contribution of MUS81 , but both Gen mus309 and mus312 mus309 double mutant larvae potentially have maternally-deposited wild-type GEN and MUS312 ( there is maternal DmBLM in all three cases ) . The weaker lethal phenotype of mus81; mus309 mutants is therefore consistent with MUS81 contributing less to DmBLM-independent pathways than either GEN or MUS312 . To gain more insights into the causes of the three synthetic lethalities , we examined several highly proliferative larval tissues . Although most larval growth is due to enlargement of cells undergoing endocycles without mitosis , there is extensive cell proliferation in several tissues , including the neuroblasts of brain , the imaginal discs , and cells in the imaginal ring of the salivary gland . These tissues all appear to be normal in mus81; mus309 double mutants . In contrast , mus312 mus309 mutants have small brains , lack imaginal discs , and have a reduced number of salivary imaginal cells ( Figure 3 ) . Nuclei of the remaining salivary imaginal cells appear larger than in wild-type larvae , suggesting increased DNA content . These phenotypes indicate that MUS312–SLX1 has a more critical role in proliferation in the early larva than MUS81–MMS4 ( Table 1 ) . Gen mus309 mutants die too early to examine these tissues; it is likely that this early lethality is due to an even more severe defect in cell proliferation . We hypothesized that the proliferation defects described above stemmed , at least in part , from unrepaired DNA damage and/or unresolved DNA repair intermediates . To determine whether there were gross chromosomal changes in the double mutants , we arrested larval neuroblasts with colchicine and examined mitotic nuclei for chromosome structural damage and aneuploidy . The frequencies of chromosome aberrations and aneuploidy in mitotic neuroblasts were indistinguishable between wild-type and single mutants for mus81 , mus312 , mus309 , or Gen ( Figure 3 , Figure 4 ) . In mus81; mus309 double mutants , there was an increased frequency of broken chromosomes ( Figure 4A ) . The mus312 mus309 double mutants showed extreme genome instability: No mitotic nuclei with completely intact chromosomes were detected , and about a third showed polyploidy ( Figure 3B , Figure 4 ) . Precise chromosome counts could not be made because of the highly fragmented nature of the chromosomes , but some nuclei appeared to be more than 4N ( tetraploid ) . The early larval death of Gen mus309 mutants precluded us from examining neuroblast chromosomes in this genotype . In S . cerevisiae , synthetic lethality between sgs1 and mus81 is suppressed by rad51 mutations , but lethality between sgs1 and slx4 is not [45] , [55] . The Drosophila ortholog of the strand exchange protein RAD51 is encoded by spn-A [56] . As in yeast , mutation of spn-A suppresses the lethality of mus81; mus309 mutants [28] . A simple interpretation is that strand exchange mediated by SPN-A leads to a toxic intermediate that must be processed by either DmBLM or MUS81; however , a more thorough analysis of multiple genotypes suggested a more complex model [28] . Suppression of lethality of mus81; mus309 by spn-A is not complete , since the triple mutant still has increased apoptosis relative to wild-type larvae and only 70% of the triple mutants survive to adulthood [28] . We found that loss of spn-A also leads to a significant decrease in the number of chromosome breaks in neuroblast cells of mus81; mus309 mutants ( Figure 4A ) . As with viability and apoptosis , suppression is not complete , which suggests that either a subset of the damage that DmBLM and MUS81 are required to process is not generated through a SPN-A-dependent process , or that pathways that operate when SPN-A is unavailable are not sufficient to repair all the spontaneous damage that would normally be processed through strand exchange-mediated pathways . Mutation of spn-A has a similarly pronounced effect on the phenotype of Gen mus309 mutants . Rather than dying at the first larval instar , the Gen mus309 spn-A mutants survive to the pupal stage ( Table 1 ) . Third instar triple mutant larvae have small brains and lack imaginal discs , though very small rudimentary discs are occasionally visible . Salivary imaginal cells are reduced in number and have enlarged nuclei ( Table 1 ) . Neuroblasts from Gen mus309 spn-A larvae have numerous chromosome breaks ( Table 1 , Figure 4A ) . Thus , preventing strand exchange ameliorates the phenotypes caused by loss of GEN and DmBLM , but the triple mutants still have cell proliferation defects . In contrast to the profound suppression of defects that arise when DmBLM and either MUS81 or GEN are absent , preventing strand invasion suppressed only some defects caused by loss of both MUS312 and DmBLM . Like mus312 mus309 double mutants , mus312 mus309 spn-A triple mutants lack imaginal discs , have small brains , and die at the pupal stage ( Table 1 ) . However , mutation of spn-A does suppresses the chromosome breakage or polyploidy phenotypes ( Figure 4 ) , as well as the defect in salivary gland imaginal cells ( data not shown ) . The differences in the effect of spn-A mutations on the various phenotypes suggests that there are multiple circumstances in which either DmBLM or MUS312–SLX1 are essential , but that only a subset of these result from strand exchange . Trowbridge et al [28] reported that mus81 mutations are viable with the separation-of-function allele mus309N2 . This mutation is an intragenic deletion predicted to remove the first 566 residues of DmBLM but leave the helicase domain intact [7] , [28] . Previous studies showed that mus309N2 are similar to null mutants in their inability to repair DNA double-strand gaps by SDSA , their hypersensitivity to ionizing radiation , and their elevated levels of spontaneous mitotic COs [7] . However , maternal-effect embryonic lethality , which is associated with extensive anaphase bridging in early-stage embryos , is substantially reduced in mus309N2 mutants compared to null mutants , though not to wild-type levels [7] . We hypothesized that DmBLM is required during the extremely rapid early embryonic S phases , particularly in the decatenation of converging replication forks , and that DmBLMN2 is capable of carrying out this process , though not with wild-type efficiency [7] . This led to the suggestion that an important function revealed by mus81; mus309 lethality is in processing blocked or regressed replication forks , either by DmBLM-catalyzed migration or by MUS81-dependent cleavage . In this model , mus81; mus309N2 mutants are viable because DmBLMN2 retains the ability to migrate these forks . Thus , the alleviation of maternal-effect lethality in mus309N2 females and the viability of mus81; mus309N2 double mutants suggests that DmBLMN2 retains some replication-fork processing functions . In contrast , the null-equivalent defect in SDSA in mus309N2 mutants suggests that the ability to disrupt D-loops during SDSA is destroyed in DmBLMN2 , while the null-equivalent elevation in mitotic COs suggests that DmBLMN2 is also unable to catalyze dHJ dissolution . To determine the extent to which the activities retained by DmBLMN2 can compensate for the loss of GEN or MUS312–SLX1 , we made Gen mus309N2 and mus312 mus309N2 double mutants . Gen mus309N2 mutants are inviable . However , rather than dying as first instar larvae , Gen mus309N2 mutants die later , as pharate adults . These double mutants have apparently normal imaginal disc size , brain size , and number/size of salivary gland imaginal cells ( Table 1 ) , but their neuroblasts frequently exhibit chromosome breaks ( Figure 4A ) . The striking differences between Gen mus309 and Gen mus309N2 mutants in their cell proliferation phenotypes and stages of lethality suggest that GEN has an important role in processing replication-associated structures when DmBLM is not available , consistent with the known biochemical activities of GEN [29] . mus312 mus309N2 mutants are also inviable , and are similar to double mutants between mus312 and null alleles of mus309 in that larvae lack imaginal discs and lethality occurs in the pupal stage ( Table 1 ) . However , several mutant phenotypes are less severe in mus312 mus309N2 double mutants . Small , severely underdeveloped imaginal discs are sometimes detected in third-instar larvae , and in metaphase neuroblasts there are fewer damaged chromosomes and polyploidy is not seen ( Figure 4 ) . These observations suggest that defects in replication contribute to the chromosome breaks , polyploidy , and , perhaps stemming from these aberrations , proliferation defects that are seen in mus312 mus309 double mutants . Gen mus309N2 mutants have fewer chromosome breaks than mus312 mus309N2 mutants ( P = 0 . 3; P = 0 . 35 for the differences in frequency of polyploidy ) , but the latter die earlier . Therefore , chromosome breaks in neuroblasts are not the sole cause of lethality . The early pupal lethality of mus312 mus309N2 mutants is most likely due to the absence of imaginal discs; the reasons for the loss of this tissue are unknown , but are likely due to poor cell proliferation , elevated apoptosis , or both .
In S . cerevisiae , mus81 yen1 double mutants have a slow growth phenotype [58] , and we found that Drosophila mus81; Gen double mutants have elevated levels of apoptosis . Thus , in both budding yeast and flies , simultaneous loss of MUS81–MMS4 and Yen1/GEN leads to spontaneous defects in cell proliferation . Although this suggests some functional overlap , the relative contributions of the two enzymes appears to be reversed in these organisms . In yeast , mus81 single mutants are hypersensitive to a number of DNA damaging agents , but yen1 mutants are not [19] , [27] , [30] , [58] , whereas in flies , Gen mutants have severe hypersensitivities and mus81 mutants have only weak hypersensitivities [28 , 54 , S . Bellendir and JS , unpublished data] . It has been proposed that Mus81–Mms4 cuts nicked HJs , but if left uncut ( as in mus81 mutants ) , these are ligated into intact HJs that are cleaved by Yen1 [30] , [31] , [49] . The in vitro biochemical activities of GEN and MUS81 and the drug sensitivities of single mutants suggest that these enzymes function in replication fork repair . GEN and MUS81–MMS4 cut different sides of replication fork-like substrates in vitro . Functional redundancy could be explained by the ability of either to cut blocked forks ( Figure 5A , i ) ; however , in both yeast and Drosophila one enzyme plays a larger role in surviving exogenous DNA damage , suggesting that these enzymes act on structures other than simple stalled forks . An obvious candidate is a regressed fork . Based on in vitro activities , MUS81–MMS4 would be expected to have a preference for forks that are regressed but have not undergone template switching ( Figure 5A , ii ) , whereas GEN would be expected to cut regressed forks in which the leading strand has undergone template switching ( Figure 5A , iii ) . The different biases in enzyme preference might be explained by differing degrees of forks regression in Saccharomyces versus Drosophila . This model assumes that Drosophila GEN , like Yen1 and human GEN1 , is an HJ resolvase . The question of whether GEN is a resolvase has important implications for understanding the partial redundancy between Drosophila GEN and MUS81–MMS4 . The rescue of S . pombe mus81 mutant phenotypes by human GEN1 and studies of knockdown of these enzymes in human cells have been interpreted with respect to the HJ-cutting activities [30] , [49]; however , a previous study of Drosophila GEN did not detect resolvase activity [53] . That study employed full-length GEN; it is possible that , like human GEN , the unstructured C-terminus must be removed to allow HJ cleavage in vitro [29] . Regardless of whether GEN cuts HJs , it remains possible that the genetic overlap between MUS81 and GEN is due at least in part to cleavage of other substrates that might arise during recombination or replication fork repair . Given that GEN and MUS81–MMS4 have some overlapping function ( s ) and yen1 sgs1 double mutants are viable in S . cerevisiae [30] , we were surprised to discover that Gen mus309 double mutants are inviable . In fact , of the three synthetic lethalities we characterized , Gen mus309 double mutants have the most severe developmental phenotype . This suggests that the structures upon which GEN can act are more frequently produced and/or more deleterious when left unprocessed by DmBLM . Conversely , mus81; mus309 mutants die the latest in development and have the least severe defects in proliferation and chromosome stability , suggesting that structures upon which MUS81–MMS4 acts are less frequently produced and/or less deleterious when left unprocessed , or that there are additional repair options available . Insights into the nature of the structures upon which either DmBLM or one of these endonucleases can act comes from the finding that double mutants with mus309N2 have much milder defects than double mutants with null alleles of mus309 ( Table 1 , Figure 4 ) . mus309N2 is thought to abolish the DSB repair and dHJ dissolution functions while leaving some replication fork function ( s ) largely intact [7] . This suggests that synthetic lethality between Gen and mus309 and between mus81 and mus309 are not due to inability to dissolve dHJs or disrupt D-loops , but to inability to process replication fork structures . Additional clues come from the observation that preventing strand exchange partially rescues Gen mus309 and mus81; mus309 . In both cases , every phenotype we studied is affected , though rescue is incomplete for each . Incomplete rescue may be because the repair methods that do not rely on strand exchange are themselves problematic , or because some repair intermediates that require either DmBLM or one of these endonucleases are generated by stand exchange and some are not . A model that is consistent with our findings is illustrated in Figure 5A . It is believed that when a replication fork encounters a block to leading strand synthesis , the fork is regressed so that it is stabilized and so the blockage is accessible for removal ( Figure 5A , ii ) . In some cases , the nascent leading strand may anneal to the nascent lagging strand ( Figure 5A , iii ) . This template switch allows the leading strand to be extended , so that after reversal of the regression the block is bypassed . We hypothesize that DmBLM is required for reversal of regression . Forks that are regressed to various degrees might be cleaved by MUS81–MMS4 ( Figure 5A , iv ) or by GEN ( Figure 5A , v ) . Regressed forks that are not reversed or cut are toxic and trigger apoptosis . In the absence of both DmBLM and MUS81–MMS4 , template switching is still an option , but in the absence of both DmBLM and GEN , there are no further options; hence , Gen mus309 mutants have a more severe defect than mus81; mus309 mutants . DmBLMN2 is capable of reversing regressed forks , and although its activity is less than that of full-length DmBLM , it is sufficient to allow survival of most mus81; mus309N2 individuals to adulthood , and survival of Gen mus309N2 to the pharate adult stage . Some studies suggest that Rad51 is required to protect blocked forks and perhaps to carry allow regression [59] , [60] , [61] . If this is true in Drosophila , then mutation of spn-A may suppress defects in double mutants by preventing fork regression , thereby blocking buildup of toxic structures . Blocked forks that are not protected by SPN-A may spontaneously break , giving rise to structures that are similar to those generated by MUS81–MMS4 or GEN cleavage ( Figure 5A , dotted line ) . Several models have been proposed to explain how these broken forks are repaired to allow replication restart [62] . These typically involve strand invasion from the broken end into the intact sister chromatid . In Drosophila , however , we propose that continued replication from adjacent forks or from de novo firing of nearby origins converts the one-ended DSB into a two-ended DSB ( Figure 5A , vi , vii ) . This proposal is consistent with the finding that substantial DNA synthesis persists after induction of S-phase damage in Drosophila [63] . Repair of the two-ended DSB would typically occur through DmBLM-dependent SDSA ( see Figure S1 ) . However , if DmBLM is not available to promote SDSA , repair occurs through a pathway that may generate a CO . As a consequence of pairing of homologous chromosomes in proliferating cells in Drosophila , repair will often involve interaction between homologs; this can contribute to the high elevation in mitotic COs in mus309 mutants [7] . In the most popular models , COs arise through resolution of HJ intermediates . It is possible that GEN plays a role in this process and that this also contributes to the early lethality of Gen mus309 double mutants . The mus312 mus309 synthetic lethality we describe is unique in that it is not alleviated by blocking strand exchange . This has also been reported for S . cerevisiae sgs1 slx4 lethality [47] . Fricke et al . [39] proposed that an important overlapping function between Sgs1 and Slx4–Slx1 is in rDNA replication: Sgs1–Top3 decatenates forks that stall during rDNA replication , but in the absence of Sgs1 these structures are cut by Slx4–Slx1 . A similar model has been suggested in S . pombe [64] . McVey et al . [7] hypothesized that DmBLM–TOP3α is required to decatenate converging replication forks during the extremely rapid S phases of early embryonic development . At this stage of development , DNA repair processes seem to be disabled [65] , so maternally-deposited DmBLM is essential for early embryonic replication . We hypothesize that DmBLM is still involved in decatenation of problematic fork convergences at later developmental stages , but that DmBLM is no longer essential because a secondary pathway is available: cleavage by MUS312–SLX1 ( Figure 5B ) . Since converging forks are not generated by strand exchange , prevention of strand exchange ( through mutation of spn-A ) does not rescue lethality . Likewise , mus312 mus309N2 mutants remain inviable because DmBLMN2 is predicted to be unable to interact with TOP3α [7] , an interaction that is expected to be essential for decatenation of converging forks . Interestingly , the chromosome breakage and aneuploidy phenotypes are milder in mus312 mus309N2 and in mus312 mus309 spn-A than in mus312 mus309 null alleles ( Table 1 , Figure 4 ) . This suggests that there are additional structures , generated by strand exchange but on which DmBLMN2 cannot act , that can be cleaved by MUS312–SLX1 . One potential additional function for SLX4–SLX1 is in repairing DNA ICLs [32] , [33] , [34] , [35] , [42] . Given the HJ resolvase activity of human SLX4–SLX1 , it seems plausible that MUS312–SLX1 cuts a single HJ intermediate or replication fork-like structures formed during ICL repair . It is unclear what defect leads to polyploidy . It has been suggested that prolonged blocks to the completion of DNA replication might prevent cytokinesis , leading to tetraploidy [66] . Consistent with this hypothesis , defects in S-phase-coupled processing of histone mRNAs leads to tetraploidy in Drosophila neuroblasts [67] . We've established that MUS81–MMS4 , GEN , and MUS312–SLX1 and are each required in the absence of DmBLM , presumably because these enzymes cleave spontaneously arising DNA structures that are usually acted upon by DmBLM . Although each of these nucleases has been considered primarily as an enzyme that cuts HJs , it is likely that the toxic intermediates that contribute to lethality also include other structures derived from replication forks . We have suggested models to explain the unique functions for each of these nucleases that become essential when DmBLM is absent . Even if these models are correct , it is likely that they describe only a subset of roles for these enzymes . Further studies of cellular phenotypes that occur in mutants lacking various combinations of these enzymes should provide additional insights into the complexities of replication fork repair and the origins and mechanisms of mitotic recombination .
Flies were raised on standard cornmeal-based media at 25°C . The following allelic combinations were used: mus312 mutants were heteroallelic for the null alleles mus312D1 ( Q226ter ) and mus312Z1973 ( K143ter ) ; mus309 mutants were heteroalleleic for the null alleles mus309D2 ( W922ter ) and mus309N1 ( Δ 2480bp N-terminus ) or mus309D2 and the separation-of-function allele mus309N2 ( Δ1537 bp N-terminus ) ; Gen mutants were hemizygous for Gen4325 ( mus324Z4325 ) or Gen5997 ( mus324Z5997 ) , over Df ( 3L ) Exel6103 ( deletes 19 genes in 64C4-64C8 , including Gen ) ; mus81 mutants were homozygous ( females ) or hemizygous ( males ) for mus1Nhe1 , which has a premature stop codon inserted by targeted recombination [28] . To generate a synthetic deletion of slx1 , we first made Df ( 3R ) HKK1 by inducing FLP-mediated recombination between the FRT sequences on P{XP}d03662 , inserted at 425 , 462 ( coordinates are from chromosome 3L in release 5 . 36 of the Drosophila genome ) and PBac[37]slx1e01051 , inserted at 470 , 260 , in the 3′ untranslated region of slx1 ( Figure S2 ) . To complement genes other than slx1 that were deleted in Df ( 3R ) HKK1 , we modified the P[acman] clone CH321-44C16 [68] , which carries sequences spanning 399 , 145 to 473 , 218 . We used recombineering to replace 469 , 261 to 470 , 077 with a gene conferring bacterial resistance to kanamycin . The deleted region contains almost the entire slx1 coding sequence , but does not overlap with MED31 . We were initially unable to get transformants of this large BAC clone , so we also replaced the 39-kb region from 399 , 284 to 438 , 520 with the bla gene that confers resistance to ampicillin . The remaining insert spans all annotated genes that are deleted in Df ( 3R ) HKK1 , but is deleted for most of slx1 . This clone was transformed into a phiC31 attP landing site on 3L ( P{CaryP}attP2 , in 68A4; injections were done by BestGene , Inc . ) . We named this integration Dp ( 3;3 ) HKK2 . Finally , we constructed a recombinant chromosome that has Dp ( 3;3 ) HKK2 and Df ( 3R ) HKK1 . This chromosome is therefore euploid except for the loss of slx1 . Flies homozygous for this chromosome are viable and fertile . Discs were harvested in Ringer's buffer from wandering third instar larvae and fixed in 4% formaldehyde in PBST ( 0 . 1% Triton-X in PBS ) for 45 min . After washing in PBST , the discs were blocked in 5% bovine serum albumin in PBST 1 hr at room temp . They were then stained overnight at 4°C with 1∶500 anti-Cleaved Caspase-3 ( Cell Signaling #966S ) in PBST . The following day , the discs were stained two hours at room temperature with 1∶1000 the 2° Alexa Fluor 488 goat anti-rabbit ( Molecular Probes #A11034 ) . Discs were then washed , fine-dissected , and mounted on a slide with Fluoromount-G ( SouthernBiotech #0100-01 ) and sealed with nail polish . Images were taken with a Nikon Eclipse E800 fluorescent microscope . Third instar larvae brains were dissected and soaked in 0 . 1 mM colchicine in 0 . 7% saline for 1 . 5 hrs , followed by 8 min in 0 . 5% sodium citrate . Brains were fixed for 20 sec in a 5 . 5∶5 . 5∶1 solution of acetic acid: methanol: water . Brains were transferred to a slide and treated with 45% acetic acid for 2 min , then squashed under a siliconized coverslip . The coverslip/slide was placed on dry ice for 10 min , then the coverslip was flicked off and the slide washed in −20°C ethanol then dried for storage at 4°C . The slide was rehydrated for 5 min in 2x SSC ( 300 mM NaCl , 30 mM sodium citrate , pH 7 . 0 ) , then stained in 2x SSC plus 1∶10 , 000 DAPI ( 1mg/mL ) for 5 min , then washed in 2xSSC and air-dried . The slide was mounted with Fluoromount-G ( SouthernBiotech #0100-01 ) and sealed with nail polish . Images were taken with a Nikon Eclipse E800 fluorescent microscope . Salivary glands were dissected from third instar larvae in Ringer's buffer and fixed for 45 min in 4% formaldehyde in PBST ( 0 . 1% Triton-X in PBS ) . After PBST washes , the glands were stained with 1∶1000 DAPI ( 1mg/ML ) 5 min and washed again . Glands were mounted on slides using Fluoromount-G ( SouthernBiotech #0100-01 ) , sealed with nail polish , and imaged on a Nikon Eclipse E800 fluorescent microscope . | The maintenance of a stable genome is crucial to organismal survival . Genome stability is perpetually threatened by spontaneous DNA damage , and DNA repair proteins are required to accurately and efficiently repair DNA damage in ways that minimize genome alterations . Some repair pathways are linked to increased risk of genome changes . One example is repair associated with the production of crossovers between homologous chromosomes . The DNA helicase BLM suppresses genome changes by promoting non-crossover forms of repair; without BLM , spontaneous crossovers , deletions , and genome rearrangements increase . Using Drosophila as a model organism , our studies reveal the complex interactions between BLM and three structure-selective endonucleases with overlapping substrate specificities and partial functional redundancy . Loss of BLM and any one of the nucleases results in severe genome instability , reduced cell proliferation , and , ultimately , death of the animal . Our work suggests that these nucleases differentially rescue the loss of functions of BLM associated with problems that arise during DNA replication , illuminating the complexity of repair mechanisms required to maintain genome stability during replication . Further , our work advances models of replication-associated repair by suggesting specific roles for BLM and structure-selective endonucleases . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"molecular",
"cell",
"biology",
"cell",
"biology",
"chromosome",
"biology",
"genetics",
"molecular",
"genetics",
"biology",
"genomics",
"genetics",
"and",
"genomics"
] | 2011 | Three Structure-Selective Endonucleases Are Essential in the Absence of BLM Helicase in Drosophila |
Rapid assignment of bacterial pathogens into predefined populations is an important first step for epidemiological tracking . For clonal species , a single allele can theoretically define a population . For non-clonal species such as Burkholderia pseudomallei , however , shared allelic states between distantly related isolates make it more difficult to identify population defining characteristics . Two distinct B . pseudomallei populations have been previously identified using multilocus sequence typing ( MLST ) . These populations correlate with the major foci of endemicity ( Australia and Southeast Asia ) . Here , we use multiple Bayesian approaches to evaluate the compositional robustness of these populations , and provide assignment results for MLST sequence types ( STs ) . Our goal was to provide a reference for assigning STs to an established population without the need for further computational analyses . We also provide allele frequency results for each population to enable estimation of population assignment even when novel STs are discovered . The ability for humans and potentially contaminated goods to move rapidly across the globe complicates the task of identifying the source of an infection or outbreak . Population genetic dynamics of B . pseudomallei are particularly complicated relative to other bacterial pathogens , but the work here provides the ability for broad scale population assignment . As there is currently no independent empirical measure of successful population assignment , we provide comprehensive analytical details of our comparisons to enable the reader to evaluate the robustness of population designations and assignments as they pertain to individual research questions . Finer scale subdivision and verification of current population compositions will likely be possible with genotyping data that more comprehensively samples the genome . The approach used here may be valuable for other non-clonal pathogens that lack simple group-defining genetic characteristics and provides a rapid reference for epidemiologists wishing to track the origin of infection without the need to compile population data and learn population assignment algorithms .
Burkholderia pseudomallei , the etiologic agent of melioidosis , is commonly isolated from soil and water in many tropical regions of the world . Endemic foci of B . pseudomallei predominantly include Southeast Asia ( particularly Thailand ) and northern Australia , although this organism is found sporadically in other equatorial regions such as South and Central America , Africa , and the Indian subcontinent [1] . Since infections are most commonly acquired from the environment , genetic differentiation is expected to occur , leading to geographic substructure within the bacterial population . Previous studies have demonstrated that B . pseudomallei populations from the melioidosis-endemic regions in Southeast Asia and Australia are not only geographically distinct but exhibit differences in clinical presentation and genetic features [2] , [3] , [4] . For example , differences in clinical manifestations include parotid abscesses , which are much more prevalent in Thailand ( 15% ) than Australia ( 4% ) . In contrast , genitourinary infections and brainstem encephalitis are more commonly seen in Australia than Thailand ( 15% vs . 2% and 2% vs . <0 . 2% , respectively ) [4] , [5] . Differences in mortality rates also differ greatly between the two endemic regions , with mortality rates of approximately 50% in Thailand compared with <20% in Australia [5] . The difference in mortality rates could reflect differences in virulence but is probably more likely to be due to differences in intensive care provisions between the two regions [6] . Despite these marked differences , none are fully diagnostic for isolates from specific geographical regions . Multilocus sequence typing ( MLST ) [7] is a bacterial genotyping method that involves the comparison of ∼450 bp-long nucleotide sequences from seven housekeeping genes . An MLST scheme has been developed for B . pseudomallei [8] and 699 sequence types ( STs ) from isolates and multiple species ( as of November 6th , 2010 ) populate the public database ( http://bpseudomallei . mlst . net/ ) . These data have shed light on the population structure of this species . It has been previously observed that B . pseudomallei STs from Australia and Southeast Asia are mutually exclusive as phylogenetic analyses show geographically correlated clusters of STs , although these analyses failed to group all samples from either region together [9] [10] . Due to relatively low levels of sequence diversity and high levels of lateral gene transfer among B . pseudomallei isolates [8] , [11] , sequence data from only seven genes are insufficient for robust phylogenetic discrimination [11] , [12] . Pearson et al . therefore used a population genetics approach to determine that B . pseudomallei STs form two distinct populations , conforming to the geographic regions of Southeast Asia and Australia [11] . Despite the phylogenetic limitations of MLST data , this large public database shows potential for population assignment using population genetic analyses . We further evaluate and update the previous population assignments [11] by comparing these results with commonly used assignment algorithms . The program Structure [13] is a Bayesian-based clustering algorithm that has been used to infer population structure within genetically diverse bacteria such as Helicobacter pylori [14] . Comparison of Structure with other population assignment software allowed us to assess the robustness of our population assignments . The B . pseudomallei population assignment results that we provide , along with a probability estimation of each assignment , can be used as a practical and immediate reference for melioidosis researchers interested in identifying geographic origins of B . pseudomallei STs and may serve as a model for other weakly clonal species .
The data used to define populations and evaluate the robustness of population assignments were downloaded from the B . pseudomallei MLST database ( http://bpseudomallei . mlst . net/ ) on January 15th , 2009 . The database consisted of 641 B . pseudomallei STs from 1802 isolates collected over 89 years from 35 countries . Approximately 44% of these isolates were collected in Southeast Asia and 53% in Australia and Papua New Guinea . The data were downloaded again on November 9th , 2010 , in order to provide more updated population assignments and population allele frequencies for all currently known STs . These most recent data consist of 664 STs from 1829 isolates , where 44% of these isolates were collected in Southeast Asia and 53% of the isolates were collected in Australia and Papua New Guinea . More detailed information on the geographical sources of isolates representing each ST can be found in the profiles datasheet in the MLST database . The program Structure [13] ( versions 2 . 2–2 . 3 . 1 due to software updates over the course of this study ) was used to analyze allelic profile data from the original 641 B . pseudomallei STs . Briefly , Structure uses MLST datasets and a Bayesian approach to identify population structure and to assign individuals to populations without a priori population descriptions . A Markov Chain Monte Carlo simulation of 100 , 000 iterations with a burn-in period of 30 , 000 was run to determine the posterior probability of the number of populations ( K ) . Where K = 2–4 , Structure analyses were repeated eight times and the posterior probabilities from each run were averaged . For populations of K = 5–17 , Structure analyses were repeated three times and the posterior probabilities averaged . Fewer repetitions were carried out for these higher K values as previous work suggests that more populations are not well supported [11] . The most statistically supported K value was selected to represent the number of populations among the STs based on the estimated log ( ln ) of the probability of the data ( ln P ( D ) ) , and the variance exhibited by each K value . All simulations were carried out using both the “no admixture” [15] and “admixture” models [16] ( comparison between these two models is shown in Supplemental Data Figure S1 ) . The posterior probability of the data ( ln P ( D ) ) for a given value of K might be expected to peak at the true value of K , however , in our runs there was no definite peak as ln P ( D ) increased slightly with an increase in K . This pattern , along with an increase in the variance of ln P ( D ) is common and has been reported by Evanno and colleagues [17] who suggest that measuring the changes in likelihood is a more accurate method for estimating the true value of K . We therefore used ΔK to determine the optimal K value of the B . pseudomallei populations . The ΔK value corresponds with the second order rate of change of all K values divided by the standard deviations from each K [17] . Calculation of ΔK is shown in Supplemental Data Text S1 . We used both BAPS and Structure results to assess population assignments [18] . BAPS ( version 4 ) is another free software package for Bayesian inference of genetic structure within a given dataset [19] , [20] , [21] , [22] . Using the “clustering of linked loci” module , BAPS determines the log likelihood in 10% increments of different population divisions and subsequently calculates the most likely K value . Thus , unlike with Structure , K is not selected a priori . The likelihood of population assignment for each ST is also calculated by BAPS . For BAPS analyses , we used sequence data from the seven B . pseudomallei MLST loci . The codon linkage model and an upward bound of 20 populations were chosen for the “clustering of linked loci” module . As with Structure , eight iterations were run where K = 2–4 and three iterations were run where K = 5–17 . As there is no empirical measure of determining the accuracy of population assignments , we further assessed Structure and BAPS assignments of B . pseudomallei using MLST data , by comparing individual ST assignments made by Structure and BAPS to the geographic information listed in the MLST database and to the likelihood of assignment into each population as calculated by Genetic Analysis in Excel ( GenAlEx ) v . 6 [23] . We also used GenAlEx to measure the degree of population differentiation among populations defined by Structure and BAPS . GenAlEx is a free Microsoft Excel add-in where datasets can be analyzed and manipulated without the requirement for multiple programs . We used the population assignment method in GenAlEx to determine the likelihood of inclusion in each population for each ST . Unlike Structure and BAPS , GenAlEx requires a priori population designations to define population allele frequencies and subsequently calculate the likelihood of population assignment for each ST . We compared the population assignment results from our Structure and BAPS results to the likelihood of population assignment calculated by GenAlEx . Also , for population defined by Structure and BAPS , we performed analyses of molecular variance ( AMOVA ) to calculate the degree and statistical significance of population differentiation . The number of populations supported by Structure and BAPS are two and three respectively . We therefore used the results from the Structure run with the highest likelihood score at K = 2 and the BAPs run with the highest likelihood score at K = 3 to infer population assignments for each ST . To show the extent of genetic differentiation among these populations , we used GenAlEx [24] to calculate ΦPT , using 999 permutations [23] . In assessing assignment results , we categorized STs according to the likelihood of assignment of each ST into a population by Structure or BAPS , allowing us to evaluate the effect of assignment confidence on discrepancies among programs . To be conservative in our assignment of STs to a population , we suggest that a ST only be considered to be from a given population if Structure or BAPS assigned it to that population ≥95% of the time . As BAPS measures likelihoods in 10% intervals , this threshold is effectively 100% for BAPS . STs assigned to either population <95% of the time were considered “undefined” even though studies using simulated datasets suggest that in some situations , assignment probabilities of >50% may be accurate [18] . We wished to provide researchers interested in B . pseudomallei population genetics with a tool for population assignment in instances where novel STs not included in this study are encountered . To achieve this goal , the frequencies of alleles belonging to STs from each population for >95% of the runs were determined . We also enumerated alleles for STs assigned to a population between 50 and 95% of the time as this measure can be useful for indicating the reliability of an allele for population assignment . Performing MLST on large bacterial collections is a time-consuming task; however , single nucleotide polymorphism ( SNP ) genotyping provides a streamlined way to characterize MLST populations even for recombining species [25] , [26] , [27] . We predicted that SNPs within MLST loci could be used to distinguish between the major B . pseudomallei ST populations . The program ‘Minimum SNPs’ [26] , with incorporated Not-N algorithm [28] , was used to search for a set of highly informative characters among the MLST alignments that could be used to distinguish between a predefined ‘ingroup’ and the remaining ‘outgroup’ population . The 566 B . pseudomallei STs determined by Structure to be assigned to one of the two populations in ≥95% of iterations were tested using the Not-N algorithm , where each population was alternately considered the ‘ingroup’ and all other STs the ‘outgroup’ . Similarly , the 607 B . pseudomallei STs identified by BAPS as belonging to any of the three populations in ≥90% of iterations were tested ( BAPS measures likelihood in 10% increments ) . In an attempt to increase the likelihood of finding a small set of population-defining SNPs , a second ‘Minimum SNPs’ analysis including only the 413 STs assigned to a population in 100% of Structure runs and a third analysis with the 560 STs assigned to a population in 100% of BAPS runs were carried out .
Structure was used to identify and characterize B . pseudomallei populations using MLST allelic profile data from 641 STs . The existence of two B . pseudomallei populations ( K = 2 ) was first proposed by Pearson and coworkers [11] as higher values of K did not break apart the two main populations and subdivisions were inconsistent between runs . Here , we confirm that when using Structure , two populations ( K = 2 ) garners the most statistical support when compared to other numbers of putative populations ( K = 1 , and 3 through 17 ) . This support is based on three criteria that have been used in other studies to justify selected K values . First , higher values of K retained the two populations ( Figure 1 ) [11] . Second , the selected K value has the lowest variance of ln P ( D ) after K = 1 ( Figure S2 ) [13] . Lastly , the ΔK shows a peak at the selected K value ( Figure S2 ) [17] . We also tested both ‘admixture’ and ‘no admixture’ analyses and obtained the same results regarding the size of K and similar results regarding population assignments for individual STs . However , the ‘no admixture’ method provided more consistent results than the ‘admixture’ approach , yielding lower variances . The results presented here are from the “no admixture model” ( see Figure S1 for a comparison of these tests ) . Using a K = 2 with Structure , the two populations were significantly distinct ( ΦPT = 0 . 123; P = 0 . 001 ) . Structure assigned 88 . 3% of STs to either Population 1 or Population 2 with ≥95% probability of assignment , with 44% and 44 . 3% of STs assigned to Population 1 and 2 , respectively ( Figure 1 ) . Population 1 is comprised of 95% Australian ( Australia and Papua New Guinea ) , 3% Southeast Asian , and 2% STs from the other parts of the world . In contrast , 89% of STs in Population 2 are from Southeast Asia , 1% from Australia , and 10% from the rest of the world ( Figure 1 ) . Only 11 . 7% of STs were not assigned to a given population based on a 95% probability of assignment threshold . This “undefined” group is comprised of STs from Southeast Asia ( 59% ) , Australia ( 25% ) , and the rest of the world ( 16% ) . We also used the population-clustering program BAPS for determining the number of B . pseudomallei populations and for assigning STs to each population . Unlike Structure we used concatenated MLST sequence data rather than the allelic data used in Structure . In BAPS , the estimated number of populations with the most statistical support was K = 3 rather than K = 2 determined by Structure . This third population defined by BAPS appears to be a sub-population of the previously identified Population 2; however , other than a mostly Asian origin , we found no geographic or epidemiological correlation among these subdivided Population 2 STs . We therefore refer to these two BAPS Asian populations as Population 2a and Population 2b . Evidence of this population subdivision was also observed in Structure when K = 3 ( Figure 1 ) ; however in Structure , both Population 1 and Population 2 were alternately subdivided depending on the run and assignments of STs to either sub-population were inconsistent . In BAPS however , Population 2 is consistently subdivided and ST assignments are consistent among runs . Therefore , it is possible that further sub-structure exists in the B . pseudomallei populations , but remain unresolved due to the limitation of having only seven MLST loci , which may not provide the genetic resolution to detect further subdivision . We compared the population assignments made by the run with the highest likelihood from Structure ( K = 2 ) and BAPS ( K = 3 ) ( Figure 2 ) . As BAPS Populations 2a and 2b are essentially subpopulations of Structure Population 2 , we searched for discrepant STs assigned to Population 1 with >50% likelihood by one program and Population 2 with >50% likelihood by the other . Of the 29 discrepancies ( Figure 2B ) , 16 were assigned by either program with a confidence level ≥95% ( one ST was assigned by both programs with a confidence level ≥95% ) . As a further measure of assignment accuracy , we compared these 16 discrepant STs to the geographical data listed in the MLST database . Eight of the nine discrepancies assigned to a population ≥95% using Structure matched the geographical data listed in the MLST database . For the discrepancies assigned to a population ≥95% with BAPS , 3/8 originated from the geographical region of the population assigned by BAPS . Even though the listed geographic source of a ST is not a perfect indicator of population , it is possible that both programs make assignment errors even when confidence values are >95% , however such errors are probably rare . The geographic sources of STs that comprise each BAPS population are shown in Figure 2C . To further evaluate Structure and BAPS assignments , we used GenAlEx to calculate the likelihood of assignment of each ST in each population . When STs with high probabilities of assignment using either Structure or BAPS were analyzed with GenAlEx , a more distinct differentiation of populations could be seen ( Figures 3 & 4 ) and the likelihood calculations from GenAlEx placed only a few STs in a different population than Structure or BAPS . As expected , differentiation among populations eroded ( reflected in a decline of ΦPT values ) and the number of discrepancies between either Structure or BAPS and GenAlEx increased as STs with lower assignment probabilities from Structure or BAPS were analyzed with GenAlEx ( Figures 3 & 4 ) . When only STs with 100% probability of assignment in Structure were analyzed with GenAlEx , there was only one discrepancy ( ST339 ) . We confirmed that ST 399 is an environmental isolate from the Darwin region of the Northern Territory , Australia . Structure assigned this ST to Population 1 , as expected , but was given a higher likelihood of belonging to Population 2 by GenAlEx ( Figure 3A & 3D ) . When STs with ≥95% probabilities of assignment with Structure were analyzed with GenAlEx , there were eight discrepancies . These discrepant STs clustered with STs from Population 1 , despite log likelihood values from GenAlEx that suggested they belonged in Population 2 , albeit with little difference in log likelihood values ( Figure 3B ) . The geographic sources of these eight discrepancies suggest that only one ST may have been erroneously assigned by Structure; specifically , ST660 is from rain water in Hong Kong and would be expected to be in Population 2 , whereas the other seven were from Northern Australia which is consistent with their position within Population 1 . As STs with decreasing probabilities of assignment with Structure were analyzed with GenAlEx , the number of discrepancies increased slightly , except for a large increase when all STs were analyzed ( Figure 3 ) . Interestingly , more discrepancies occurred with Population 1 than Population 2 . It has been previously observed that the Southeast Asian B . pseudomallei population ( i . e . Population 2 ) has high levels of recombination but low allelic diversity , due to a monophyletic introduction of B . pseudomallei into Southeast Asia . In contrast , the Australian population appears to be paraphyletic with greater allelic diversity in spite of lower recombination between STs [11] . Therefore , the greater diversity of Australian alleles may make Bayesian assignment of STs into Population 1 more complex than Population 2 . Our cut-off value of ≥95% is likely to result in very few erroneous assignments using Structure . Indeed , ST660 is the only potentially inaccurate assignment that we identified at this cut-off value . When GenAlEx was compared against the BAPS K = 3 dataset , there were 14 discrepancies when only STs with 100% probability of assignment were analyzed with GenAlEx ( Figure 4A & 4D ) . For only one of these discrepancies ( ST514 ) , the assignment by BAPS into Population 2a is not consistent with the geographic origin listed in the MLST database ( Australia ) , representing a potentially erroneous assignment by BAPS . Four STs were assigned to Population 2a by one program and Population 2b by the other . As geographic correlates for these two populations are unknown , it is impossible to determine which assignment is more likely . For the remaining nine discrepancies between BAPS and GenAlEx , the geographic origin listed in the MLST database is consistent with the BAPs population assignment . When STs at the ≥90% assignment probability with BAPS were analyzed with GenAlEx , there were 19 discrepancies . Two of these discrepancies ( ST 514 and ST 660 ) are likely erroneous assignments by BAPS into Populations 2a and 1 respectively as their geographic origins as listed in the MLST database are Australia and Hong Kong , respectively . The number of discrepancies continues to rise as more STs are analyzed and the threshold for inclusion drops to ≥50% assignment probability with BAPS . At all levels of assignment probability by BAPS , most discrepancies involved assignments by BAPS into Population 1 while few discrepancies occurred with STs assigned by BAPS into Population 2b . This is similar to the pattern of discrepancies found with Structure assignments . This observation suggests that assignments into Population 1 are the most challenging while assignments into Population 2b are least difficult and probably more robust . In comparison to the Structure-GenAlEx comparisons , there were more overall discrepancies for GenAlEx and BAPS; however , this was expected as BAPS is splitting STs into three populations rather than just two . In addition to evolutionary dynamics and computer algorithms , discrepant population assignment of certain STs can occasionally be attributed to database errors . Indeed , it has been shown that the listed origins for some B . pseudomallei STs are not always accurate due to curation difficulties or by not being able to account for patient travel histories . For example , several isolates recovered in the USA were likely from infections acquired during travel in Southeast Asia [10] . Using our population assignment data , we have identified and corrected some database errors , however , it is possible that other errors remain . There are discrepancies between Structure and BAPS assignments and the listed origin of a ST in the MLST database . We therefore paid particular attention to those STs where both GenAlEx and the MLST database suggested a different population assignment than Structure ( Figure 3D ) or BAPS ( Figure 4D ) . At the 95% likelihood level for Structure , only one such discrepancy ( ST660 ) exists . Although erroneous attribution must always be considered , it is possible that this ST is derived from a recent , but ecologically established introduction into Hong Kong . Another possibility is that this ST was erroneously assigned by Structure to Population 1 . However , BAPS similarly assigned ST660 to Population 1 albeit with 82% likelihood . At or above the 95% likelihood level , we could therefore find only one potential example of an inappropriate assignment by Structure . At ≥90% likelihood level for BAPS , we found one potential discrepancy when compared to Structure , the MLST database and GenAlEx . Sequence type 514 was assigned by Structure at 100% confidence in Population 1 . However , BAPS assigned ST514 with 100%confidence into Population 2a . The MLST database lists ST514 as being collected from a human source in Australia . Unfortunately , this information does not confirm the origin since travel between Thailand and Australia is prevalent . Whole genome sequencing of this ST will help resolve uncertainties regarding Australian and Southeast Asian population assignments as phylogenetic analyses can be expected to reflect population subdivisions as they have for the Australian and Southeast Asian populations [11] . Of the discrepancies between Structure and BAPS versus GenAlEx , the Structure results were most closely aligned with the geographical origin of STs as listed in the MLST database . However , both BAPS and GenAlEx were able to identify instances where Structure population assignments were inconsistent with the epidemiological data , indicating that no single program was 100% effective in B . pseudomallei ST population assignment . Therefore , we suggest , where possible , that Structure and BAPS are used in concert with large epidemiological datasets for highly recombinant organisms to make the most robust population assignments . The addition of more loci and more thoroughly sampling isolates not assigned to either population with high confidence will likely lead to a better understanding of the intricacies of B . pseudomallei population structure . Given the genetic delineation of up to three populations using population assignment software , we hypothesized that a combination of SNPs might be identified that readily differentiate between these B . pseudomallei populations . We used the program ‘Minimum SNPs’ , with incorporated Not-N algorithm [28] , to find population-specific SNPs from both Structure and BAPS defined populations . Using STs with ≥95% population assignment from Structure , we identified a set of 25 SNPs that were needed to discriminate STs from Population 2 from all other STs , albeit with a confidence of only 92 . 5% . In other words , even with a set of 25 SNPs , only 92 . 5% of the Population 2 STs could be distinguished from the Population 1 STs . No addition SNPs could be added by the algorithm to increase the percentage of Population 2 STs that could be distinguished from Population 1 STs . In order to increase the likelihood of identifying a smaller number of SNPs for population differentiation , we narrowed down our population definition by including only STs assigned to each population in 100% of Structure runs . Our results showed that a set of 16 SNPs were needed to separate the Population 2 STs from the Population 1 STs at a confidence level of 97 . 6% . As inaccurately assigned STs can hamper the ability of ‘Minimum SNPs’ to find population specific SNPs , we also used the BAPS population designations at both the ≥90% and 100% thresholds for population assignment . For STs assigned to each population in 90% of BAPS runs , the Not-N algorithm identified a set of 26 SNPs that discriminated Populations 2a and 2b apart from Population 1 with a confidence of 81 . 1% . For STs assigned to a given population in 100% of BAPS runs , a set of 26 SNPs discriminated Populations 2a and 2b apart from Population 1 with 84 . 3% confidence . A set of 21 SNPs discriminated Population 2a apart from Populations 1 and 2b with 95 . 5% confidence while a set of 13 SNPs discriminated population 2b from the others with 99 . 2% confidence . Finally , by analyzing only the Population 2 STs identified at the 100% threshold with BAPS , we found a single SNP ( at position 192 in the narK locus ) that distinguishes all STs in Population 2b ( C nucleotide ) from all STs in Population 2a ( G or T nucleotide ) . These results suggest that complete population identification of all members of all populations by a combination of SNPs from MLST data is not possible . A more recent version of the MLST database was downloaded and used to repeat our Structure and BAPS analyses . Once the analyses on the updated database were complete ( November 6th , 2010 ) these data were compared to the database originally downloaded for this study ( January 15th , 2009 ) . This comparison verified the consistency of Structure and BAPS results between the temporal datasets . Of note , however , is the identification by BAPS of a fourth population consisting of three STs , two of which were included in the original database and were formerly placed in Population 1 . The third ST in this new population ( ST698 ) is novel and is a human isolate from the USA . Because this population appears to be part of the Australian population , we refer to it as Population 1b and the other Australian population as Population 1a . Population assignments and likelihood values for each ST based on the updated MLST database are shown in Table 1 . This table provides a resource that can be used by researchers interested in determining the geographic source population of B . pseudomallei STs . Comparisons with other population assignment methods and with geographic source information listed on the MLST database suggest that the risk of assignment by Structure and BAPS into the incorrect population is low when a high percentage of iterations result in the same assignment . In addition , there appear to be fewer potential errors with STs assigned to Population 2 by Structure and 2a or 2b by BAPS . We therefore suggest that a cut off value of ≥95% ( ≥90% for BAPS ) assignment probability can serve as a conservative threshold above which assignment errors are not likely and which include a large proportion ( ∼90% ) of the entire ST populations . The threshold used by different investigators does not need to be universal , and our recommendation of ≥95% is solely intended as a conservative guide . Indeed , for STs assigned to Population 2 ( or 2a/2b ) , which is a monophyletic population , it is likely that a lower threshold of even ≥60% assignment probability is not likely to result in erroneous assignments . While we present here a list of STs and the likelihood of assignment into each population , we recognize that new STs will be found with future sampling , limiting the long-term utility of our analyses . However , due to the relatively low diversity and high recombination rates relative to mutation in B . pseudomallei [11] , it is likely that many new STs will not contain novel alleles , but rather will comprise new combinations of characterized alleles . As population assignments with Structure are based on allele frequencies in a population , we include this information here with the expectation that this resource will continue to be useful even as novel STs are discovered ( Figure S3 and Table S1 ) . We suggest that alleles that are predominantly associated with population 1 or population 2 can be used to estimate population assignment for novel STs . Of 50 randomly selected STs , all but three could be assigned based on the presence of alleles predominantly associated with one population ( ≥95% of their occurrence is attributed to one population ) . These three STs do not have a high affinity to either populations as all were originally assigned with <95% confidence by Structure and BAPS . Of the 664 B . pseudomallei STs , 80% have alleles that are exclusively found in one of the two main populations and 93% have alleles that are associated with one of these populations in ≥95% of their occurrences . Thus for new STs , allele frequency data can shed light on appropriate population assignments . As lateral gene transfer is increasingly found to play an important role in the population dynamics of a range of bacterial species , population genetics tools such as Structure and BAPS will become more widely used by epidemiologists . The approach described here facilitates rapid assignment of isolates to established populations without needing to compile data , or learn and run a new application . Population assignment is one of the first steps in epidemiological tracking of disease and can be used to identify and track bacterial introductions into new regions . We have expanded on our previous work [11] by rigorously exploring the composition of the two major populations of B . pseudomallei . Our results suggest that the programs Structure and BAPS are both sensitive and accurate for population assignment of B . pseudomallei using MLST data , as the two programs provide similar results . The relative rate of recombination to mutation at MLST loci for B . pseudomallei is higher than for any other bacterial species yet reported [11] , meaning that allele frequency differences among populations is an appropriate method for determining population structure . Examining allele frequencies when deciphering population structure is standard for eukaryotes , where high recombination rates cause allelic frequency differences among populations through genetic drift [29] . Population assignment is an important aspect of epidemiological and forensic attribution . As knowledge of population dynamics and geographical distribution of a species increases , attribution can be attempted at an increasingly fine scale , allowing investigators to focus their attention on a very small and well-defined population and geographic region . For B . pseudomallei , little is currently known about population dynamics , evolution and even geographical distribution . High relative rates of recombination to mutation complicate attempts to discern population structure for this species using strictly phylogenetic approaches . MLST analyses are popular for bacterial pathogens and the large data set collected for B . pseudomallei has allowed for the robust identification of two main populations that correspond to the endemic geographical regions of Southeast Asia and Australia . While substructure within these two populations likely exists , such as the third population identified by BAPS , the seven MLST genes and the current set of STs do not provide enough resolution for further robust differentiation among subpopulations . Genotype interrogation at more loci or great numbers of STs will increase our knowledge of subpopulation dynamics , but in the meantime our current ability to differentiate between the two or three major populations is an important first step for epidemiological attribution . Increasing knowledge of the geographic distribution and population structure of B . pseudomallei STs form the foundation for future work on the evolution , population dynamics and geographical distribution of subpopulations of this bacterium . | Burkholderia pseudomallei is a soil-dwelling bacterium that can infect a large range of hosts . In humans , B . pseudomallei causes melioidosis , and typical routes of entry include open wounds , inhalation , or ingestion . Clinical features are diverse , although pneumonia and abscess formation are common . High rates of recombination within the genome of this bacterium have confounded attempts to match clinical samples to geographically defined populations . Here we provide a reference that simplifies source attribution issues . We applied population assignment software to previously generated sequence data from seven B . pseudomallei genes to define the major geographic populations within this species . We evaluated the robustness of our results by comparison with two additional population assignment programs . We present the likelihood that each variant is assigned to a particular geographic population . This information can be used to assign novel B . pseudomallei isolates to a geographic population without needing to learn and run cumbersome population assignment applications . This method can also be used for other bacteria that are difficult to source-attribute due to high levels of genomic variation and recombination . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"and",
"Discussion"
] | [
"emerging",
"infectious",
"diseases",
"population",
"genetics",
"biology",
"evolutionary",
"biology",
"microbiology",
"bacterial",
"pathogens"
] | 2011 | Epidemiological Tracking and Population Assignment of the Non-Clonal Bacterium, Burkholderia pseudomallei |
Cyclosporin A ( CsA ) has important anti-microbial activity against parasites of the genus Leishmania , suggesting CsA-binding cyclophilins ( CyPs ) as potential drug targets . However , no information is available on the genetic diversity of this important protein family , and the mechanisms underlying the cytotoxic effects of CsA on intracellular amastigotes are only poorly understood . Here , we performed a first genome-wide analysis of Leishmania CyPs and investigated the effects of CsA on host-free L . donovani amastigotes in order to elucidate the relevance of these parasite proteins for drug development . Multiple sequence alignment and cluster analysis identified 17 Leishmania CyPs with significant sequence differences to human CyPs , but with highly conserved functional residues implicated in PPIase function and CsA binding . CsA treatment of promastigotes resulted in a dose-dependent inhibition of cell growth with an IC50 between 15 and 20 µM as demonstrated by proliferation assay and cell cycle analysis . Scanning electron microscopy revealed striking morphological changes in CsA treated promastigotes reminiscent to developing amastigotes , suggesting a role for parasite CyPs in Leishmania differentiation . In contrast to promastigotes , CsA was highly toxic to amastigotes with an IC50 between 5 and 10 µM , revealing for the first time a direct lethal effect of CsA on the pathogenic mammalian stage linked to parasite thermotolerance , independent from host CyPs . Structural modeling , enrichment of CsA-binding proteins from parasite extracts by FPLC , and PPIase activity assays revealed direct interaction of the inhibitor with LmaCyP40 , a bifunctional cyclophilin with potential co-chaperone function . The evolutionary expansion of the Leishmania CyP protein family and the toxicity of CsA on host-free amastigotes suggest important roles of PPIases in parasite biology and implicate Leishmania CyPs in key processes relevant for parasite proliferation and viability . The requirement of Leishmania CyP functions for intracellular parasite survival and their substantial divergence form host CyPs defines these proteins as prime drug targets .
The cyclophilin ( CyP ) protein family consists of highly conserved proteins that share a common signature region of approximately 109 amino acids , the cyclophilin-like domain ( CLD , Prosite access number: PS50072 ) . The CLD carries the peptidylprolyl isomerase ( PPIase ) activity characteristic of CyPs [1] , which has been implicated in protein folding , assembly of multi-protein complexes , and signal transduction [2]–[4] . CyPs are characterized by the binding of the cyclic peptide inhibitor cyclosporin A ( CsA ) , which inhibits the protein phosphatase calcineurin and finds application for example as immune-suppressive drug in organ transplantation [5] . In addition to its inhibitory effect on T cell-mediated immunity [6]–[8] , CsA displays anti-microbial activity against a variety of protozoan pathogens [9]–[11] , including Leishmania [12]–[15] . Parasites of the genus Leishmania cause important human diseases collectively termed leishmaniasis , which range from mild , self-healing cutaneous lesions generated by L . major to fatal visceral infection of liver and spleen caused by L . donovani [16] , [17] . Leishmania is transmitted by infected sand flies , which harbor the proliferating flagellate promastigote form of the parasite . Highly infectious metacyclic promastigotes are inoculated into the mammalian host during sand fly blood feeding , where they are engulfed by phagocytes of the endo-reticular system and develop inside the phagolysosome into amastigotes , which subvert the host immune response and cause the immunopathologies characteristic of the various forms of leishmaniasis [18] , [19] . CsA has been shown to exert a leishmanicidal effect on intracellular L . tropica [12] and L . major in mouse and macrophage infection [13]–[15] . Although these findings define members of the Leishmania CyP protein family as potential important drug targets , only little is known on this protein family in trypanosomatids and the mechanisms of the anti-parasitic effects of CsA on intracellular Leishmania remain elusive . A potential role of Leishmania CyPs in amastigote differentiation and virulence can be postulated based on the role of Leishmania donovani LdCyP in disaggregation of adenosine kinase aggregates [20] , an important enzyme in the Leishmania purine salvage pathway , whose activity substantially increases during the pro- to amastigote differentiation [21] . Furthermore , the amastigote-specific phosphorylation of cyclophilin 40 [22] , [23] suggests that activity , localization , and interaction of this protein may be regulated in a stage-specific manner by post-translational modification . The use of CsA for anti-leishmanial chemotherapy is limited by its suppressive action on host immunity , which leads to aggravation of experimental visceral leishmaniasis [24] . A better understanding on the biology of Leishmania CyPs , and their structural and functional differences to human CyPs is required to pave the way for the identification of new inhibitors with increased specificity for parasite CyPs . Here we initiated a first genome-wide analysis of the Leishmania CyP protein family and used the L . donovani axenic culture system [25] , [26] to investigate the effects of CsA on promastigotes and amastigotes in culture . Our data indicate substantial evolutionary divergence between parasite and host CyPs , which may be exploitable for drug development . We provide evidence for stage-specific functions of Leishmania CyPs in regulation of promastigote cell shape and proliferation , and amastigote thermotolerance . We demonstrate for the first time a stage-specific and direct toxic effect of CsA on host-free amastigotes , validating Leishmania CyPs as drug targets .
Leishmania donovani strain 1S2D ( MHOM/SD/62/1S-CL2D ) clone LdB [27] was maintained at 26°C , pH 7 . 4 in M199 medium supplemented with 10% FCS , 20 mM HEPES pH 6 . 9 , 12 mM NaHCO3 , 2 mM glutamine , 1× RPMI 1640 vitamin mix , 10 µg/ml folic acid , 100 µM adenine , 30 µM hemin , 8 µM biopterin , 100 U/ml penicillin and 100 µg/ml streptomycin . Axenic amastigotes were differentiated at 37°C with 5% CO2 , in RPMI 1640 medium pH 5 . 5 supplemented with 20% FCS , 2 mM glutamine , 28 mM MES , 1× RPMI 1640 vitamin mix , 10 µg/ml folic acid , 100 µM adenine , 1× RPMI 1640 amino acid mix , 100 U/ml of penicillin and 100 µg/ml of streptomycin . Both cyclosporin A ( CsA ) isolated from Tolypocladium inflatum ( Calbiochem ) and FK506 isolated from Streptomyces tsukubaensis ( A . G . Scientific ) were dissolved in absolute ethanol at a final concentration of 10 mM and the stock was stored at −20°C . Log-phase promastiogtes ( 2×106/ml ) were cultured in medium containing either solvent , CsA or FK506 at concentrations ranging from 5 to 25 µM and incubated at 26°C , pH 7 . 4 for 48 hours unless otherwise specified . Axenic amastigotes were differentiated at 37°C for 72 hours and were incubated at a density of 2×106 parasites/ml at 37°C with 5% CO2 , pH 5 . 5 for 48 hours in medium containing either solvent , CsA or FK506 unless otherwise specified . The growth of solvent treated cells controls and CsA treated parasites was determined using a CASY cell counter ( Schärfe System ) or determined microscopically by cell counting using 2% glutaraldeyhde fixed cells . Cell proliferation was determined by CellTiter-Blue assay ( Promega ) according to the manufacturer's instructions . Briefly , 20 µl of CellTiter-Blue was added to the cells in 96-well plate and incubated at 37°C for 4 hours . Fluorescence was measured ( exλ = 560 nm; emλ = 590 nm ) using a spectrometer SP-2000 ( Safas ) . Results were expressed in % of fluorescence intensity compared to solvent treated cells control . The tests were performed in quadruplicate . The sequences of human and Leishmania cyclophilins were retrieved using the UniProt ( www . uniprot . org ) and GeneDB ( www . genedb . org ) databases , respectively , and conserved protein domains were identified by ScanProsite ( www . expasy . ch/tools/scanprosite ) . In order to determine the level of conservation of CLD domains across human and trypanosomatid parasites , all putative CLD containing proteins of the sequenced genomes of L . major , L . infantum , L . barsiliensis , T . brucei , and T . cruzi [28]–[30] were retrieved from the TriTrypDB database ( http://tritrypdb . org/tritrypdb/ ) using HUMAN_PPIA as an initial query for PSI-BLAST . After three cycles , all hits with a significant E-value ( <10E-5 ) and more than 70% coverage of the CLD domain were selected , and their putative CLD domain was then extracted using the alignment to HUMAN_PPIA as a guide . Given the high level of conservation of the CLD domains , it is realistic to consider this dataset as a complete set of the CLD proteins present in the species covered by the current release of TriTrypDB ( Release 1 . 1 ) . These sequences were aligned with T-Coffee ( default mode ) [31] , and a Neighbor-Joining tree was computed with 500 bootstrap replicates . Positions in contact between CsA and cyclophilin A were identified on the multiple sequence alignment and the corresponding columns were extracted . The resulting functional residues were compared and clustered for similarity using UPGMA . We first identified Leishmania CyPs that are predicted to bind CsA using multiple sequence alignment of human and Leishmania major cyclophilins , and assessing the conservation of the residues known to be involved in cyclosporin A binding in known complexes . Based on these criteria , six Leishmania major cyclophilins shared the CsA binding residues with human PPIA or PPID ( LmaCyP1 , LmaCyP2 , LmaCyP4 , LmaCyP5 , LmaCyP11 and LmaCyP40 ) and were selected for further analysis . The leishmanial cyclophilins were modelled with the automated mode of the Swiss-Model tool [32] using the following PDB structures as templates: 2 bit [33] for LmaCyp1; 3eov [34] for LmaCyP2 , 2hqj ( Arakaki and Merritt , unpublished ) , corresponding to LmaCyP11 , for LmaCyP4 and LmaCyP5; 1ihg [35] for LmaCyP40 . For each model or structure , the corresponding putative model complex with cyclosporin A was built based on the complex of L . donovani cyclophilin with CsA ( 3eov ) as a template using the program insightII . Each model complex was subjected to a very limited energy refinement ( 100 cycles with the insightII Discover Module , steepest descent algorithm ) . The 3eov CsA binding residues ( R78 , I80 , F83 , M84 , Q86 , G95 , T96 , A123 , N124 , A125 , G126 , Q133 , F135 , W143 , L144 , H148 ) at less than 4 Å from CsA , were used for the superposition . The subsequent analyses of the 3D model complexes and evaluations of the putative interaction with the CsA were performed with the program insightII . Cell death was assessed by propidium iodide exclusion assay [36] . Briefly , 107 promastigotes or axenic amastigotes from control or CsA treated cultures were washed and resuspended in PBS containing 2 µg/ml of propidium iodide and incubated at room temperature for 15 min in the dark . The stained cells were subjected to FACS analysis ( exλ = 488 nm; emλ = 617 nm ) . 10 , 000 events were analyzed . For cell cycle analysis , 107 late-log phase promastigotes were washed once with cold PBS and resuspended in pre-chilled 90% methanol in PBS and kept at −20°C overnight . The fixed cells were washed once with cold PBS and then resuspended in propidium iodide staining solution ( 10 µg/ml PI , 100 µg/ml RNase A in PBS ) and incubated at 37°C for 30 min in the dark . The stained cells were subjected to FACS analysis as described above . Cell cycle distribution was calculated by FlowJo ( Tree Star , Inc . ) using the Dean-Jett-Fox model . For Giemsa staining , 107 promastigotes or axenic amastigotes were immobilized on poly-L-lysine coated cover slips , fixed with methanol and stained with Giemsa reagent ( Sigma ) according to the manufacturer's instructions . The stained cells were mounted with Mowiol 4-88 ( Sigma ) [37] and observed with a Zeiss Axioplan 2 wide field light microscope . Cells were prepared for scanning electron microscopy as described [38] . Briefly , parasites were washed in PBS , fixed with 2 . 5% glutaraldehyde in PBS , and treated with 1% OsO4 . The cells were then dehydrated and critical-point dried ( Emitech K850 or Balzers Union CPD30 ) and coated with gold ( Joel JFC-1200 or Gatan Ion Beam Coater 681 ) . Samples were visualized with scanning microscope Joel JM6700 F . Indirect immunofluorescence staining was performed with 107 promastigotes or axenic amastigotes that were settled on poly-L-lysine coated coverslips and fixed in methanol at -20°C for 5 min . The fixed cells were rehydrated with PBS , and sequentially incubated with a mouse anti-α-tubulin antibody ( Sigma , 1∶2500 dilution ) and an anti-mouse IgG-rhodamine antibody ( Molecular Probes , 1∶250 dilution ) . Nuclei and kinetoplasts were stained with DAPI and the slides were mounted with Prolong ( Molecular Probes ) . Modified CsA with a primary amine side chain was provided by the Texas A&M Natural Products LINCHPIN Laboratory , Assistant Director Dr . Jing Li [39] . The CsA-amine was coupled to the Affi-Gel®10 resin ( Bio-Rad ) by reaction with the N-hydroxysuccinimide ester groups of the resin . Briefly , 7 . 5 µmol of CsA-amine was mixed with 500 µl of Affi-Gel® 10 and incubated at room temperature for 6 hours . The coupling reaction was quenched by removing the CsA-amine and blocking the unreacted Affi-Gel® 10 with 0 . 2 M ethanolamine . Logarithmic promastigotes were lysed with lysis buffer ( 50 mM HEPES , 100 mM NaCl , 10% glycerol , 0 . 5% NP-40 and 1 mM PMSF ) followed by sonication on ice ( 30 s sonication with 15 s pause for 5 min ) . Insoluble debris was removed by centrifugation . The cleared cell lysate ( 1 mg protein/ml ) was mixed with the CsA-Affi-Gel and incubated at 4°C for 3 hours . Bound proteins were eluted using hot Laemmli buffer . The elution was subjected to 10% SDS-PAGE , stained with SyproRuby®protein gel stain ( Invitrogen ) , and CsA-binding proteins were identified by MS analysis as described [22] and Western blotting . Leishmania major CyP40 was amplified from L . major Friedlin V1 ( MHOM/JL/80/Friedlin ) genomic DNA using the primers 5′-CTCGAGGGAGGAATGCCGAACACATACTGC-3′ ( XhoI site and 2 glycine residues are underlined ) and 5′-GCGGCCGCAACCCTCACGAGAACATC-3′ ( NotI site is underlined ) and ligated to pGEM-T ( Promega ) . The insert was then released by XhoI and NotI and ligated into pGEM-HAstrep . The intermediate construct was digested with BamHI and NotI to release the strep::CyP40 and ligated into pGEX-5X-1 ( Amersham Biosciences ) . The resulting plasmid pGEX-5X-Strep::CYP40 was replicated in E . coli BL21 . Recombinant GST::strep::CyP40 was induced with 0 . 2 mM IPTG overnight at room temperature and then purified with GSH-sepharose and strep-tactin sepharose ( Fig . S1 ) using an Äkta Purifier FPLC system ( Amersham Biosciences ) . Measurements were performed according to [40] . Briefly , the peptidyl prolyl cis/trans isomerization reaction was initiated by diluting the peptide Abz-Ala-Ala-Pro-Phe-pNA in an anhydrous 0 . 5 M LiCl/TFE mixture with 35 mM HEPES pH 7 . 8 . Inhibition of PPIase activity was measured by pre-incubating CsA with the enzyme ( 29 . 5 nM ) for 5 min at 10°C before starting the reaction by the addition of the substrate . Data analysis was performed by single exponential non-linear regression using Sigma Plot Scientific Graphing System . Parasites ( 108 cells ) were lysed in 1× Laemmli buffer ( 1×109 cells/ml ) and vortexed vigorously for 30 seconds . The lysates were sonicated for 1 min on ice and boiled for 5 min . Soluble fractions were collected as protein extracts after brief centrifugation . Twenty microliters of samples ( equivalent to 2×107 cells ) were separated by 10% SDS-PAGE and then transferred to PVDF membrane . Mouse anti-LPG antibody ( clone CA74E , 1∶5000 dilution ) [41] , mouse anti-A2 antibody ( clone C9 , 1∶200 dilution ) [42] , rabbit polyclonal anti-CyP40 ( established using recombinant strep::CyP40 as antigen , 1∶5000 dilution , Eurogentec ) , and appropriate HRP-conjugated secondary antibodies were used to probe the membrane using dilutions of 1∶10000 and 1∶50000 , respectively , and signals were revealed by SuperSignal ECL from ThermoFisher .
PPIases are classified according to the binding of the inhibitors cyclosporin A ( CsA ) and FK506 in two major protein families , cyclophilins ( CyPs ) and FK506 binding proteins ( FKBPs ) , respectively [43]-[45] . A third PPIase family is represented by PpiC/parvulin-like proteins implicated in proline-directed phosphorylation [46] , [47] . Based on the presence of a conserved CyP-type PPIase signature sequence , termed cyclophilin-like domain , CLD , ( Prosite accession number: PS50072 , FY-xx-STCNLVA-x-FV-H-RH-LIVMNS-LIVM-xx-F-LIVM-x-Q-AGFT ) , the Leishmania major genome encodes for 17 cyclophilin-like proteins ( LmaCyPs ) , five FKBP-like LmaFKBPs , and two PpiC/parvulin-like LmaPPICs ( Fig . 1 and Table 1 ) , all of which are conserved in the L . infantum and L . braziliensis genomes ( Fig . 2 ) . According to the current nomenclature [2] , the LmaCyPs were distinguished by numbering from the smallest to the highest predicted molecular weight ( Table 1 ) . Based on length and domain structure , three types of L . major cyclophilins ( LmaCyPs ) can be distinguished . A first group of four proteins ( LmaCyP1–3 , 6 ) is characterized by a single CsA-binding domain without any significant N- or C-terminal sequence extensions ( Fig . 1 and Table 1 ) . A second group of 11 proteins shows significant ( 50 or more amino acids ) N-terminal ( LmaCyP4 , 5 , 8 , 10 , 12 , 16 ) , C-terminal ( LmaCyP7 , 11 ) , or both N- and C-terminal extensions ( LmaCyP9 , 13–15 ) . These extensions are unique and not conserved in human CyPs , but are mostly conserved across other trypanosomatids , including L . infantum , T . brucei and T . cruzi . Exceptions are the C-terminus of LmaCyP13 and the N-termini of LmaCyP8 , 10 , and 14 , which are unique to Leishmania suggesting highly parasite specific functions absent in Trypanosoma . Finally , two LmaCYPs are characterized by the presence of additional functional domains , including LmaCyP5 containing a conserved prokaryotic lipid attachment domain ( PLD , prosite access number PS5125 ) , and LmaCyP40 , the cyclophilin-40 homolog containing two tetratricopeptide repeat domains ( TPR , prosite accession number PS50005 ) known to interact with HSP90 [48]–[51] . We investigated the relationship between human and trypanosomatid CyPs by multiple alignment and cluster analysis using the sequence of the conserved CLD or the functional residues implicated in PPIase function and CsA binding . The clustering tree obtained for the CLD demonstrates that all LmaCyPs have conserved homologs in L . infantum , L . braziliensis , T . brucei , and T . cruzi , which cluster together with highly significant bootstrap values ( Fig . 2A ) . All LmaCyPs have one homologue in the other Leishmania or Trypanosoma species , with the exception of LmaCyP5 , which underwent expansion in the T . brucei genome with five sequentially arranged copies of the gene . It is interesting to speculate that the expansion of the only cyclophilin family member that contains a conserved lipid binding domain may be a reflection of the T . brucei biology , with a potential role for example for the expression of abundant gpi-anchored VSG proteins [52] . Many of the nodes support the existence of CyP subclasses across the trypanosomatids with a significant bootstrap value . In contrast , the nodes that cluster these subclasses together with their human homologues have only poor bootstrap support . This observation suggests that the various classes of CLDs encountered in trypanosomatid cyclophilins are quite distinct from one subclass to another and to their human counterparts . Substantial conservation however was observed in the cluster analysis performed with the functional CyP residues implicated in PPIase function and CsA binding ( Fig . 2B ) . For instance , eight human CyPs and five LmaCyPs are clustering together showing a complete conservation of the canonical signature sequence characteristic for the human CsA-binding protein PPIA ( Fig . 2B and Table 2 ) . This represents a significant conservation when considering that the overall CLD domain is only 64% conserved between the Leishmania and Human CyPs . These results indicate that a subset of Leishmania CyPs are likely functionally conserved and implicated in PPIase function , while other , less conserved LmaCyPs may carry different enzymatic activities . In conclusion , our analysis reveals a large Leishmania CyP protein family suggesting an important role of PPIases in parasite biology , and identifies unique sequence elements in the LmaCyP CsA-binding domains that may be exploitable for drug development . Identification of 5 out of 17 LmaCyPs with a highly conserved CsA binding motif strongly suggests inhibitor-binding to multiple LmaCyPs with potentially important consequences on the biological functions of these proteins and Leishmania infectivity . In the following we investigate this possibility studying the effects of CsA on L . donovani promastigotes and amastigotes in culture . CsA has been previously shown to reduce the intracellular growth of L . major amastigotes [13] , [14] . To further elucidate the mechanisms underlying this inhibition , we investigated the effects of CsA treatment on cultured L . donovani promastigotes and axenic amastigotes . Log-phase parasites from both stages ( 2×106/ml ) were cultured in medium containing either ethanol ( vehicle ) or CsA at concentrations ranging from 5 to 25 µM , and incubated at 26°C , pH 7 . 4 ( promastigote ) or 37°C , pH 5 . 5 ( amastigote ) for 48 hours . At the time points indicated , the cells were fixed and counted microscopically or processed for CellTiter-Blue assay to test for proliferation . CsA-treated promastigotes showed a dose-dependent , progressive reduction of growth with an IC50 at 48 hours between 15 and 20 µM and a more than 5-fold decrease in growth at the highest inhibitor concentration compared to the control ( Fig . 3A and B , left panels ) . Growth reduction was associated with a strong inhibition of resazurin reduction indicating reduced cell proliferation or cell viability ( Fig . 3B , right panel ) . CsA-mediated growth reduction was reversible , as parasite growth resumed normally after removal of the drug after 48 hours of treatment ( data not shown ) . Likewise , CsA had a striking effect on the growth of L . donovani axenic amastigotes . The parasites showed substantially higher susceptibility to CsA at this stage with an IC50 between 5 and 10 µM ( Fig . 3A , right panel , and Fig . 3B , left panel ) , and strongly reduced resazurin reduction ( Fig . 3B , right panel ) . Together , our data demonstrate that CsA interferes with the in vitro growth of both L . donovani promastigotes and axenic amastigotes . In the following we used FACS-based approaches to investigate the mechanisms underlying this growth defect . To elucidate the mechanisms of CsA-mediated growth inhibition , we first investigated the effects of CsA on the viability of treated promastigotes and axenic amastigotes using a propidium iodide ( PI ) exclusion assay [36] . The percentages of PI positive , dead promastigotes and axenic amastigotes after 48 hours of CsA treatment was revealed by FACS analysis . Promastigotes did not show any significant increase in PI positive cells after incubation with CsA ranging from 5 to 15 µM ( Fig . 4A ) , and more than 80% of cells were viable even at 25 µM CsA . In contrast , the proportion of PI positive axenic amastigotes increased dramatically with increasing CsA concentration , with a 4-fold decrease in cell viability at only 10 µM CsA ( Fig . 4A ) . Thus , the decrease in cell number of CsA-treated promastigotes results from a slow-down in proliferation rather than parasite killing . This result was further confirmed by cell cycle analysis . Late-log phase promastigotes were fixed with 90% methanol in PBS , stained with PI , and cell cycle phase distribution was determined by FACS analysis . Treatment of the parasites with 15 µM and 20 µM CsA did not affect the cell cycle distribution ( Fig . 4B ) , suggesting that inhibition of parasite proliferation results from a non-synchronous slow-down in cell cycle progression . CsA-treatment of promastigote cultures induced a striking effect on parasite morphology . We documented these alterations by microscopic analysis using Giemsa staining and scanning electron microscopy . Treatment of promastigotes with 10 to 20 µM CsA induced morphological changes reminiscent of axenic amastigotes , including increased aggregate formation ( Fig . 5A ) , oval cell shape ( Fig . 5B ) , and shortened and retracted flagella ( Fig . 5C ) . The CsA effects on L . donovani promastigotes are reminiscent to parasites treated with the HSP90 inhibitor geldanamycin , which results in amastigote differentiation [53] . We evaluated the effect of CsA on the differentiation state by following the expression of two markers , the promastigote specific surface glycoconjugates lipophosphoglycan ( LPG ) [54] , which is lost during amastigote differentiation , and the A2 protein , which is induced during the pro- to amastigote conversion [42] , [55] . Logarithmic promastigotes were incubated with vehicle alone ( 0 . 15% ethanol ) or 15 µM CsA at 26°C , pH 7 . 4 for 72 hours , and the expression profile was compared to axenic amastigotes by Western blotting using monoclonal anti-lipophosphoglycan antibody CA7AE [41] and anti-A2 antibody C9 [42] . Despite the amastigote-like morphology , CsA-treated promastigotes maintain expression of LPG , comparable to the level of solvent treated cells promastigotes , and do not show induction of the amastigote marker protein A2 ( Fig . 5D ) . CsA treatment of promastigotes at pH 5 . 5 did not result in further differentiation as judged by morphology and expression of LPG , nor did it have an effect on parasite viability ( data not shown ) . These results demonstrate that unlike geldanamycin , CsA induces morphological features similar to amastigotes without inducing the appropriate expression profile . CsA exerts its inhibitory action through binding of CyPs and inactivation of the cellular phosphatase calcineurin by CsA/CyP complexes [8] , [56] . In the following , we used the unrelated calcineurin inhibitor FK506 to analyze if the CsA effects on the parasite are mediated through inhibition of this phosphatase , a test that has been previously applied on Leishmania [57] . Log-phase promastigotes and axenic amastigotes ( 2×106/ml ) were cultured for 48 hours in medium containing either ethanol ( vehicle ) or FK506 at concentrations ranging from 5 to 25 µM , and incubated at 26°C , pH 7 . 4 ( promastigote ) or 37°C , pH 5 . 5 ( amastigote ) . FK506 treatment of promastigotes induced morphological changes similar to CsA treated parasites , and strongly reduced in vitro growth and cell proliferation in a dose-dependent manner ( Fig . 6A and B , left panels , and data not shown ) . Like CsA , FK506 did not significantly affect promastigote cell viability at the lower drug concentrations ( Fig . 6B , left panel ) . In contrast , FK506 treatment of axenic amastigotes did not reproduce the CsA effects . First , as judged by proliferation and viability assay , amastigotes were more resistant to FK506 , with an IC50 between 15 and 20 µM , compared to ca . 7 µM for CsA ( Fig . 6B , right panel ) . Second , unlike CsA , FK506 did not induce massive cell death in amastigotes even at the highest concentration ( Fig . 6C , right panel ) . These data show that CsA and FK506 have different effects on L . donovani axenic amastigotes , which may be due to either stage-specific differences in inhibitor uptake or distinct intracellular cellular targets . Based on previously published observations , Leishmania CyPs may have important amastigote-specific chaperone functions and participate in protein disaggregation [20] . We tested if CsA treatment affects thermotolerance of promastigotes and amastigotes following the number of propidium iodide stained , dead parasites as a read out . Log-phase promastigotes or amastigotes were treated with 15 µM CsA and parasites were simultaneously incubated for various time periods at either 26°C or 37°C . As expected , CsA treated amastigotes showed increased cell death in the presence of CsA during the 20 hours time course experiment ( Fig . 7 , right panel ) . Significantly , CsA-treatment of amastigotes at 26°C completely abrogated the toxic effect of the inhibitor . This data shows that CsA-mediated amastigote killing is temperature dependent . We confirmed this result using the complementary set up , incubating CsA-treated promastigotes at high temperature . Just like amastigotes , CsA-treated promastigotes underwent cell death as soon as 10 hours after temperature shift ( Fig . 7 , left panel ) . CsA alone or heat shock alone had no significant effect on promastigote viability . Thus , CsA affects thermotolerance of both the promastigote and amastigote stages . The effect of CsA on parasite thermotolerance primed us to investigate the potential interaction between this inhibitor and LmaCyP40 , a bifunctional cyclophilin that has both PPIase and co-chaperone function and interacts with members of the HSP protein family through TPR domains [58] . We first used a structural approach applied on six leishmanial cyclophilins selected for their similarity to the cyclosporin A binding pocket of human orthologs . We built the corresponding model complexes with CsA and evaluated their geometric fit and ability to establish inter-molecular hydrogen bonds with the ligand . The experimentally identified CsA binding residues of the L . donovani cyclophilin ( 3eov ) and the putative binding residues of the L . major 3D model complexes , including the one for LmaCyP40 , are highly conserved ( Fig . 8A ) . All models , even if built on different templates , display a root mean square deviation of less than 0 . 6 angstrom on the CsA binding residues of the experimentally determined complex structure . Consequently , all models can accommodate the CsA ligand with no molecular clash and the hydrogen-bonding pattern is conserved with respect to the experimental structure ( Fig . 8A , lower panel ) . Furthermore , manual inspection of the model complexes revealed a good geometric complementarity between the protein and the ligand . All these evidences support the hypothesis that these L . major cyclophilins , including LmaCyP40 , are good candidates for CsA binding . We confirmed binding of the CsA ligand to LmaCyP40 by studying the proposed interaction by affinity chromatography using CsA-loaded resin . L . donovani promastigote extracts were incubated with the resin and bound proteins were separated by SDS-PAGE . One major band , specifically retained on the CsA-resin , was revealed by fluorescent protein gel staining , and identified as CyP2 by MS analysis ( Fig . 8B , left panel , and Dataset S1 ) . Western blot analysis of the gel revealed cyclophilin 40 ( Fig . 8B , right panel ) , thus confirming the CsA-CyP40 interaction suggested by the structural modelling . We next analyzed the biochemical characteristics of the LmaCyP40-CsA interaction using GST::Strep::CyP40 purified from recombinant bacteria ( Fig . S1 ) . We first determined the kcat/Km of Leishmania major GST::Strep::CyP40 PPIase activity by evaluating the linear dependency between kenz and enzyme concentration ranging from 14 . 7 to 59 nM . The catalytic efficiency of Leishmania major GST::Strep::CyP40 for Abz-Ala-Ala-Pro-Phe-pNa was found to be kcat/KM = ( 3 . 725±0 . 16 ) ×105 M−1 s−1 ( Fig . 8C , upper panel ) . We then tested direct inhibition of the LmaCyP enzymatic activity by CsA using the substrate Abz-Ala-Ala-Pro-Phe-pNA and increasing amounts of inhibitor . The IC50 value of CsA was determined to be 162±46 nM CsA ( Fig . 8C , lower panel ) and thus similar to human CyP40 with an IC50 value of 195 nM [59] .
The leishmanicidal activity of CsA has been first demonstrated in L . tropica infected BALB/c mice , which showed a dose-dependent inhibition of parasite burden and reduction in lesion formation [12] . This anti-parasitic activity has been subsequently confirmed for L . major in mouse and macrophage infection assays , and various modes of CsA action have been proposed [13] , [14] , [57] . The observation that CsA has no overt anti-microbial activity against L . major promastigotes in culture , but efficiently kills amastigotes in infected macrophages , provided support to the idea that the toxic effect of CsA on intracellular parasites depends on inhibition of host rather than Leishmania CyPs . This hypothesis was further supported by findings showing that the phosphatase calcineurin , the prime target of the inhibitory CsA/CyP complex , is expressed at very low levels and is not recognized by Leishmania LmaCyP19 ( corresponding to LmaCyP1 according to our nomenclature ) , although this protein efficiently bound CsA [60] , [61] . In contrast to these previous reports , our data provide several lines of evidence for a direct action of CsA on Leishmania CyPs . A first line of evidence resulted from the bio-informatics analysis and structural modeling of Leishmania CyPs . Blast search of the L . major and L . infantum genome databases ( www . genedb . org ) identified a surprisingly large family of 17 CyP-like proteins in these protozoan , compared to yeast , Drosophila , and human with 8 , 14 and 19 CyPs , respectively ( Table 1 , Fig . 2 ) [62]–[64] . Multiple sequence alignment of trypanosomatid and human CyPs , cluster analysis of the functional residues implicated in PPIase catalytic activity and CsA binding of the CLD , and structural modelling revealed the presence of six Leishmania CyPs that showed conservation of the functional residues ( Table 2 , Figs . 2 and 8A ) and were predicted to form a complex with CsA . This remarkable conservation indicates that multiple Leishmania CyPs are likely binding to CsA , a fact that we subsequently confirmed by affinity chromatography and Western blotting , revealing direct interaction of the inhibitor with Leishmana CyP2 and CyP40 ( Fig . 8 B ) . The effects of CsA on L . donovani promastigotes and axenic amastigotes further support this possibility and provided a second line of evidence for a direct action of CsA on Leishmania CyPs in vitro . We showed that inhibitor treatment of L . donovani promastigotes leads to dose-dependent , reversible inhibition of proliferation ( Figs . 3A and B ) , without significant effects on cell viability ( Fig . 4A ) and cell cycle distribution ( Fig . 4B ) . These results confirmed previous observations that CsA does not exert a toxic effect on Leishmania promastigotes , but revealed a strong effect on promastigote in vitro growth that escaped previous analysis , likely due to the lower CsA concentration ( 4 µM ) used in these studies [13] , [14] . In contrast to promastigotes , CsA showed a direct toxic effect on L . donovani axenic amastigotes with more than 50% of parasite death in the presence of 10 µM inhibitor ( Fig . 4A ) . This result demonstrates for the first time that the observed anti-leishmanial effect on intracellular amastigotes in mouse and macrophage infection [13] , [14] , [57] may rely mainly on direct inhibition of parasite CyPs by CsA , although a participation of host CyPs can not be excluded . We further investigated the mechanisms underlying the stage-specific effects of CsA using the unrelated antifungal macrolide inhibitor FK506 . FK506 binds to FKBPs , a second class of PPIases ( Table 1 ) , which similar to the CsA/CyP complexes inhibit calcineurin [8] . FK506 treatment reproduced the effects observed in CsA-treated promastigotes , suggesting inhibition of calcineurin as one of the mechanisms underlying the observed growth defect of this parasite stage ( Fig . 6 ) . To our surprise , unlike CsA , FK506 did not exert a toxic effect on axenic amastigotes at concentrations between 5 and 15 µM ( Fig . 6B ) , a fact previously observed in intracellular L . major amastigotes [57] . These data indicate that the toxic effect of CsA on amastigotes occurs likely through calcineurin-independent mechanisms , which may be directly linked to inhibition of stage-specific enzymatic functions of Leishmania CyPs . Cyclophilins are protein chaperones with PPIase activity , which catalyzes the cis-trans isomerization of peptidyl-prolyl bonds , affecting stability , activity , and localization of client proteins [2] , [65] . Thus , inhibition of CyP functions by CsA may provoke pleiotropic downstream effects that may lead to the observed growth inhibition and loss of viability . In the context of the current literature , two pathways may be singled out with potential relevance for the CsA-dependent toxicity . First , L . donovani adenosine kinase aggregates have been identified as clients for CyP2 , which disaggregates complexes of this protein [20] , [66] , thereby playing an important function in the purine salvage pathway [67] . Inhibition of this important CyP2 chaperone function may limit the intracellular concentration of adenosine and affect DNA synthesis with consequences for promastigote growth and amastigote viability . Second , cyclophilins have been reported to participate in the response to heat stress in other microbial pathogens . In the human pathogenic fungi Cryptococcus neoformans , CsA treatment prevents growth at elevated temperatures [68] , [69] and the CyP-related protein Cp1a is required for full expression of fungal virulence [70] . Our data indeed established a direct link between the sensitivity of Leishmania to CsA and the parasite thermotolerance . We demonstrated that CsA-treated amastigotes are insensitive to the drug when incubated at 26°C , while CsA-resistant promastigotes are efficiently killed by the inhibitor at 37°C ( Fig . 7 ) . A second observation linked Leishmania CyPs with the response to increased temperature . We observed a striking effect of CsA on promastigote morphology , which acquired an oval cell shape and shortened their flagella , thus showing some ( but not all ) features characteristic for amastigote differentiation ( Fig . 5 ) . A similar morphogenic effect has been previously observed on promastigotes treated with the HSP90 inhibitor geldanamycin [53] . It is possible that both CsA and geldanamycin target different proteins are part of the same heat shock complex implicated in Leishmania differentiation and thermotolerance , such as cyclophilin 40 , a multifunctional protein that interacts with various members of the HSP family through conserved TPR domains [58] . Indeed , our data identified LmaCyP40 as a direct target for CsA as judged from the direct interaction between the enzyme and the inhibitor ( Fig . 8B ) and CsA-dependent inhibition of LmaCyP40 PPIase activity ( Fig . 8C ) . It is interesting to speculate that the temperature-dependent CsA effect on Leishmania viability is the result of CyP40 inhibition . Future studies employing LmaCyP40 conditional null mutants with the aim to dissociate the PPIase and chaperone functions of this enzyme may allow testing this hypothesis and shed important new light on the function of LmaCyP40 in parasite thermotolerance and infectivity . In conclusion , our data revealed for the first time a direct cytostatic and cytotoxic effect of CsA on L . donovani in culture . We provided evidence that the stage-specific effects of CsA are governed by independent mechanisms linked to inhibition of calcineurin phosphatase activity in promastigotes , and inhibition of CyP functions relevant for thermotolerance in amastigote . We identified unique sequence elements in Leishmania CyPs and documented a considerable evolutionary expansion of this protein family , compared to other organisms , emphasizing the importance of this class of molecules for trypanosomatid-specific biology . The requirement of Leishmania CyP functions for intracellular parasite survival and their substantial divergence from host CyPs defines these proteins as prime drug targets . The suppressive action of CsA on host immunity and its exacerbating effects on murine toxoplasmosis , trypanosomiasis , and visceral leishmaniasis [24] , [71] , [72] obviously eliminates this drug for anti-parasitic intervention . Hence , the focus of future research should lie on the identification of novel CyP inhibitors that specifically target parasite CyPs without altering the host immune status . | Visceral leishmanisasis , also known as Kala Azar , is caused by the protozoan parasite Leishmania donovani . The L . donovani infectious cycle comprises two developmental stages , a motile promastigote stage that proliferates inside the digestive tract of the phlebotomine insect host , and a non-motile amastigote stage that differentiates inside the macrophages of mammalian hosts . Intracellular parasite survival in mouse and macrophage infection assays has been shown to be strongly compromised in the presence of the inhibitor cyclosporin A ( CsA ) , which binds to members of the cyclophilin ( CyP ) protein family . It has been suggested that the toxic effects of CsA on amastigotes occurs indirectly via host cyclophilins , which may be required for intracellular parasite development and growth . Using a host-free L . donovani culture system we revealed for the first time a direct and stage-specific effect of CsA on promastigote growth and amastigote viability . We provided evidence that parasite killing occurs through a heat sensitivity mechanism likely due to direct inhibition of the co-chaperone cyclophilin 40 . Our data allow important new insights into the function of the Leishmania CyP protein family in differentiation , growth , and intracellular survival , and define this class of molecules as important drug targets . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"computational",
"biology/sequence",
"motif",
"analysis",
"cell",
"biology/cellular",
"death",
"and",
"stress",
"responses",
"infectious",
"diseases/neglected",
"tropical",
"diseases",
"infectious",
"diseases/protozoal",
"infections",
"infectious",
"diseases/antimicrobials",
"a... | 2010 | Cyclosporin A Treatment of Leishmania donovani Reveals Stage-Specific Functions of Cyclophilins in Parasite Proliferation and Viability |
Albeit genetically highly heterogeneous , muscular dystrophies ( MDs ) share a convergent pathology leading to muscle wasting accompanied by proliferation of fibrous and fatty tissue , suggesting a common MD–pathomechanism . Here we show that mutations in muscular dystrophy genes ( Dmd , Dysf , Capn3 , Large ) lead to the spontaneous formation of skeletal muscle-derived malignant tumors in mice , presenting as mixed rhabdomyo- , fibro- , and liposarcomas . Primary MD–gene defects and strain background strongly influence sarcoma incidence , latency , localization , and gender prevalence . Combined loss of dystrophin and dysferlin , as well as dystrophin and calpain-3 , leads to accelerated tumor formation . Irrespective of the primary gene defects , all MD sarcomas share non-random genomic alterations including frequent losses of tumor suppressors ( Cdkn2a , Nf1 ) , amplification of oncogenes ( Met , Jun ) , recurrent duplications of whole chromosomes 8 and 15 , and DNA damage . Remarkably , these sarcoma-specific genetic lesions are already regularly present in skeletal muscles in aged MD mice even prior to sarcoma development . Accordingly , we show also that skeletal muscle from human muscular dystrophy patients is affected by gross genomic instability , represented by DNA double-strand breaks and age-related accumulation of aneusomies . These novel aspects of molecular pathologies common to muscular dystrophies and tumor biology will potentially influence the strategies to combat these diseases .
Muscular dystrophies ( MDs ) comprise a group of inherited disorders , characterized by progressive muscle wasting and weakness , frequently causing premature death due to lack of effective therapies . More than 150 years ago , Edward Meryon was the first to characterize the detrimental “fatty degeneration of the voluntary muscles” in Duchenne MD ( Meryon E . Lancet 2:588 , 1851 ) . Today it is well accepted that the progressive loss of functional muscle tissue and its replacement by adipose and fibrous tissue represent a pathology common to all MDs despite their heterogeneous genetic etiology . Most MDs are caused by gene mutations that lead to absence or dysfunction of structurally and/or functionally important molecules of the muscle fiber [1] , [2] . The sarcoplasmic spectrin-related protein dystrophin is thought to structurally stabilize the muscle fiber sarcolemma by linking the actin-based cytoskeleton to the extracellular matrix via interaction with the dystroglycan ( DG ) -complex . Lack or vast reduction of dystrophin causes severe Duchenne muscular dystrophy ( DMD ) [3] in humans and myopathy in corresponding mouse models , such as the mdx [4] or mdx-3Cv [5] mouse . Mutations in several glycosyltransferase-encoding genes , such as the fukutin related protein ( FKRP ) or LARGE lead to defective glycosylation of the α-subunit of DG . This molecular defect underlies the second most common group of MDs , the so called “secondary dystroglycanopathies” . Numerous other MD-related molecules are not known to directly interact with the DG complex , such as dysferlin or calpain-3 . Defective expression of dysferlin , a ubiquitously expressed 230-kDa transmembrane protein that has been shown to be involved in resealing muscle fiber membranes , causes limb-girdle muscular dystrophy type 2B ( LGMD2B ) or Miyoshi-myopathy in humans [6] , [7] . An inbred mutation in the murine dysferlin ( Dysf ) gene makes the SJL-mouse a naturally occurring animal model for the human dysferlinopathies [8] . Mutations of the CAPN3 gene encoding the muscle-specific calcium-activated neutral protease calpain-3 , a proteolytic switch in muscle remodeling [9] , cause LGMD2A , a MD with a wide clinical spectrum [10] . Again , the corresponding animal model , the Capn3-deficient mouse is only affected by a mild progressive muscular dystrophy [11] . Given the diverse and obviously unrelated functions of these proteins , whose absence or dysfunction causes MDs , a common pathomechanism driving the complex events of parallel muscle regeneration and degeneration and progressive proliferation of fibrous and fatty tissue seen in all MDs is likely but still remains elusive . In the light of the fact that nearly 25 years ago the DMD gene was identified as the molecular basis for Duchenne MD , the lack of causative therapies has dampened earlier therapeutic promises based on the discovery of molecular defects underlying several MDs and underscores the imperative need for a comprehensive understanding of pathology involved in these rare but lethal diseases . When starting to study age-related phenotypes of murine MDs , we have observed the frequent and spontaneous occurrence of skeletal muscle-derived tumors in our colony of C57BL/10-mdx mice , suggesting a tumor-suppressive role of dystrophin in mice . Therefore we extended our studies to other dystrophin mutations , mouse strains , and even to other MD-mouse models for the most frequent MDs in humans , like dysferlin , calpain-3 and Large , respectively . We show that all of these MD-mouse lines are prone to develop mixed soft-tissue sarcomas containing tumor elements displaying histological and molecular characteristics of rhabdomyo- , fibro- , and liposarcoma . These MD-associated tumors share complex , non-random genomic alterations affecting well-known tumor suppressor as well as oncogenes and these cancer signatures are already detectable in dystrophic muscle tissue , independent of the underlying mutation . Consequently , we show that genomic instability and DNA damage are present also in muscle of human MD patients . Collectively , these data strongly support an unprecedented general link between muscular dystrophy and cancer , driven by the accumulation of DNA damage , chromosome copy number aberrations , and finally the origin of cell clones harboring cancer-like mutations in dystrophic muscle tissue . We propose that - similar to pre-neoplastic lesions - the dystrophic muscle is characterized by genomic instability , which contributes to a common hyperproliferative pathomechanism promoting the degenerative process in human MDs and favoring age-related tumorigenesis in the respective mouse models .
During the last two decades we have observed the spontaneous occurrence of soft tissue tumors arising from various skeletal limb and trunk muscles in our dystrophin-deficient C57BL/10 mdx-mouse [4] cohort . These tumors arose in aged mdx mice ( mean age-of-onset: ∼540 d ) with an incidence of almost 40% , whereas we never observed the occurrence of such tumors in our C57BL/10 wild-type mice . In our colony of another dystrophin-deficient mouse line , mdx-3Cv , which lacks both the muscle 427 kDa and non-muscle 71 kDa dystrophin isoforms due to a mutation at the intron-exon 66 junction [5] , we observed the spontaneous occurrence of skeletal muscle-derived tumors indistinguishable from those observed in C57BL/10-mdx mice . However , mdx-3Cv developed skeletal muscle-tumors at a significant older age ( ∼660 d ) and a decreased incidence of only 5% as opposed to our mdx colony . Because we could not figure out if these differences were due to the different genetic backgrounds ( mdx: C57BL/10 , mdx-3Cv: C57BL/6 x B6C3Fe ) or due to the different dystrophin-mutations , we generated two novel mdx inbred strains , i . e . BALB/c-mdx , and C3H-mdx , respectively , and further studied mdx-mice on mixed C57BL/6 x BALB/c and C57BL/10 x B6C3Fe backgrounds . Indeed we observed the spontaneous occurrence of skeletal muscle-tumors also in these mdx-mice , underlining a strain-independent tumor-suppressor role of dystrophin . Mean ages-of-onset , incidences and gender distributions of tumor-formation were strongly strain-dependent , whereby the C57BL/10 background was most tumor-susceptible ( Table 1 ) . The spontaneous occurrence of skeletal muscle-associated tumors in different dystrophin-deficient mouse lines independent of the underlying dystrophin gene mutation supported a candidate tumor suppressor role of dystrophin . In order to learn whether other MD-genes , which are not directly related to dystrophin , might also suppress tumor formation , we studied mice lacking dysferlin ( DysfSJL mutation [8]; Dysf −/− ) , calpain-3 ( Capn3 −/−; knockout [11] ) , or Large ( Largemyd mutation [12]; Large −/− ) . In a colony of Dysf −/− mice inbred onto C57BL/10 ( n = 151 ) , we also observed high incidence ( 23%; male-to-female ratio ∼3:1 ) of age-related sarcomas ( ∼640 d ) , which mainly arose from proximal hind limb muscles . Also for Dysf −/− mice a strain-dependent effect with respect to mean age of sarcoma-onset was detected , which was more than 100 days later ( ∼755 d ) when the mutation was bred on a mixed C57BL/10 x B6C3Fe background , whereas the sarcoma incidence remained unchanged ( 22% ) . Notably , dysferlin-deficiency on the mixed C57BL/10 x B6C3Fe background resulted in a predominant abdominal wall location of sarcomas ( Figure 1A , Table 1 ) . Based on the spontaneous occurrence of skeletal muscle-tumors in mice deficient for the so far molecularly unrelated genes dystrophin and dysferlin , we hypothesized that MD-genes in more general might act as tumor suppressors . To this end , we conducted a life-span study with mice lacking calpain-3 , the animal model for LGMD2A in humans . Indeed , also Capn3 −/− mice developed skeletal muscle-derived sarcomas at an incidence of 5% . Finally , we also observed the rare occurrence of sarcoma formation even in myd mice ( representing a model for a severe congenital MD in humans ) , in spite of their considerably short lifespan ( Table 1 ) . In order to test if dystrophin , dysferlin and calpain-3 have tumor-suppressor effects in vivo , we generated double-mutant mouse lines , i . e . dystrophin-deficient ( mdx ) mice with additional lack of either dysferlin ( Dmd −/− Dysf −/− ) or calpain-3 ( Dmd −/− Capn3 −/− ) . Dmd −/− Dysf −/− mice ( C57BL/10 ) clinically presented with significant weakness characterized by severe dystrophic signs in the skeletal muscle ( R . B . , manuscript in preparation ) , and had a severely reduced life-span of ∼13 months . Remarkably , malignant skeletal muscle-derived sarcomas ( Figure 1B ) constituted the main cause of premature death in this condition . While penetrance was sharply increased in male mice , 63% of which developed sarcomas , a dramatic decrease in tumor latency was observed in both genders , with the mean age-of-onset reduced to ∼390 d ( compared to 540 d in Dmd −/− and 640 d in Dysf −/− ) . The combined effect of Dmd −/− Capn3 −/− in double-knockout mice , which also presented with a severe MD-phenotype leading to a shortened life-span of ∼13 months ( R . B . , manuscript in preparation ) , resulted in spontaneous sarcoma-formation in 44% of the animals with a mean-age of onset of ∼390 days ( Figure 1C ) . Thus , additional loss of both dysferlin and calpain-3 in dystrophin-deficient mdx mice dramatically reduced sarcoma latency ( Figure 1D ) . Because the macroscopic appearances of the skeletal muscle-derived tumors showed areas of different colorings and varying consistencies ( Figure 1E ) , we speculated that this might be due to a mixed composition of diversely differentiated tumor-cell lineages . Indeed , careful histopathological examinations revealed that all tumors independent from the underlying MD mutation ( s ) resembled mixed sarcomas , comprising variably sized coexisting compartments of rhabdomyosarcoma ( RMS ) , fibrosarcoma ( FS ) and liposarcoma ( LS ) , respectively ( Figure 1F–1K ) . Histopathology of RMS mainly presented as embryonic ( ERMS ) or spindle-cell tumors , which expressed myogenic factors to various degrees , such as myogenin ( Figure 1G ) , Myf5 , or desmin ( not shown ) . Also at the ultrastructural level , these tumor compartments were composed of cells with myofibrils , which were partly arranged in sarcomeric manner ( Figure 1L ) . Tumor-compartments identified as FS displayed bundles of collagen fibres and immature , proliferating fibroblasts , which were arranged in a typical herringbone pattern , hereby recapitulating the histopathological hallmark of human FS ( Figure 1F , 1I ) . The third identifiable compartment was LS consisting of lipocytes , which showed both well-differentiated and de-differentiated morphologies . The unifying characteristics of all LS-cells were positivity for lipid staining by Sudan Black ( Figure 1H ) and , moreover , immunoreactivity for Cdk4 ( a human LS biomarker ) ( Figure 1K ) . By electron microscopy , LS-cells characteristically contained numerous fat droplets ( Figure 1M ) . In line with the histological findings , propagated tumor cell cultures also revealed co-existence of different cell types ( Figure 1N ) , most prominently myogenin-positive cells and lipogenic but myogenin-negative cells ( Figure 1O ) , providing further support that tumors arising in MD-mice are mixed-type sarcomas . Because these findings disclosed all MD-tumors as complex , mixed sarcomas , we next studied the expression of select human sarcoma-related genes [13] . Indeed we found increased expression levels for RMS-markers ( Myog , Myl4 , Igf2 , Prox1 ) , a FS-gene ( Vcan ) , and LS-related genes ( Pparg , Myo1e , Hoxa5 , Plau ) , which further established the MD-tumors as mixed sarcomas consisting of RMS , FS and LS-compartments ( Figure S1 ) . In order to characterize the emerging link between MD and sarcoma susceptibility , we investigated genetic lesions in tumors originating in our MD-mouse strains . DNA extracted from solid tumors from Dmd −/− or Dysf −/− mice was subjected to an arrayCGH-based screen ( n = 8 ) , which revealed that the majority of Dmd −/− tumors were characterized by multiple segmental chromosomal changes , chromosome number aberrations , and amplification of loci harboring the Met ( encoding the Met proto oncogene hepatocyte growth factor receptor ) or Jun oncogene , while tumors from Dysf −/− mice typically displayed less genomic instability ( Figure 2A ) . Frequent disruption of the tumor suppressor loci Cdkn2a , encoding p16INK4a and p19ARF , Nf1 , encoding neurofibromin 1 , and Trp53 , together with whole chromosome 8 and/or 15 gains represented key non-random alterations of sarcomas in both MD models . Quantitative PCR ( qPCR ) experiments ( Figure 2B and 2C ) of DNA extracted from tumors of Dmd −/− , Dysf −/− , Dmd −/− Dysf −/− , and Dmd −/− Capn3 −/− mice ( n = 98 ) revealed that these genetic lesions were common but occurred at different degrees , depending on the specific gene defect ( s ) . Frequent amplification of Met or Jun oncogenes was observed in Dmd −/− ( 41% ) and Dmd −/− Dysf sarcomas ( 44% ) . In contrast , amplifications of Mdm2 and/or Cdk4 ( which were additionally tested because of their frequent amplification in human sarcomas , most prominently liposarcomas [14] ) were rare ( <5%; not shown ) . Lesions of the Nf1 gene ( exons 23 and/or 56 ) were more frequently found in Dmd −/− ( 34% ) and Dmd −/− Dysf −/− sarcomas ( 31% ) as opposed to Dysf −/− ( 14% ) or Dmd −/− Capn3 −/− ( 18% ) . Conversely , exon 2 of the Cdkn2a tumor suppressor gene , which encodes parts of both p16INK4a and p19ARF , was reduced in 73% of Dysf −/− tumors whereas ∼50% of Dmd −/− Dysf −/− and Dmd −/− Capn3 −/− tumors carried this deletion . Notably , many of the qPCR-ratios obtained for Cdkn2a and Nf1 were consistent with losses throughout the tumor . In 25% of DNA samples from sarcomas with qPCR values indicating Cdkn2a loss , exon 2 copy numbers were <0 . 2 , which suggested the presence of a homozygous deletion in ∼80% of tumor cells , compatible with an early event in tumorigenesis . Based on the arrayCGH-findings we screened a large cohort of tumors also for chromosome 8 and 15 copy number aberrations . We found gains of either or both chromosomes in the vast majority ( 80% ) of sarcomas . While ∼40–60% of tumors from all MD models displayed gains of both chromosomes , chromosome 8 alone was preferably gained in Dysf −/− and chromosome 15 in Dmd −/− tumors indicating a probable MD-specific preference ( Figure 2C ) . More than 50% of the measured chromosome 8/15 ratios were consistent with gains throughout the tumor , implying the presence of trisomies in more than 90% of tumor cells . This suggested that together with losses at Cdkn2a and Nf1 loci the recurrent duplications of these chromosomes belong to early events in sarcoma development . Thus , we argued that such events might occur in skeletal muscles of MD-mice prior to formation of clinically identifiable tumors . To test this hypothesis , we assessed chromosome 8 and chromosome 15 copy numbers in DNA samples extracted from a panel of typically tumor-prone limb muscles ( n = 101 ) , which were obtained from different animals ( n = 31 ) that were sacrificed at advanced ages ( comparable to the mean age of mice with sarcomas in the respective MD models ) but had not developed visible tumors until then ( Figure 3A ) . We found elevated levels of chromosome 8 and/or 15 in ∼30% of muscles from MD-mice but never in wild-type mice ( Figure 3B ) . Also occasional copy number aberrations of the Cdkn2a , Nf1 , Met and Jun genes were detected in dystrophic muscles ( ∼12% ) . Because the extents of some of these findings were clearly compatible with the presence of malignant cell clones within the tested muscles , we next analyzed these muscles microscopically . Indeed we found variably sized microscopic tumor masses residing between muscle groups and within single muscle fascicles ( Figure 3C ) . Immunohistochemical examination of these tumors in situ revealed intense staining of cell proliferation markers ( p27 , PCNA ) as well as Cdk4 ( Figure 3D ) , compatible with high proliferative activity . These findings clearly showed ( i ) that tumor pre-stages and pre-neoplastic lesions are already present in dystrophic muscle and ( ii ) that the actual sarcoma incidence of MD-mice is much higher than that solely based on the occurrence of visible tumors . Because none of these DNA-abnormalities were present in non-muscle tissues ( i . e . brain , liver , and lung ) we concluded that these somatic aberrations are specific to dystrophic skeletal muscle . To address whether aneuploidy affects also human MD , we analyzed primary myoblast lines from DMD and LGMD2B patients . In myoblast DNA samples from a DMD and a LGMD2B patient as well , arrayCGH revealed profiles indicating borderline gains of several chromosomes . In particular , aberration scores indicated gains of chromosome 19 ( Figure 4A ) . To confirm this finding , interphase fluorescent in situ hybridization ( I-FISH ) experiments were performed on cytospin preparations from early-passage myoblast cell cultures of DMD ( n = 4 ) , and LGMD2B ( n = 3 ) patients , as well as healthy donors ( n = 2 ) . In contrast to normal cells , myoblasts from DMD and LGMD2B patients frequently harbored tetrasomies of chromosome 19 ( 13–27%; Figure 4B ) . Additional analyses for other chromosomes revealed multiple aneusomies , such as tri- and tetrasomies of chromosomes 1 ( 5–8% ) , 2 ( 3–6% ) , and 8 ( 4–20%; Figure 4C ) . In a DMD myoblast cell line , metaphase spreads displayed formation of diplochromosomes ( i . e . pairs of sister chromosomes , generated by endoreduplication ) ( Figure 4D ) , which are indicative for heterogeneous chromosomal instability and aneusomies . DNA content analyses by FACS profiling of propidium iodide-stained cells revealed that myoblasts from DMD and LGMD2B patients contained abnormally high proportions of nuclei with aberrant DNA-content , indicated by prominent G0+ peaks ( Figure 4E ) . Targeted FISH analysis of nuclei isolated through sorting of such G0+ peaks verified the presence of genomes harboring chromosome 8 aneusomies ( Figure 4E , insets ) . Moreover , the occasional presence of micronuclei implied the continual induction of numerical or structural chromosomal damage in MD-myoblast lines . In order to preclude that the observed chromosomal copy number aberrations had been acquired or at least amplified in vitro , as reported for embryonic stem cells [15] and committed progenitor cells [16] , we asked whether aneusomies also represent an in vivo genotype and do exist in skeletal muscle tissue of MD patients . To this end , interphase nuclei from frozen muscle biopsies from human MD-patients were isolated and probed by I-FISH . We detected tri- and/or tetrasomies of chromosomes 2 and/or 19 in ∼5–12% of the nuclei isolated from DMD muscle ( n = 4 ) ( Figure 4F ) . In contrast , counts of chromosome 13 , for which normal copy numbers were found in myoblasts , were readily comparable to control muscles ( Table 2 ) . Similarly , aberrant chromosome 2 and 19 counts were detectable in muscle biopsies from patients with LGMD2A ( n = 3 , CAPN3 mutations ) , LGMD2I ( n = 3 , FKRP mutations ) , as well as LGMD2B ( n = 1 , DYSF mutations ) ( Figure 4G ) . Notably , LGMD2A muscles exhibited slightly aberrant counts also for chromosome 13 ( 4 . 6% versus 1% in controls ) . Generally , poly-/aneusomic nuclei further displayed features like enlargement , more irregular shape , and micronucleus formation , when compared to disomic nuclei . I-FISH signals in nuclei with aneusomic configurations frequently appeared either as highly condensed doublet signals ( in particular for chromosome 19 ) or as bizarre structures with highly elongated conformation , indicating increased variability of differential ( probably abnormal ) states of chromatin condensation ( Figure 4F , 4G ) . In order to learn if the degree of aneusomies correlates with the disease progression of muscular dystrophies , we also studied fetal muscle obtained during autopsy of aborted fetuses with prenatal diagnosis of DMD or MDC1C . Indeed , these fetal muscle tissues contained much less chromosomal copy number aberrations ( chr2: ∼1% versus 0 . 2% in controls; chr13: 0 . 6% versus 0%; chr19: ∼3% versus 1% ) . Thus , compared to age-matched control muscles , we observed an age-dependent increase of the frequency of aneusomic nuclei in MD patients ( Figure 4H , 4I ) . The finding of cancer-like mutations and somatic aneuploidy in dystrophic muscle prompted us to speculate that this might be caused by damage to DNA induced e . g . by oxidative or replication stress . The formation of interstitial deletions and intrachromosomal amplifications , which we found in pre-neoplastic lesions and sarcomas arising in murine MDs , belong to typical genetic aberrations that result from unrepaired DNA double-strand breaks ( DSB ) [17] and represent early events in the development of cancer [18] . To explore whether damage to genomic DNA precedes sarcoma development , we studied the canonical DNA damage response pathways in skeletal muscle from dystrophic mice . When analyzing muscle tissue from Dmd −/− mice , pronounced activation of the two major DNA damage response pathways was observed , characterized by high expression of Ser1981-posphorylated ATM ( p-ATM , ataxia-telangiectasia mutated kinase ) and Ser428-posphorylated ATR ( p-ATR , ATM and Rad3-related ) , and of their downstream signaling targets Chk1 and Chk2 ( not shown ) . We next investigated histone H2A . x , which represents a target of the ATM pathway that signals the presence of DSBs and constitutes a key protein of the DNA damage response by accumulating at large stretches of chromatin surrounding DSBs and recruiting repair factors [19] . In contrast to normal controls , muscle from MD mice was characterized by intense immunoreactivity with an antibody specifically detecting Ser139-phosphorylated histone H2A . x ( γ-H2A . x ) , similar to the reactivity observed in sarcomas ( Figure 5A–5C ) . We then examined the DNA damage response in muscle biopsies obtained from human DMD patients . In contrast to healthy control muscles , γ-H2A . x immunostainings revealed high levels of DSBs in muscle biopsies from all DMD patients ( n = 4 ) tested , with multiple nucleoplasmic foci formation belonging to muscle fibres and moreover to non-muscle cells within the endomysial connective tissue , such as interstitial fibroblasts and endothelial cells ( Figure 5D , 5E ) . We further found that DNA-damage response was already present in pre-pathologic muscle from very young patients ( 9–11 months ) and a DMD fetus , which suggested that DNA-double strand breaks very likely occur prior to clinical onset of muscle weakness , wasting , and the concomitant inflammatory response . Also muscle tissue in samples from LGMD2A ( CAPN3 , n = 3 ) , LGMD2I ( FKRP , n = 2 ) , MDC1C ( FKRP ) , and LGMD2B patients ( DYSF ) exhibited intense γ-H2A . x immunoreactivity and multiple nucleoplasmic foci formation ( not shown ) . That both muscle-fiber nuclei and non-muscle cell nuclei displayed massive γ-H2A . x accumulation prompted us to specifically assess the DSB response in myogenic precursor cells . We investigated primary muscle cell cultures generated from DMD and LGMD2B patients . In contrast to myoblasts from healthy donors , nuclei from DMD and LGMD2B myoblasts showed pronounced accumulation of γ-H2A . x foci ( Figure 5F–5I ) . The formation of distinct nuclear immunofluorescent foci was observed in 49% of cells from DMD and 59% from LGMD2B myoblasts ( compared to 24% in controls ) and the number of cells with multiple ( ≥3 ) foci was also markedly increased ( DMD: 32%; LGMD2B: 45%; controls: 10% ) .
Here we show that different types of MD mouse models develop with increasing age mixed soft-tissue sarcomas ( STS ) , presenting as rhabdo-fibro-liposarcomas . While the spontaneous occurrence of RMS has been previously reported in mdx mice [20] and in addition in mice deficient of α-sarcoglycan [21] ( Sgca −/− , a model for the human LGMD2D ) , this is the first report of sarcomas in mice lacking dysferlin , calpain 3 , or Large . Our work further shows for the first time that also mice lacking dystrophin due to other mutations than mdx and on different genetic backgrounds are prone to develop age-related STS . In contrast to the previous reports , we found that virtually all sarcomas from MD mice histologically present as mixed sarcomas consisting of RMS and of two additional components with fibro- and liposarcomatous differentiation . Macroscopically , sarcomas feature considerable heterogeneity regarding visual appearance and consistency of tumor mass . Similar to the high complexity and histological diversity inherent to human sarcomas , we found it extremely difficult to exactly stage individual tumors due the highly complex and heterogeneous structure and significant sectional plane divergence . Therefore , our finding of mixed sarcomas in mdx and other MD mice rather extends than rebuts the previous reports by Chamberlain et al . [20] , who reported alveolar RMS in mdx , and Fernandez et al . [21] , who described embryonal RMS in mdx and also Sgca −/−mice . As a further difference , sarcoma incidence in our C57BL/10 mdx mice ( 39% ) was clearly higher compared to the previously reported RMS incidences ( ∼6–9% ) . It remains elusive if these differences are due to different housing conditions or other unknown environmental or strain-specific factors . It is , however , remarkable that the three main components of malignant cell-types , i . e . myo- , fibro- , and lipocytes , which we observed in our MD-mouse tumors , correspond exactly to the same cell- and tissue types that are crucially characterized by progressive proliferation in MDs . Thus , the MD-associated proliferation of fat and connective tissue might create the molecular context permitting sarcoma development arising from a multipotent mesenchymal or muscle-derived stem cell . Several observations in our study lend support to the speculative view that MD-genes might have a role as tumor suppressors . We found that strain backgrounds with C57BL/6 proportions obviously exerted protective effects with regard to tumor latency and that tumor penetrance was lower in Dmd −/− mice on C3H or BALB/c backgrounds compared to C57BL/10 . In line with our observation , C57BL/6 is known for its resistance to Ptch1+/−-induced rhabdomyosarcomas [22] . Genetic background also clearly influenced tumor gender specificity in Dmd – mice ( male preference in BALB/c , female in C3H ) and tumor site predilection in Dysf −/− mice ( ∼60% abdominal wall tumors in C57BL/10 x B6C3Fe compared to ∼20% in C57BL/10 ) . Such strain-specific modulation of incidence , latency , location spectrum , and gender preference has been well documented for other cancer models , such as the p53-deficient mouse [23] . The significantly reduced sarcoma latency in double-mutant Dmd −/−Dysf−/−- and Dmd −/−Capn3 −/− mice also resembles a common feature of tumor suppressor mouse models , as exemplified by the synergistic effect of a combined loss of p53 and Nf1 , which accelerates soft-tissue sarcoma development [24] . Thus , the effects we observed for MD-gene losses represent classical credentials of tumor suppressor genes . In support of this view , dystrophin has been linked to human cancer , as its frequent inactivation was shown to be involved in the pathogenesis of malignant melanoma [25] . Notably , in melanoma cell lines dystrophin knock-down enhanced migration and invasion , whereas re-expression attenuated migration and induced a senescent phenotype , fully in line with a tumor suppressor role of dystrophin [25] . Moreover , utrophin , the highly related autosomal paralogue of dystrophin , represents a tumor suppressor candidate , owing to its frequent disruption in human malignant tumors and its capability to inhibit breast cancer cell growth [26] . Notably , aberrations of the DG have been associated with several types of human cancer [27]–[30] , suggesting a potential role also in tumorigenesis . In particular , a tumor suppressor function has been suggested for laminin-binding glycans on α-dystroglycan [31] , whose loss can be caused by silencing of the LARGE gene in several metastatic epithelial cell lines [30] . For both , dystrophin [32] and dysferlin [33] interactions with the microtubule network have been recently described , which suggests their hypothetical implication in microtubule-mediated cell functions , such as mitosis and cell migration . Future studies will be needed to clarify whether MD-genes act as tumor suppressors , which is suggested but not proven by our data . We found that murine sarcomas from MD-mice frequently harbor non-random , recurrent genetic lesions that provide links to human mesenchymal cancers . The pivotal p53 and retinoblastoma ( RB ) cell cycle control pathways were frequently incapacitated by the disruption of the Cdkn2a locus , which encodes two different tumor suppressors , the Cdk4 kinase inhibitor p16INK4a and the Mdm2-p53 regulator p19ARF , both of which play an important role in the development and progression of many human cancer types . Deletions at Trp53 and Nf1 loci established a genetic link to human soft-tissue sarcomas , which are characterized by frequent p53 mutations [34]–[36] , as well as to syndromes associated with increased RMS incidence due to germ-line disruption of these tumor suppressor genes ( Li-Fraumeni , TP53; Neurofibromatosis type I , NF1 ) [37] . More recently , human myxofibrosarcoma and pleomorphic liposarcomas were shown to frequently harbor NF1 mutations [14] . Thus , the disruption of Nf1 in sarcomas from MD-mice parallels specific - non myogenic - subtypes of human soft-tissue sarcomas and suggests a more general role for Nf1-lesions in the genesis of mesenchymal cancers . A high fraction of sarcomas from MD-mice harbored amplifications of the Met or Jun oncogenes . The Met oncogene amplification constitutes a critical path to aberrant activation of the Hgf/c-Met axis , which is known to promote tumorigenesis and to be involved in the progression and spread of multiple human cancers . Amplification of the JUN oncogene has been reported in human liposarcomas [38]–[39] , in sound accordance with herein discovered frequent Jun amplification in MD mixed sarcomas . Our finding of recurrent chromosome 8 and/or 15 gains in MD sarcomas provides a link to other murine cancers . Chromosomes 8 and/or 15 are frequently duplicated in T cell tumors [40]–[41] or transgenic mouse models of acute promyelocytic leukemia [42] , and probably contribute to elevated expression of the Junb and/or Myc oncogenes , as suggested for Myc in T cell lymphomas [41] . Notably , several human malignancies , amongst them myxoid/round cell liposarcoma [14] , are known to harbor recurrent gains of chromosome 8 . Most importantly , the human chromosome 8 harbors multiple regions that are syntenic to both murine chromosomes 8 and 15 , which we found to be regularly gained in the MD-sarcomas . We discovered that the most frequent and most prominent genetic alterations that characterize full-blown skeletal muscle-derived sarcomas are already present in dystrophic skeletal muscle of clinically tumor-free mice . We also demonstrated DNA damage and showed that skeletal muscle of MD-mice harbors microscopic tumor infiltrates prior to the development of macroscopically visible tumors . In particular , our findings suggested that somatic aneuploidy , indicated by recurrent gains of chromosomes 8 and 15 , contributes to sarcoma susceptibility in murine MD . Thus , the frequent occurrence of chromosome 8/15 gains together with specific losses at the Cdkn2a locus might represent early events occurring in cancer pre-stages and promoting malignant transformation [43] . Importantly , these findings also suggested that the actual sarcoma incidence of MD-mice is much higher than that solely based on the occurrence of visible tumors . In the light of our results , sarcoma formation might be regarded as the disease end-stage of a MD in mice . The finding of cancer-like genomic aberrations and DNA damage in the skeletal muscle from MD-mice inspired us to search for such aberrations in skeletal muscle of human MD patients . We focused on DMD and LGMDs caused by DYSF , CAPN3 , or FKRP mutations , representing the most frequent MDs , and found that all of them are associated with somatic aneuploidy and widespread DNA damage in skeletal muscle tissue in vivo . Also in vitro , cultured myogenic stem cells from DMD and LGMD2B patients exhibited DNA damage and aneuploidy . In our study , somatic aneuploidy appeared to be a feature concurring with the outbreak of pathology in dystrophic muscle and to increase with age in human MD patients . In contrast , high levels of DSBs were already evident in fetal muscle from DMD and MDC1C individuals and in muscle biopsies from DMD infants ( <1 year ) , which suggested that DNA damage precedes the clinical manifestation and therefore cannot be solely related to replication stress . While somatic aneuploidy has been reported in multiple human pathologies , such as Alzheimer's disease , this is the first report on gross somatic aneuploidy in MDs . Genomic instability has been reported in laminopathy-based premature ageing [44] , a condition caused by mutations in lamin A/C , notably another MD-related molecule . DNA damage was shown recently in Friedreich's ataxia , a neurodegenerative disorder [45] . Depending on the context , aneuploidy not only can promote tumorigenesis [18] , [46]–[47] , but also can impair proliferation , cause premature replicative senescence [48] , or can even suppress tumorigenesis [49] . Under the assumption that aneuploidy affects cells destined for muscle regeneration and/or function , aneuploidy could therefore represent an important pathological feature causing a propensity for malignant transformation in murine MDs and contributing to tissue malfunction and diminished regenerative capacity in human MDs . Unrepaired DNA damage activates cellular senescence [50] and could therefore be also associated with the known generalized diminished replicative capacity of DMD myoblasts [51] , contributing to the progressive exhaustion of the muscle's regenerative potential [52] . In the murine condition , senescence could also underlie sarcoma susceptibility as secreted senescence associated factors can contribute to a pro-tumourigenic inflammatory environment [50] , thereby promoting the occurrence of age-related cancer [53] . In this context , it will be interesting to study sarcoma formation in mdx mice lacking the RNA component of telomerase ( mdx/mTR ) that have very recently been shown to have shortened telomeres in muscle cells and a severe progressive muscular dystrophy [52] . For the time being , we have no answer for why murine MDs frequently end up in sarcoma formation while in human MD patients increased muscle-tumor susceptibility has not been reported . But it is interesting to note that it is also not fully understood why loss of dystrophin causes a fatal MD in humans while only a mild myopathy in mice . Also , Dysf −/− and Capn3 −/− deficient mice are largely spared the severe symptoms of the patients with LGMD due to defects in these two genes . We speculate , however , that essential differences in tumor biology between men and mice could account for this difference: while humans are prone to epithelial carcinomas , mice commonly develop mesenchymal sarcomas , which might be due to profound differences in telomere biology between the two species [54] . Also , fewer genetic events are required to induce malignant transformation in mice compared to humans [43] , [55] . Collectively , our findings that genetically distinct MDs in mice and humans share a common molecular pathology characterized by DNA damage and genomic instability similar to pre-cancerous lesions suggests the existence of a novel , unifying pathomechanism that might contribute to disease progression through erosion of the replicative capacity of muscle stem cells and could therefore help to explain the common fatal progression of degeneration and wasting in MD . This is a novel aspect , which contributes to our understanding of MD , and moves an orphan disease close to the common disease cancer , thereby hopefully opening novel therapeutic avenues .
Samples for this study were collected from diagnostic skeletal muscle biopsies , which had been conducted in patients assigned for evaluation of musculoskeletal disorders at our department . Patients or their legal guardians gave informed consent for scientific purpose use of left-over tissue samples . DMD patients included in this study had a confirmed molecular diagnosis of DMD , ascertained by lack of dystrophin staining in immunohistochemistry ( IH ) and Western Blot ( WB ) , and in most cases a genetic diagnosis . Muscle biopsy samples used in this study were from a total of n = 6 different DMD patients: M2006 ( age at muscle biopsy: 9 m; DMD gene mutation: c . 3053_3087del ) , M2008 ( 11 m; c . 8669-1G>T ) , M1633 ( 6 a; c . 858T>G p . Tyr286X ) , M1994 ( 7a; unknown DMD mutation ) , M1895 ( 8 a; dup_ex3-7 ) , M1959 ( 15 a; del_ex17 ) , and two samples from aborted fetuses with DMD . LGM2DA patients ( n = 3 ) had a genetic diagnosis and WB exhibited absence of calpain-3 specific bands in muscle tissue: M1883 ( 9 a; c . 550delA p . Thr184ArgfsX36 ) , M2207 ( 13 a; c . 550delA ) , M2219 ( 25 a; c . 1342C>T p . Arg448Cys ) . LGMD2I patients ( n = 3 ) had a confirmed diagnosis by FKRP gene sequencing: M1787 ( 10 a; c . 854A>C p . Glu285Ala ) , M2190 ( 28 a; c . 826C>A , p . Leu276Ile ) , M2166 ( prenatal; c . [962C>A]+[1086C>G] p . Ala321Glu + p . Asp362Glu ) . LGMD2B in one patient was confirmed by reduced dysferlin reactivity in IH and WB , and DYSF gene sequencing: M2057 ( 62 a , c . 509C>A p . Ala170Glu ) . All tissue samples were snap-frozen in dry ice-cooled 2-methylbutane within 1 h after biopsy and stored at -80°C until use . Primary myoblast cultures were obtained from the Muscle Tissue Culture Collection , Friedrich-Baur-Institute , Department of Neurology , Ludwig-Maximilians-University Munich ( Germany ) . DMD: “Essen 88/07” ( 14 a , del45_50 ) ; “72/05” ( 7 a , dup_ex8-29 ) ; “Essen 8/02” ( 4 a , del_ex51-55 ) ; “166/00” ( 6 a , 2bp-deletion in exon 6 ) ; LGMD2B: “90/01” ( 36 a , female , c . [638C>T]+ [5249delG] ) ; “176/01” ( 32 a , male , c . [2367C>A]+ [5979dupA] ) ; “362/03” ( male , 33 a , c . [exon 5 p . Pro134Leu]+ [5022delT] ) ; controls: “363/07” ( 21a , male ) ; “179/07” ( 21a , female ) . Cells were maintained in Ham's F-12 medium supplemented with 15% fetal bovine serum , GlutaMax ( L-glutamine 200 mM ) , glucose ( 6 . 6 mM ) , fetuin ( 0 . 47 mg/mL ) , bovine serum albumin ( 0 . 47 mg/mL ) , dexamethasone ( 0 . 38 µg/mL ) , insulin ( 0 . 2 µg/mL ) , epidermal growth factor ( 10 ng/mL ) , Pen-Strep ( penicillin G 5000 units/mL , streptomycin 5 mg/mL ) , and fungizone ( amphotericine B 0 . 5 µg/mL ) at 37°C in a humidified atmosphere of 5% CO2: 95% air . For experimental purposes , cells were harvested after 3 or 4 passages . DNA was isolated using the QIAamp DNA Mini Kit ( Qiagen , Hilden , Germany ) according to the manufacturer's recommendations . RNA was isolated using TRI Reagent ( Sigma-Aldrich , St . Louis , MO ) . Cells were stained with BD Cycletest Plus , DNA Reagent Kit for DNA content analysis by flow cytometry ( BD Biosciences , San Jose , CA ) . Mice stocks were maintained at the Division for Laboratory Animal Science and Genetics ( Medical University Vienna , Himberg , Lower Austria ) under institutionally approved protocols for the humane treatment of animals . Mice were cared for in our facilities under conventional housing conditions and received food and tap water ad libitum . Mice presenting with weakness received intensive care , were fed with food pellets soaked in tap water , and were examined daily . Aged mice were checked daily for the development of tumors . In general , tumors were characterized by rapid growth , necessitating the killing of affected mice within few days after visual identification of sarcomas . The Dmdmdx C57BL/10 ( mdx ) , Dmdmdx-3cv C57BL/6 ( mdx-3cv ) , SJL Dysfim ( SJL-Dysf ) , and B6C3Fe Largemyd ( myd ) mice were originally obtained from The Jackson Laboratory ( Bar Harbor , ME ) . To study dystrophin deficiency on other strains , we inbred the Dmdmdx mutation to C3H and BALB/c ( >20 consecutive backcross generations; residual heterozygosity <0 . 01 ) . Some mdx mice were maintained on a mixed C57BL/6 x BALB/c background . Further , we inbred the SJL-Dysf mutation onto the C57BL/10 background , where a prolonged life span compared to SJL was observed , which enabled us to study late-onset stages of dysferlin-deficiency . Capn3 knockout mice ( Capn3tm1Jsb ) [11] on the 129/Sv x C57BL/6 background were obtained from Isabelle Richard and crossed to C57BL/6 mice . To generate Dmd −/− Dysf −/− and Dmd −/− Capn3 −/− double-mutants , mdx mice were crossed to SJL-Dysf ( C57BL/10 ) and Capn3 knockout mice . Some mdx C57BL/10 , mdx-3cv C57BL/6 , as well as SJL-dysf C57BL/10 mice were crossed to B6C3Fe myd mice . Of these animals , only Large+/− and Large+/+ mice were used for analysis of tumorigenesis , which were indistinguishable from pure mdx , mdx-3cv , and SJL-dysf mice , respectively . In all cases , Large+/− heterozygosity had no influence on tumorigenesis and conferred no overt additional phenotype with regard to muscle pathology . After sacrifice by cervical dislocation , mice were dissected , and muscles , other tissues , and ( where applicable ) tumors were excised and snap-frozen in dry ice-cooled 2-methylbutane . All samples were stored at -80°C . Using sterile techniques , parts of excised tumors were washed in PBS , cut into small pieces , and cultured in primary medium , containing DMEM ( Dulbecco's Modified Eagle's Medium , 4 . 5 g/L glucose; PAA Laboratories , Pasching , Austria ) , 20% fetal bovine serum ( FBS “GOLD” Origin: USA; PAA Laboratories ) , 200 U/l PenStrep ( Penicillin , Streptomycin; Lonza , Cologne , Germany ) and 2 . 5 µg/ml Fungizone ( Gibco , Invitrogen Ltd , Paisley , UK ) . After sporadic adhesion of tumor cells , remaining tissue parts were removed , and the primary medium was replaced by growth medium ( DMEM , 20% FBS , 50 U/l PenStrep ) . To activate differentiation , FBS was replaced by 2% horse serum ( PAA Laboratories ) . DNA/RNA extraction was performed as described above for cultured myoblasts . DNA was isolated from serial 5 µm-cryosections prepared from dissected skeletal muscle ( ∼5 mg ) or tumor ( ∼10 mg ) specimens . Reference sections from the sampling procedure were HE-stained for histomorphological examination . Sections were stored at −80°C and then subjected to tissue lysis and nucleic acid purification according to the QIAamp DNA Mini Kit protocol ( Qiagen ) . Mouse tail DNA was isolated using the same protocol starting from lysates prepared by directly lysing 2–3 mm tail tips . DNA concentrations were measured using the NanoDrop spectrophotometer ( Peqlab , Erlangen , Germany ) , DNA samples were diluted ( 10 ng/µl ) and stored at −20°C until use . RNA was extracted from serial 10 µm-cryosections by lysis in 1 TRI Reagent ( Sigma-Aldrich ) , chloroform extraction , and precipitation with isopropanol . RNA samples were measured by spectrophotometry ( NanoDrop ) and quality controlled using BioAnalyzer LabChips ( Agilent Technologies , Santa Clara , CA ) . Cryosections were stained with haematoxylin and eosin ( HE ) . Sudan Black B was used for lipid staining . For immunohistochemistry , 10 µm cryosections were fixed using 3 . 7% paraformaldehyde ( 5 min ) , treated with 0 . 1% Triton-X100 ( 5 min ) , rinsed in PBS , and subsequently incubated with primary antibodies . For immunocytochemistry , cytospins were prepared from cell suspensions and subjected to methanol/acetic acid ( 3:1 ) fixation before antibody incubation . Primary antibodies used in this study were as follows: Myogenin ( Santa Cruz Biotechnology , CA; sc-576 ) , Myf-5 ( sc-302 ) , desmin ( Millipore , Billerica , MA , MAB3430 ) , Cdk4 ( sc-260 ) , PCNA ( sc-7907 ) , p27 ( sc-776 ) , p-Ser1981-ATM ( Cell Signaling Technology , Danvers , MA; #4526 ) , phopsho-Ser428-ATR ( #2853 ) , p-Ser296-Chk1 ( #2349 ) , p-Thr68-Chk2 ( #2661 ) , p-Ser139-Histone H2A . X ( #9718 ) . Secondary antibodies were conjugated to Alexa-Fluor 488 , Alexa-Fluor 594 ( Molecular Probes Invitrogen , Carlsbad , CA ) , Cy3 ( Dianova , Hamburg , Germany ) , or to horseradish peroxidase . Where indicated , immunostained sections were counterstained with 4′ , 6-diamidino-2-phenylindole ( DAPI ) and then analyzed by confocal microscopy using either Olympus Fluoview or Zeiss Axioplan2 microscopes . Fully automated software-assisted quantification of DNA damage ( γ-H2A . x foci ) in myoblasts was performed using the software Metafer ( MetaSystems , Altlussheim , Germany ) . Graphical representations ( plots of fluorescence intensity versus foci numbers ) were generated in R . Matched pairs of sarcoma and tail-tip DNA ( as reference ) samples from the same mice were analyzed using the Agilent mouse genome CGH 44K ( design ID 015028 ) and 244K ( 014695 ) oligonucleotide microarrays ( Agilent Technologies ) . Human myoblast DNA samples were analyzed on Agilent human genome CGH 44K arrays ( 014950 ) , using as reference human genomic DNA from multiple anonymous male donors that was purchased from Promega ( G147A; Madison , WI ) . Labeling and hybridization procedures were performed according to the instructions provided by Agilent . In brief , 200 ng of test and reference DNA were digested with AluI and RsaI ( both Promega ) and then subjected to differential labeling by random priming with incorporation of either Cyanine 3- or Cyanine 5-dUTP ( PerkinElmer , Waltham , MA ) using the BioPrime Array CGH Genomic Labeling System ( Invitrogen , Carlsbad , CA ) . After purification with Microcon YM-30 centrifugal filter units ( Millipore ) , the labeled products were combined , mixed with blocking agent , Hi-RPM hybridization buffer ( both included in the Oligo aCGH/ChIP-on-Chip Hybridization Kit , Agilent ) , human Cot-1 DNA ( Roche Diagnostics , Mannheim , Germany ) or mouse Cot-1 DNA ( Invitrogen ) , and hybridized onto respective microarray slides . Hybridization was carried out for 48 h at 65°C in a hybridization oven . Slides were washed according to the protocol by Agilent , scanned using the Agilent Technologies Scanner G2505B and analyzed using the Feature Extraction and Genomic Workbench 5 . 01 ( formerly DNA Analytics 4 . 0 ) software . To screen for Met ( chromosome [chr] 6 ) and/or Jun ( chr 4 ) as well as Cdk4 and/or Mdm2 ( chr 10 ) oncogene amplification , tumor DNA samples ( 25 ng ) were subjected to a quantitative endpoint PCR , consisting of 0 . 4 µM each primer , 0 . 2 mM dNTPs , 1 . 5 mM MgCl2 , ( NH4 ) 2SO4-containing amplification buffer , and 0 . 5 units Taq DNA polymerase ( reagents from Fermentas , St . Leon-Rot , Germany ) . Competitive co-amplification of internal control targets ( with similar amplicon size ) allowed the unambiguous determination of ≥4-fold amplification levels . Primer sequences ( 5′->3′ ) were as follows: Met_f AAC TGT TCT TGG AAA AGT GAT CGT; Met_r TTT GAA ACC ATC TCT GTA GTT GGA; S100a8_f CGT TTG AAA GGA AAT CTT TCG TGA; S100A8_r TAT CCA GGG ACC CAG CCC TA; Jun_f AAA GCA GAC ACT TTG GTT GAA AG; Jun_r CGC TAT TAT AAA TAT GCA CAA GCA A; Mdm2_f CAT CGC TGA GTG AGA GCA GA; Mdm2_r AAG ATG AAG GTT TCT CTT CTG GTG; Cdk4_f AGT TTC TAA GCG GCC TGG AT; Cdk4_r TCT CTG CAA AGA TAC AGC CAA C; Lig3_f AGG AGA GAA GCT GGC TGT GA; Lig3_r AGC TTT CCT TCC TCT TTG CC . After cycling ( 3 min 95°C , 30× [40 sec 95°C , 40 sec 60°C , 1 min 72°C] , 3 min 72°C ) , 5 µl aliquots of reaction products were analyzed on ethidium bromide-stained 1 . 5% agarose gels and quantified from captured images using Image J . Relative Met , Jun , Cdk4 , and Mdm2 copy number levels were calculated by normalization to the internal standard ( S100a8 on chr3 and Lig3 on chr11 , respectively ) . Tumor samples with copy numbers indicating oncogene amplification were also subjected to verification by real-time PCR ( see below ) . Deletions at the Cdkn2a ( chr 4 ) and Nf1 ( chr 11 ) loci were measured using a quantitative real-time PCR ( qRT-PCR ) SybrGreen assay ( ΔCt method ) , involving separate amplification of target genes and an internal reference ( Lig3 ) . Primers were designed for Cdkn2a exon 2 , which encodes parts of both p16INK4a and p19ARF . CGH 244K data from one Dysf -/- tumor revealed a compound loss at the Nf1 locus consisting of a large ∼0 . 4 Mbp deletion encompassing the whole gene and a smaller ∼42 kb deletion spanning exons 9–28 . To screen for Nf1 deletions in other tumors , two different exons ( 23 and 56 ) were chosen as qRT-PCR targets . Primer sequences were as follows: Cdkn2a_f: GTA GCA GCT CTT CTG CTC AAC TAC; Cdkn2a_r AAT ATC GCA CGA TGT CTT GAT GT; Nf1_I22_f TGA TGA AGT AGT TTG CCA TTG TTT; Nf1_E23_r TTG CCA TCA TGA CTT CAA CTA ACT; Nf1_I55_f CTC TCG CTC TTC ATT TCA TCT TCT; Nf1_E56_r GCC ATA AGC CAT TAA AAC CAA AAC . Met and Jun targets were the same as above . 15 ng of DNA template were amplified in the presence of 0 . 5 µM primers and components of the SensiMixPlus SYBR universal mix ( Quantace , London , UK ) using the Stratagene Mx3005P cycler ( Agilent ) . Cycling conditions: 10 min 95°C , 40× [40 sec 95°C , 40 sec 60°C , 1 min 72°C] , followed by a dissociation segment for melting curve analysis . Chromosome 8 and 15 gains were also assessed by qRT-PCR choosing Junb ( chr 8 ) and Myc ( chr 15 ) as targets , respectively . Gene dosage was normalized to an arbitrary gene on chr 12 ( Prima1 ) , whose copy number appeared widely stable in the CGH screen . Primer sequences were as follows: Junb_f GCA GCT ACT TTT CGG GTC AG; Junb_r GTG GTT CAT CTT GTG CAG GTC; Myc_f CCA CCT CCA GCC TGT ACC T; Myc_r GTG TCT CCT CAT GCA GCA CTA; Prima1_f GTT TCC ATA TCT GCA GGT GAC A; Prima1_r CTC TCG TTC ATC AGC TGT TCC T . Reactions were carried out as above but with 30 sec extension steps . Fluorescence data were analyzed using the MxPro 4 . 1 software ( Stratagene ) . After verification of primer performance , relative quantification was obtained using the threshold cycle method; ΔCt values were calibrated to wild-type ( C57BL/10 ) tail-tip DNA . Plots of ΔΔCt ( sarcomas ) values were done in R and graphical representations of ΔCt values from skeletal muscle DNA samples were made in Microsoft Excel 2007 . To study whether mixed sarcomas from MD-mice express select human sarcoma-related genes , we subjected RNA isolated from primary tumor samples as well as from tumor cell cultures to quantitative RT-PCR . Total RNA ( 1 µg ) was reverse-transcribed by standard oligo-dT primed cDNA synthesis using M-MuLV Reverse Transcriptase in a reaction buffer containing 50 mM Tris-HCl ( pH 8 . 3 at 25°C ) , 50 mM KCl , 4 mM MgCl2 , 10 mM DTT , and 1 mM dNTPs ( Fermentas ) . An aliquot corresponding to 10 ng of the initial RNA sample was subjected to a quantitative endpoint PCR , consisting of 0 . 4 µM each primer , 0 . 2 mM dNTPs , 2 mM MgCl2 , ( NH4 ) 2SO4-containing amplification buffer , and 0 . 25 units DreamTaq Green DNA Polymerase ( reagents from Fermentas ) in a 25 µl reaction volume . Primer sequences for the human rhabdomyosarcoma-marker genes ( Myog , Myl4 , Igf2 , Prox1 ) , a fibrosarcoma gene ( Vcan ) , and liposarcoma-related genes ( Pparg , Myo1e , Hoxa5 , Plau ) are available from the authors on request . After cycling ( 3 min 95°C , 35× [20 sec 95°C , 20 sec 60°C , 40 sec 72°C] , 3 min 72°C ) , 10 µl aliquots of reaction products were analyzed on ethidium bromide-stained 1 . 5% agarose gels and quantified from captured images using Image J . Relative RNA abundance was calculated by normalization to the Gapdh transcript levels and compared to skeletal muscle samples isolated from wild-type and mdx mice . RNA abundance in tumor cell lines was compared to murine C2C12 myoblast cells . Visualization of gene expression was accomplished by heatmaps made in R using the heatmap . 2 function . For I-FISH experiments on myoblasts , cells were fixed using 4% formaldehyde . FISH analysis on interphase nuclei extracted from cryofixed tissues was performed according to a previously published protocol with modifications [56] . In brief , thirty 20 µm-cryosections were fixed in PBS-buffered 4% paraformaldehyde ( 2–3 h at ambient temperature ) , rinsed twice with 0 . 9% NaCl and stored at 4°C overnight until further use . Fixed tissue sections were then transferred into a 90 µm nylon mesh and subjected to proteinase K digestion ( 0 . 05%; 10–15 min 37°C ) . After harvesting by cytospinning through the mesh , nuclei were air-dried , fixed with paraformaldehyde solution ( 4% in PBS ) , washed with 1× PBS ( 2×3 min ) , pre-treated with sodium thiocyanate ( 1 M , 80°C 1 min ) , and subjected to digestion with proteinase K ( 1 min at 37°C ) . After fixation , slides were air-dried , followed by heating to 78°C ( 8 min ) for denaturing . Slides were then incubated with digoxigenin or biotin labeled chromosome probes ( 2p , 18cen , 19q from Dr . M . Rocchi , Molecular Cytogenetic Resource Centre , Bari , Italy; chr1 from Dr . Howard J . Cooke [57]; 8cen purchased from Kreatech Diagnostics , Amsterdam , The Netherlands; chr13 FKHR and 19p/19q from Vysis , Abbott Laboratories , IL ) for hybridization overnight at 37°C . Slides were washed in 2× SSC 50% formamide , and 2× SSC at 42°C , and incubated with Cy3-labelled anti-biotin ( Dianova , Hamburg , Germany ) or FITC-labeled anti-digoxigenin antibodies in 2% BSA for 30 min at 37°C in a humid chamber . After washing in 4× SSC 0 . 1% Tween-20 ( 2×7 min at 42°C ) , slides were incubated with secondary antibodies labeled with Cy3 or FITC ( Dianova ) in 2% BSA for 30 min at 37°C , washed again as above , ethanol-dried , and mounted using Vectashield with DAPI ( Vector Laboratories , Burlingame , CA ) . Slides were analyzed using an Axioplan2 ( Zeiss ) microscope and I-FISH signals were captured using the ISIS software and quantification of the I-FISH spots was achieved with the Metafer software ( both from , MetaSystems , Altlussheim , Germany ) . For each sample 300 nuclei were automatically detected by the software and subsequently visually inspected by two independent investigators . Data presented were calculated from an average of 200 nuclei eligible for analysis . | All kinds of muscular dystrophies ( MDs ) are characterized by progressive muscle wasting due to life-long proliferation of precursor cells of myo- ( muscle ) , fibro- ( connective tissue ) , and lipogenic ( fat ) origin . Despite discovery of many MD genes over the past 25 years , MDs still represent debilitating , incurable diseases , which frequently lead to premature death . Thus , it is imperative to gain novel insights into the underlying MD pathomechanisms . Here , we show that different mouse models for the most common human MDs frequently develop skeletal musculature-associated tumors , presenting as complex sarcomas , consisting of myo- , lipo- , and fibrogenic compartments . Collectively , these tumors are characterized by profound genomic instability such as DNA damage , recurring mutations in cancer genes , and aberrant chromosome copy numbers . We also demonstrate the presence of these cancer-related aberrations in dystrophic muscles from MD mice prior to formation of visible sarcomas . Moreover , we discovered corresponding genomic lesions also in skeletal muscles from human MD patients , as well as stem cells cultured thereof , and show that genomic instability precedes muscle degeneration in MDs . We thus propose that cancer-like genomic instability represents a novel , unifying pathomechanism underlying the entire group of genetically distinct MDs , which will hopefully open new therapeutic avenues . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"animal",
"genetics",
"cancer",
"genetics",
"genetic",
"mutation",
"duchenne",
"muscular",
"dystrophy",
"cytogenetic",
"analysis",
"animal",
"models",
"genetics",
"of",
"disease",
"model",
"organisms",
"biology",
"mouse",
"x-linked",
"congenital",
"hereditary",
"myopath... | 2011 | DNA Damage, Somatic Aneuploidy, and Malignant Sarcoma Susceptibility in Muscular Dystrophies |
Insulin-like growth factor I receptor ( Igf1r ) signaling controls proliferation , differentiation , growth , and cell survival in many tissues; and its deregulated activity is involved in tumorigenesis . Although important during fetal growth and postnatal life , a function for the Igf pathway during preimplantation development has not been described . We show that abrogating Igf1r signaling with specific inhibitors blocks trophectoderm formation and compromises embryo survival during murine blastocyst formation . In normal embryos total Igf1r is present throughout the membrane , whereas the activated form is found exclusively at cell contact sites , colocalizing with E-cadherin . Using genetic domain switching , we show a requirement for E-cadherin to maintain proper activation of Igf1r . Embryos expressing exclusively a cadherin chimera with N-cadherin extracellular and E-cadherin intracellular domains ( NcEc ) fail to form a trophectoderm and cells die by apoptosis . In contrast , homozygous mutant embryos expressing a reverse-structured chimera ( EcNc ) show trophectoderm survival and blastocoel cavitation , indicating a crucial and non-substitutable role of the E-cadherin ectodomain for these processes . Strikingly , blastocyst formation can be rescued in homozygous NcEc embryos by restoring Igf1r signaling , which enhances cell survival . Hence , perturbation of E-cadherin extracellular integrity , independent of its cell-adhesion function , blocked Igf1r signaling and induced cell death in the trophectoderm . Our results reveal an important and yet undiscovered function of Igf1r during preimplantation development mediated by a unique physical interaction between Igf1r and E-cadherin indispensable for proper receptor activation and anti-apoptotic signaling . We provide novel insights into how ligand-dependent Igf1r activity is additionally gated to sense developmental potential in utero and into a bifunctional role of adhesion molecules in contact formation and signaling .
The ultimate goal of the mammalian preimplantation development is the formation of a hollow shaped embryo called blastocyst , crucial for all stages of subsequent development . It is generated by a highly organized interplay of multiple signaling pathways guiding development from a fertilized egg to an 128-cell staged embryo . At the end of this important process three distinct cell lineages are established – the epiblast , that will give rise to the embryo proper , the primitive endoderm , that forms some of the extraembryonic membranes and the trophectoderm ( TE ) that contributes to the placenta [1] , [2] . In mice , at embryonic day ( E ) 4 . 5 after segregation of the early lineages is completed the blastocyst hatches from its glycoprotein envelop ( zona pellucida ) in order to invade the uterine epithelium and implant . Besides the orchestrated interplay of transcription factor networks that regulate expression of Oct4 , Cdx2 , Nanog , Gata4 and Gata6 , indispensable for correct lineage segregation [3]–[7] , the formation of a proper blastocyst strongly depends on tightly controlled cell adhesion mainly mediated by E-cadherin ( E-cad , also known as Cdh1 ) [8] . Mice deficient for E-cad ( Cdh1 ) are incapable of forming a proper trophectodermal epithelium [9] , [10] . Compaction , however , is accomplished by residual maternally provided gene expression and is lost upon maternal/zygotic E-cad ( Cdh1 ) depletion [11] , [12] . In contrast , N-cadherin ( N-cad , also known as Cdh2 ) , another crucial member of classical cadherins is first detected after implantation and its gene ablation demonstrates distinct functions as well . N-cad ( Cdh2 ) expression is initiated when the first mesoderm cells start to emerge at the primitive streak during gastrulation [8] . Although the mesoderm is properly formed in N-cad ( Cdh2 ) -deficient mice , patterning of the somites and the neural tube is severely affected , and embryos die due to a heart defect [13] . Interestingly , the cardiac phenotype is rescued by ectopic expression of E-cad ( Cdh1 ) in the developing heart , indicating that the adhesive function of cadherins is at least in part interchangeable [14] . In agreement with this finding , E-cad and N-cad have similar properties in the degree of conservation of amino acids ( aa ) , in mediating homophilic adhesion , and in binding to the same intracellular interaction partners , such as β-catenin , Plakoglobin and p120ctn [15] , [16] . However , E-cad ( Cdh1 ) and N-cad ( Cdh2 ) are usually expressed in a mutually exclusive pattern and induce different cellular properties . Cell polarity and an epithelial sessile shape are established in cells that express E-cad ( Cdh1 ) , whereas cell migration can be induced in cells that gain N-cad ( Cdh2 ) expression [17] , [18] . This phenomenon reflects the contribution of cadherins to epithelial-mesenchymal transition during gastrulation , as well as during carcinogenesis [8] , [19] , [20] . Detailed knowledge about how the unique properties of the two related cadherins are translated in molecular terms is still limited . By ectopically switching cadherin expression from E-cad to N-cad using a previously reported gene replacement approach , we were able to maintain cell adhesion and analyze specific and unique molecular features of either E-cad or N-cad [21] . Interestingly , embryos carrying two N-cad knock-in alleles in the E-cad ( Cdh1 ) locus ( N-cad ki/ki ) were not able to form a proper TE and died within the zona pellucida , similar to E-cad−/− embryos . Since control embryos that harbor an HA-tagged E-cad ( Cdh1 ) cDNA knock-in allele were able to form blastocysts and implant properly , the result of N-cad ki/ki embryos suggests that E-cad has a unique function during blastocyst formation [21] , [22] . However , how this crucial and unique function of E-cad is implemented and why it cannot be replaced by N-cad is a complete enigma . The insulin-like growth factor I receptor ( Igf1r ) belongs to the protein family of receptor tyrosine kinases and is mainly activated by Igf1 and Igf2 acting in an autocrine and paracrine manner . The downstream signaling cascade regulates proliferation , differentiation , metabolism and survival of most cell types during fetal growth and postnatal life [23] , [24] . Blocking kinase activity by either loss-of-function mutation of the receptor or of both ligands result in reduced body weight and size combined with multiple defects including muscle dystrophy and impaired survival of newborn pups [25] , [26] . The Igf1/Igf2/Igf1r axis provides growth promoting , anti-apoptotic functions in almost all tissues and organs and treatment of mouse preimplantation embryos with Igf1 enhanced blastocyst formation in vitro by supporting PI3K/Akt activity [27] , [28] . However , detailed knowledge about a role of this pathway during preimplantation development is lacking . Here , we further addressed the question about the unique function of E-cad by replacing its expression with chimeric cadherin genes using similar knock-in approaches as for N-cad ki/+ mice and identified a novel fundamental and cell-adhesion independent function of E-cad in promoting cell survival of the TE by facilitating Igf1r activity .
To elucidate the function of E-cad during TE formation , the protein was divided into two parts , its N-terminal extracellular adhesive region and its C-terminal transmembrane and intracellular portion , the latter of which mediates its interaction with catenins . These regions were combined with the matching portions of the N-cad molecule to generate artificial chimeric cadherins ( Figure 1A ) . Cloned cDNAs encoding EcNc ( corresponding to aa 1–710 of the E-cad precursor peptide and 725–906 of N-cad ) and NcEc ( aa 1–724 of N-cad and 711–884 of E-cad ) were fused to a sequence encoding an HA-tag and inserted into the E-cad ( Cdh1 ) locus to replace E-cad as described previously ( Figure 1B–1D ) [21] , [22] , [29] . For both EcNc and NcEc approaches , two independent ES-cell clones were used to generate the corresponding knock-in strains . Proper expression of the chimeric molecules was confirmed on mRNA level in ES cells and by immunofluorescence and immunohistochemistry of embryos after deletion of the selection cassette . RNA levels of the two knock-in alleles were comparable to the amount of N-cad ki and EcHA transcripts ( Figure S1A ) . Distribution of both chimeric proteins completely overlapped with endogenous E-cad staining in TE and inner cell mass ( ICM ) cells of heterozygous preimplantation embryos ( Figure 1E ) and accurately recapitulated E-cad ( Cdh1 ) expression in the epithelia of post-implantation stages ( Figure S1B ) . The analysis confirmed successful gene replacement and correct spatiotemporal expression of both knock-in alleles . Similar to the N-cad ki/+ mice , no phenotype was detected in heterozygous NcEc or EcNc animals , indicating that the chimeric proteins did not interfere with E-cad-mediated adhesion . At E2 . 5 , EcNc and NcEc homozygous embryos were observed in a 24-h time-lapse experiment to monitor blastocyst formation . Both homozygous mutants underwent compaction normally and were indistinguishable from their heterozygous littermates ( Figure 2A and 2C , 0–6 h ) . EcNc homozygous mutants ( EcNc ki/ki ) , which express the cadherin with the adhesive domain of E-cad , properly segregated the TE from the ICM cells and formed a blastocoel cavity similar to control embryos ( Figure 2A , 6–24 h , and 2E ) . However , the TE was more fragile as pulsing caused by sequential expansion and collapse was observed more frequently in EcNc homozygous mutants than in control littermates ( Figure 2A and Video S1 ) . Proper protein localization to the basolateral membranes of TE cells was confirmed by immunofluorescence ( Figure 2B ) . In contrast , homozygous NcEc mutants ( NcEc ki/ki ) , which express the cadherin with the extracellular domain of N-cad , were incapable of establishing a blastocyst . Cells on the outside were shuffled around , rounded up , vacuolated and became scattered at the surface of the embryo , whereas the control littermates formed blastocysts during the 24-h time-lapse recording ( Figure 2C , 2F and Video S2 ) . Confocal optical plane sections of immunofluorescently labeled embryos revealed that the NcEc protein was evenly distributed on the scattered cells on the surface ( Figure 2D , yellow arrow ) . The results indicated that TE and blastocoel cavity formation required the presence of the extracellular domain of E-cad . However , the reduced stability of the TE layer in homozygous EcNc mutants showed that the intracellular domain of the E-cad molecule also contributed to the process and has an important function that cannot be performed by N-cad . To confirm that lineage segregation , cell polarity and expression of molecules playing a key role during the cavitation process were not affected in NcEc homozygous embryos , we analyzed the expression of specific marker genes by immunofluorescence labeling . No difference in the expression or localization of essential proteins was found in NcEc embryos , indicating that the ICM and the TE were correctly specified and that the cavitation machinery was present ( Figure 3A and 3B ) . Cell polarity was correctly established based on detection of apical staining of ZO-1 and Ezrin and basolateral localization of the chimeric cadherins and Na+/K+-ATPase ( Figure 3C ) . In contrast to E-cad−/− , both EcNc ki/ki and NcEc ki/ki embryos show proper expression and membrane localization of β-catenin , Plakoglobin and p120ctn ( Figure S2A ) . They were connected to both chimeric cadherin molecules in a similar manner as detected by immunoprecipitation ( Figure S2B ) . In addition , embryo-derived homozygous TE cells differentiated into trophoblast giant cells , and ES cells showed proper adhesive colony formation . Similar differentiation capacities were observed for EcNc ki/ki and NcEc ki/ki genotypes , in stark contrast to Ecad−/− ES cells ( Figure S3 ) . This indicated that cell polarity , adhesion , cadherin complex composition and the cavitation machinery are well established in both homozygous mutants . One major difference between the NcEc ki/ki embryos and the EcNc ki/ki embryos was that the failure of proper TE formation in NcEc mutants was accompanied by cell scattering and vacuolation in the outside cells , both of which indicate the induction of programmed cell death ( PCD ) . To verify an aberrant induction of PCD in NcEc mutants , embryos were labeled for cleaved Caspase 3 , a general marker for the activation of apoptosis . A substantial increase in number of Caspase 3-positive cells was detected in the TROMA-1 labeled outer cells of the mutants ( Figure 4A and 4B ) . In contrast , no apoptosis was found in the TE cells of homozygous EcNc embryos or control littermates . Moreover , EcNc ki/ki embryos did not show a delayed onset of apoptosis as identified in prolonged embryo cultures for additional 24 h , indicating that in these mutants TE was not prone to PCD ( Figure S4B and S4C ) . Interestingly , the induction of PCD in homozygous NcEc mutants was phenocopied if wildtype ( wt ) embryos were incubated with staurosporine , a bacterial-derived alcaloid which activates PCD by inducing Caspase 3 . Treating wt embryos with 50 nM staurosporine severely compromised blastocyst formation ( Figure S5A , S5B and S5D ) . This result strongly indicated that in NcEc mutants , the fragile equilibrium between cell survival and cell death was shifted towards apoptosis due to the misexpression of the NcEc chimeric cadherin in the E-cad ( Cdh1 ) expression domain . Moreover , since increased PCD was not detected in EcNc mutants , we concluded that replacing the extracellular domain of E-cad with N-cad specifically caused this imbalance . In the following experiments , we sought to re-establish this fine-tuned equilibrium to promote cell survival . Proper preimplantation development in mice and humans relies on the orchestrated program of various growth factors and small secreted molecules , such as prostaglandins , which are produced by the embryo and the oviduct , [30] , [31] . In vitro , activation of prostacyclin-dependent signaling enhances embryo survival and hatching of mouse , human and pig embryos by suppressing Caspase 3 activation via PPARδ and 14-3-3ε and by blocking cytochrome C release from the mitochondria [31] . Here , we used iloprost , a stable synthetic analogue of prostacyclin ( PGI2 ) and analyzed the effect on blastocyst formation in NcEc homozygous mutants [32] , [33] . Strikingly , if homozygous NcEc mutants were cultured between E2 . 5 and E3 . 5 in the presence of 1 µM iloprost , an accurate blastocoel cavity formed within the 24-h time-lapse recording ( Figure 4C , 4D , 4H; Figure S5C; Video S3 ) . Hence , blocking Caspase 3-mediated activation of PCD rescued this phenotype . We re-investigated N-cad ki/ki embryos that also failed to form a TE under standard conditions [21] . Interestingly , treatment of those embryos rescued blastocyst formation in a similar fashion , whereas there was only a moderate effect on homozygous EcNc mutants , resulting in enhanced stability of the TE ( Figure 4E , 4F and Video S4 ) . In contrast , blastocyst formation was not rescued in in vitro cultured E-cad-null embryos , presumably due to the entire absence of cadherin-mediated adhesion , which is indispensable for TE formation ( Figure 4G , 4H and Video S5 ) . The proper spatial organization and the epithelial nature of the TE in iloprost-stimulated mutants were confirmed by TROMA-1 staining . An intact TE layer of TROMA-1-positive cells that was correctly separated from TROMA-1-negative ICM cells was detected in homozygous NcEc and N-cad ki/ki mutants ( Figure 4I ) . Manipulation of other branches of the PCD pathway resulted similarly in rescue of NcEc embryos . Either blocking p53 by cyclic pifithrin alpha ( cPFT ) or directly inhibiting Caspase 3 by Z-DEVD-FMK enabled both TE formation and the maintenance of blastocyst integrity ( Figure S4A ) . These results indicated that prosurvival cues need to be active during blastocyst formation and that in homozygous NcEc mutants the shifted balance of cell survival and PCD was artificially returned to its equilibrium by inhibiting the apoptotic program at different levels . However , this rescue was only possible in a cadherin-mediated manner since blocking apoptosis rescued only cadherin-expressing embryos . Since all heterozygous mutant mice analyzed here did not show defects and developed normally until adulthood , a dominant effect of an inappropriate cadherin in TE cells is very unlikely . This was confirmed by analysis of compound NcEc/EcNc mice that show proper blastocyst formation capacity ( Figure S4E ) . When searching for a putative prosurvival signal that is triggered by the presence of the extracellular domain of E-cad , we focused on receptor tyrosine kinase ( RTK ) signaling cascades . In specific cellular contexts , cadherins interact with RTKs like Egfr and Fgfr2 and thus modulate downstream signaling activities [34]–[36] . Since incubation with bFGF did not improve blastocyst formation ( data not shown ) , and mutations in Egfr do not reveal TE defects [37] , we focused on Igf1r-mediated signaling . In previous studies , Igf1 enhanced blastocyst formation by providing a survival signal through PI3K/Akt [27] , [28] . In agreement with these data , blocking Igf1r signaling in wt embryos at the morula stage with a specific inhibitor ( Tyrphostin AG1024 ) induced cell fragmentation of outer cells and blocked TE formation ( Figure S5E ) . When NcEc homozygous mutant embryos were treated with 100 ng/ml Igf1 for 24 h and recorded with time-lapse microscopy , these embryos formed a stable TE and even initiated hatching at the end of the recording ( Figure 5A , 5F; Figure S4A , S4B; Video S6 ) . In agreement with our previous results , Igf1-mediated rescue was observed in homozygous NcEc and N-cad ki/ki mutants but not in E-cad−/− embryos ( Figure 5C–5F and Videos S8 and S9 ) . To rule out the possibility that Igf1 simply delayed the induction of PCD we generated prolonged embryonic cultures of NcEc ki/ki . After initial 24 h incubation , embryos were transferred to fresh medium and kept for additional 24 h in the incubator . In the presence of Igf1 NcEc ki/ki embryos formed a stable TE without showing indications of apoptosis ( Figure S4B ) . Thus , the treatment of NcEc ki/ki embryos with either Igf1 , iloprost or cPFT rescued apoptosis as indicated by absence of active Caspase 3 staining ( Figure S4D ) . In utero as well as in vitro , preimplantation embryos receive insulin ( Ins1 ) - and insulin receptor ( Insr ) -mediated signals [23] . Thus , homozygous mutants were incubated with 25 µg/ml insulin to determine whether this pathway contributes to cell survival . Although there was a significant improvement in the formation of a blastocoel cavity , the effect was modest in comparison to Igf1 treatment ( Figure 5B and Video S7 ) . This result revealed that Igf1 and the activation of its receptor Igf1r plays a crucial role during preimplantation development and the receptor kinase activity provides the endogenous survival signal in wt embryos that is blocked or attenuated in NcEc and N-cad ki/ki mutants . Our previous analysis indicated a functional link between E-cad and Igf1r . In vitro , a direct interaction between these two proteins was observed in MCF-7 cells [38] , [39] , but whether this interaction also influences Igf1r activity in a ligand-dependent or ligand-independent manner is unknown . To further study the putative role of Igf1r in preimplantation development and whether Igf1r kinase activity is facilitated by E-cad to regulate survival of TE cells , we first analyzed the expression of Igf1r and the amount of its activated form ( pIgf1r ) . In wt embryos , the receptor was detected in preimplantation stages . It localized to basolateral membranes and additionally to the apical membrane of TE cells , showing a partial overlap with E-cad at cell contact sites ( Figure 5H and Figure S5F ) . Interestingly , analysis of pIgf1r with a phospho-specific antibody revealed that the receptor was only activated at cell contact sites that showed substantial overlap with anti-E-cad staining at lateral membranes ( Figure 5H and Figure S5F ) . Treatment of wt embryos with Igf1 hyperactivated Igf1r resulting in ectopic pIgf1r detection at apical sites , whereas blocking of receptor activation by Tyrphostin AG1024 abolished pIgf1r detection ( Figure 5G and Figure S5F ) . Strikingly , and in contrast to wt or EcNc embryos , NcEc and E-cad−/− embryos showed weaker or absent pIgf1r staining intensities , although the overall amount of Igf1r was not changed ( Figure 5I ) . Moreover , treatment of wt embryos with a chelating agent , like EGTA to deplete Ca2+-ions and to interfere with cadherin conformation and function [40] led to a decrease in pIgf1r levels mimicking the lack of Igf1r activation of homozygous NcEc mutant embryos ( Figure S5G ) . These results suggested that the activity of the receptor is reduced in NcEc mutants due to lack of facilitation by interaction of E-cad and Igf1r . To test this idea , we performed a Duolink proximity ligation assay , a fluorescence-based method to show protein-protein interaction in situ . As a control , we analyzed the known interaction between E-cad and β-catenin in wt blastocysts , which showed fluorescent signals at sites of interaction at basolateral membranes as expected ( Figure 6A , 6 ) . Analysis of the interaction of E-cad and Igf1r by Duolink revealed fluorescent labeling in wt blastocysts . In agreement with the cellular distribution of pIgf1r , interaction was detected at lateral membranes ( Figure 6A , 1 ) . A similar assay in wt embryos using an antibody against pIgf1r gave a comparable result , suggesting that E-cad interacts with the activated form of the receptor or that only E-cad-bound receptor becomes activated ( Figure 6A , 4 ) . However , In NcEc ki/ki embryos only a reduced and almost absent signal was present , although anti-pIgf1r and anti-E-cad antibodies were able to detected both proteins . The analysis demonstrated a lack of interaction in NcEc ki/ki embryos in agreement with our hypothesis ( Figure 6A , 3 ) . No signal was obtained in E-cad-null embryos ( Figure 6A , 2 , 5 ) . In a second approach complexes between endogenously expressed E-cad and Igf1r were analyzed by anti-E-cad immunoprecipitation ( IP ) . Wt and NcEc ki/ki trophoblast stem cells ( TS cells ) were isolated from blastocyst outgrowths and from ES cells after transient induction of ectopic Cdx2 expression , respectively . Binding of E-cad to Igf1r was identified upon co-immunoprecipitation of lysates from wt TS cell lysates ( Figure 6B , upper panel ) . Two additional fragments of lower molecular weight were specifically co-precipitated and detected by two individual anti-Igf1r antibodies ( Figure 6B and data not shown ) . The additional bands were very faint in the input samples . They presumably represent γ-secretase processed forms of the receptor as observed previously [41] generated upon activation and are enriched in the immunoprecipitation ( Figure 6B , upper panel ) . In contrast to that , no interaction was seen when chimeric NcEc was precipitated with the same anti-E-cad antibody from lysates of NcEc ki/ki TS cells . Specific Igf1r signals were not detectable in IgG and E-cad IPs . Our data suggest that E-cad and Igf1r interact in the TE at sites of cell-cell contact . This interaction is indispensable for TE formation via facilitation of RTK signaling activity , which in turn promotes cell survival and keeps apoptosis at bay .
Cadherins are bona fide adhesion molecules that are involved in clustering cells of the same type together . Additionally , a role in cadherin-mediated RTK signal transduction through their interactions with different receptors has been suggested . These interactions either attenuate or enhance RTK activation in a ligand-dependent or ligand-independent manner [34]–[36] , [42] , [43] . The Igf/insulin-like growth factor I receptor axis controls growth , differentiation and cell survival and comprises Igf1 , Igf2 and insulin as ligands and Igf1r , Igf2r and insulin receptor ( Insr ) [23] . The activity of this pathway is further regulated by the Igf1-binding proteins Igfbp3 , Igfbp4 and Igfbp5 [44] , [45] . In addition , Igf1 binds to Insr and insulin to Igf1r with lower affinity , and Igf2 signaling is transduced through both Igf1r and Insr simultaneously [25] . In contrast , the major role of Igf2r is to attenuate Igf1 and Igf2 signals since it lacks a kinase domain [46] . During preimplantation development , Igf1r , Igf2r and Insr are expressed , and Igf1 and insulin are provided both maternally and zygotically [47] . Treatment of embryos in vitro with Igf1 enhances embryo viability via mitogenic and anti-apoptotic responses , indicating that Igf1 has a role in providing survival signals [27] , [28] , [30] . In this study , we unraveled a link between E-cad and Igf1r that promotes cell survival in the TE . Our data indicate a previously unknown function of Igf1r during preimplantation development . Furthermore , full activation of the Igf1r kinase domain likely requires physical interaction to the extracellular domain of E-cad . Abrogating cadherin function by conformational changes upon Ca2+-withdrawal [40] , [48] results in reduced phosphorylation of Igf1r . If E-cad is replaced by either N-cad or a chimera harboring the extracellular domain of N-cad , no interaction of the two proteins is detected , and Igf1r is only inefficiently activated . Consequently , the balance between cell survival and cell death is shifted towards PCD , and embryos cannot form a functional TE . Homozygous NcEc and N-cad ki/ki embryos are rescued by an excess of Igf1 ligand because the activity of the receptor is artificially raised to normal levels . However , a full rescue of blastocyst formation is possible only if cadherin-mediated cell adhesion is also present ( Figure 7 ) . We hypothesize that E-cad , in addition to its role in mediating homophilic cell adhesion , has a novel function during preimplantation development and triggers the survival signal initiated by Igf1/Igf1r activation . Mice in which components of the Igf axis are knocked out have been described by several groups , and deficiencies in this axis lead to reductions in body size and weight [44] . Igf1 mutants are severely affected and display muscle dystrophies , with only 5% of offspring reaching adulthood , and these mice are infertile [26] . Although single , double and even triple knockouts of components of the Igf axis do not show preimplantation defects [25] , [46] , certain combinations of mutations always result in infertility . These mice may not show preimplantation defects , due to residual maternal activity: that is , the mice still receive Igf/insulin signals from maternal tissues , and they are also provided with maternal mRNA and protein for the receptors of the Igf axis during oocyte maturation . A combined maternal/zygotic loss-of function experiment has not been performed yet to show whether the complete depletion of ligands and receptors results in TE formation defects . By treatment with Tyrphostin AG1024 we here targeted Igf1r and Insr of both maternal and zygotic origin . This simultaneous blocking of entire downstream signaling results in the induction of apoptosis . In addition to previous observations , our data support the importance of Igf1r signaling already during preimplantation development , since also E-cad dependent loss of Igf1r activation results in inefficient maintenance of survival signals . Analyses of loss- and gain-of-function mutations are indicating a general underlying mechanism that controls cell survival . Cells isolated from Igf2-null or Igfr1-null animals display increased apoptosis as indicated by the small body size [25] , [49] . In contrast , overexpression of Igf1 or Igf2 led to a decrease in apoptosis , observed as non-involuting mammary glands and pancreas hyperplasia , indicating a dramatic shift towards cell survival [50] , [51] . Combined with our new findings Igf1 has a general role in controlling cell survival and cell death , a function that is active during preimplantation development as well . Cadherin-mediated modulation of RTKs has been previously shown for Egfr and Fgfr2 [34] , [35] , [42] , and it may be an intrinsic property of RTKs that they need to be clustered by or interact with cadherins to be efficiently activated . Interestingly , soluble E-cad isolated from the serum of cancer patients blocks apoptosis via activation of Egfr in MDCK cells [52] . This suggests a comparable role of the E-cad/Egfr interaction to that found in our analysis for Igf1r . It is tempting to speculate that during preimplantation development control of PCD by Igf1r is regulated in an E-cad-dependent manner and has an important function . The coupling of both proteins may act as a sensor to eliminate embryos that are unable to manage the crucial step during the morula to blastocyst transition , in which the embryo switches from depending on pyruvate to glucose [53] . This switch is correlated with increasing demands for ATP within the embryo , to support Na+/K+-ATPase activity , and is mediated by Igf1r [54] . In line of our hypothesis , the fragile balance between survival and apoptosis is linked to an interaction of Igf1r and E-cad , which acts as a checkpoint to assess the viability of the embryos . Saving nutrients and energy by eliminating abnormal embryos at an early stage is a favored strategy . Our analysis suggests a hitherto unknown preimplantation checkpoint that couples integrity of the TE to embryo survival . Additionally , there is evidence that connects E-cad and the regulation of the PCD-cell survival balance in other tissues . Hyperactivated Igf1r in the mammary gland shifted the balance towards cell survival [50] , whereas an opposite effect that promotes apoptosis was observed after E-cad ( Cdh1 ) depletion . During lactation milk production is hampered due to precocious involution , resulting in a shift towards PCD [55] . It will be interesting to address whether loss of E-cad also impairs proper Igf1r function in the mammary gland by a similar mechanism as described here . Although our data are in favor of this model , we cannot fully exclude a different mechanism and contribution of secondary effects . Many RTKs utilize co-receptors like integrins or adhesion molecules , such as CD44 for proper function [56] . In our mutants adhesion and cell polarity may be altered resulting in improper localization of Igf1r . However , staining for total Igf1r and cell polarity markers of NcEc ki/ki mutants indicated that Igf1r expression and protein localization were not changed and the TE cells still maintain apical-basal polarity . Nevertheless , secondary effects may occur due to reduced cell adhesion based on the artificial nature of the chimeric cadherins and/or to altered connection to the cytoskeleton , essential for proper adhesion . This may differ between the two molecules EcNc and NcEc and may have escaped from our analysis . Interestingly , our data suggest that in addition to the extracellular domain the intercellular domain of E-cad contributes to TE formation and normal development as well since the EcNc homozygous mutants are incapable of hatching . Unique molecular features may reside in differential affinities to bind to β-catenin or other interacting proteins , as has been suggested previously [57] . E-cad and N-cad differ in their interactions with p120ctn isoforms , which may influence p120ctn-mediated small GTPase activity and the flexibility of adherens junctions [58] . Very likely , unique intracellular interaction partners exist for both cadherins but need to be identified in future experiments . Our study has unraveled a novel role of Igf1r activity and a crucial mechanism that provides a link of how E-cad may control the balance between cell survival and PCD . Igf1r activation is essential to promote cell survival in the TE lineage , but requires E-cad-mediated facilitation of the signal via protein interaction . Unraveling the role of this function and its implication for morphogenesis and differentiation will have a significant impact on our understanding of cadherin-mediated signaling during embryogenesis and in human diseases , such as cancer .
Animal husbandry and all experiments were performed according to the German Animal Welfare guidelines and approved by the local authorities . cDNAs of E-cad and N-cad were used to generate coding sequences for chimeric proteins EcNc and NcEc corresponding to the extracellular domain of E-cad ( amino acids 1–710 ) fused to intracellular domain including transmembrane portion of N-cad ( aa 725–906 ) and aa 1–724 of N-cad fused to aa 711–884 of E-cad , respectively . Both sequences were combined with a C-terminal HA-tag and inserted into the ATG codon of the previously described targeting vector ( pBluescriptII , Stratagene ) using standard molecular cloning techniques [21] , [22] . Homologous recombination and analysis of surviving ES cells by Southern blot was done as described [21] , [22] . Two independent clones were used for injection into blastocysts to generate chimeric mice . After backcrossing to Zp3-cre mice to delete the neomycin resistance cassette during oocyte maturation EcNc and NcEc heterozygous mouse lines were established [59] , backcrossed to C57BL/6 and inter se to obtain homozygous mutant embryos . Genotyping was performed by PCR using tail biopsies , yolk sacs or entire embryos with the following primers: wt allele ( Ecad5′UTR_s , CCC AAG AAC TTC TGC TAG AC/Ecad1_as , TAC GTC CGC GCT ACT TCA ) , EcNc and NcEc alleles ( ENcad3′_s , AAG CTG GCG GAC ATG TAC/Ecad1_as ) , EcNc ( Ecad_s , ATC GCC ACA CTC AAA GTG/Ncad1_as , CTG TGG CTC AGC ATG GAT ) , NcEc ( Ncad_s , TGG AAG CTG GTA TCT ATG/Ecad2_as , TCA TCA GGA TTG GCA GGA ) , Ncad ki ( Ecad5′UTR_s/Ncad2_as , TGG CAA GTT GTC TAG GGA ) . Preimplantation embryos were isolated by flushing the oviducts or uteri with M2 medium , transferred into 10 µl KSOM droplets covered with mineral oil ( Fluka ) . Time-lapse microscopy was performed as described with minor modifications using a Zeiss Axiovert 200 M microscope equipped with Narishige manipulators , Incubator XL and Tempcontrol together with a humidifier connected to a heating stage E100 ( Zeiss ) at 37°C and 7 . 5% CO2 [21] , [22] . Embryos were photographed every 15 min for 24 h . Zeiss AxioVision ver . 4 . 8 software and Uniblitz shutters were used for the acquisition of time-lapse images . Embryos were treated with specific inhibitors and growth hormones as indicated in the following concentrations: 1 µM iloprost ( Cayman chemical ) , 30 µM cPFTalpha ( Sigma ) , 50 µM Z-DEVD-FMK ( Enzo ) , 1–50 nM staurosporine ( Enzo ) , 100 ng/ml Igf1 ( eBioscience ) , 25 µg/ml insulin ( Sigma ) , 10 µM Tyrphostin AG1024 ( Alexis biochemicals ) , 2 mM EGTA . Each experiment was repeated at least five times . After isolation embryos were washed in PBT ( 0 . 05% Tween/PBS ) and fixed with 2% PFA/PBS for 10 min . Cellular permeabilization was carried out for 5 min with 0 . 3% Triton X-100/PBS and embryos were incubated in primary antibody in 2 . 5% BSA/PBT for 2 h to overnight at room temperature . Subsequently , alexa488 or alexa594-conjugated secondary antibodies were applied for 1 h . Embryos were stained with DAPI to visualize nuclei ( 1∶1000 , Invitrogen ) and mounted in PBS droplets covered with mineral oil in glass bottom petri dishes ( Willco wells ) . Confocal microscopy was performed using Leica TCS SP2 laser scan head attached to a Leica DM IRE2 inverted microscope . Images were processed using IMARIS software ( Bitplane ) . Antibodies: anti-E-cadherin ( intracellular ) , anti-N-cadherin , anti-β-catenin , anti-Plakoglobin , ( BD Bioscience ) , HA . 11 ( Covance ) , anti-Ezrin , anti-cleaved Caspase 3 ( Cell Signaling ) , anti-E-cadherin ( extracellular , gp84 ) [60] , TROMA-1 [61] , anti-p120ctn , anti-ZO-1 ( Zymed ) , anti-Na+/K+-ATPase ( Millipore ) , anti-AQP3 , anti-Sox2 ( Calbiochem ) , anti-Oct4 ( Santa Cruz ) , anti-Nanog [62] , anti-Cdx2 ( Biogenex ) , anti-Igf1r ( α-subunit , abcam ) , anti-Igf1r ( β-subunit , Cell Signaling ) , anti-pIgf1r ( abcam ) . The Duolink assay ( Olink Bioscience ) was performed according to the manufacturers instructions in 10 µl droplets covered with mineral oil at 37°C . Immunoblotting and immunoprecipitation ( IP ) was performed as described for ES cell lysates ( 500 ng protein , 500 ng antibody ) or with minor modifications for TS cells [22] . Briefly , TS cells were stimulated with 50 ng/ml Igf1 for 10 min , incubated in crosslinking buffer ( 6 mM KCl , 2 mM Bissulfosuccinimidyl suberate/PBS ) for 30 min at 4°C , followed by quenching in 100 mM Glycine/PBS and harvested in lysis buffer ( 20 mM Tris-HCl pH 7 . 9 , 137 mM NaCl , 2 mM MgCl2 , 5 mM EDTA , 1 mM EGTA , 1% Triton X-100 , 10% Glycerol , 10 mM Na3VO4 , 10 mM NaF , 1× Complete protease inhibitor , Roche , 1 mM PMSF ) [43] . For IP , 2 mg of protein were incubated overnight at 4°C using 1 µg anti-E-cad ( BD ) antibody and 25 µl slurry of protein-G coupled Dynabeads ( Invitrogen ) . After washing in washing buffer ( 25 mM Tris-HCl pH 7 . 4 , 100 mM NaCl , 0 . 1% Tween-20 ) IP samples were separated by 8–10% SDS-PAGE . Proteins of the IP or from whole cell lysates were transferred to nitrocellulose membranes using a semi-dry electroblotter ( Biorad ) with 200 mA for 45 min . After blocking ( 2% dry milk in TBS/0 . 1% Tween-20 ) for 30 min , membranes were incubated with antibody solution for 2 h to overnight and subsequently with a secondary horseradish peroxidase-conjugated antibody for 1 h . Anti-Igf1r ( β-subunit ) , anti-Igf1r ( total , α-IR3 , Calbiochem ) , anti-E-cad ( BD ) and anti-Gapdh ( Calbiochem ) were used . Proteins were detected by soaking membranes in ECL Plus ( GE Healthcare ) exposed to autoradiography films . Generation of homozygous EcNc and NcEc ES cells was performed as described previously using morulae or blastocysts from heterozygous intercrosses , plated on mytomycin-treated embryonic fibroblasts [22] . Alkaline phosphatase ( AP ) staining of isolated ES cells was used to verify undifferentiated pluripotent status . Cells were fixed in 4% PFA/PBS for 15 min , washed two times with PBS and incubated for 30 min in AP staining solution ( 25 mM Tris-maleic acid pH 9 . 0 , 0 . 4 mg/ml α-naphtyl phosphate , 1 mg/ml Fast Red TR Salt , 8 mM MgCl2 , 0 . 01% Na-deoxycholate , 0 . 02% NP40 ) . Wt TS cells were generated similar to ES cells by blastocyst outgrowth in RPMI 1640 , 20% FCS , 2 mM glutamine , 1 mM pyruvate , 50 µg/ml penicillin/streptomycin , 100 µM β-mercaptoethanol , 25 ng/ml FGF4 ( Sigma ) , 1 µg/ml Heparin as described [63] . NcEc ki/ki TS cells were derived from transdifferentiated ES cells by stable transfection of an inducible Cdx2 plasmid ( Cdx2ER ) and treated with 1 µg/ml 4-OH-tamoxifen for 10 days [64] . Teratoma formation was induced by subcutaneous injection of 1×107 trypsinized ES cells into BALB/c nude mice as described [21] , [22] . Specimen were fixed at 4°C overnight in 4% PFA/PBS and dehydrated in 30% , 50% , 70% , 100% Ethanol/PBS series for 1 h each , followed by two 10 min incubations in 100% Xylene before transferring them to paraffin overnight . Casted into paraffin blocks samples were sectioned using a RM2155 microtome ( Leica ) at 7 µm and stored at 4°C until further processing . Hematoxylin/eosin ( H&E ) staining and immunohistochemistry was carried out as described previously using epitope retrieval by 20 min boiling in Tris-EDTA pH 9 . 0 buffer [22] , [57] . Analysis of transcripts of the knock-in alleles was carried out as described previously [22] . For detection of individual transgenic transcripts the following primers were used to detect sequences of: 5′ E-cad wt or knock-in ( 5′EcadUPL_s , AGT GTT TGC TCG GCG TCT/5′EcadUPL_as , GCA AAG CCA TGA GGA GAC C ) ; 3′ E-cad knock-in ( 3′EcadUPL_s , CAC CCC CTT ACG ACT CTC TG/HA-UPL_as , GAC GTC ATA AGG ATA TCC AGC A ) ; 5′ N-cad knock-in ( 5′NcadKIUPL_s , CCA TGG CCA CTA GTA TGT GC/5′NcadKIUPL_as , AAT TTC ACC AGA AGC CTC CA ) ; 3′ N-cad knock-in ( 3′NcadKIUPL_s , GGC CTT AAA GCT GCT GAC AA/3′NcadKIUPL_as , AAC CAT TAT AAG CTG CAA TAA ACA A ) ; Actb ( bActUPL_s , AAG GCC AAC CGT GAA AAG AT/bActUPL_as , GTG GTA CGA CCA GAG GCA TAC ) . | One of the most important steps during mammalian development is the formation of a blastocyst before implantation . Proper blastocyst development is fundamentally reliant on the function of the E-cadherin adhesion molecule , which cannot be replaced by another highly related member of the cadherin family . We have addressed the question of how E-cadherin unfolds its unique function during this central embryonic process . We generated mouse mutants that allow specific domain swapping of extra- and intracellular protein domains of E-cadherin with the corresponding portion of N-cadherin . Upon E-cadherin ( Cdh1 ) depletion , apoptosis is induced in cells that are required to form the trophectoderm , the outer cells of a functional blastocyst . Uncoupling of the two E-cadherin domains demonstrated that specifically the presence of the extracellular domain is indispensable in providing essential survival cues . To establish a proper trophectoderm the insulin-like growth factor I receptor ( Igf1r ) is intimately connected to the E-cadherin–mediated suppression of apoptosis . By interaction of the two proteins Igf1r is efficiently activated to allow embryo survival , blastocyst formation , and implantation . This novel and adhesion-independent function of E-cadherin may serve as paradigm for bifunctionality of adhesion molecules and how they are particularly utilized to interpret signal transduction activities in specific cellular contexts . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetic",
"mutation",
"mechanisms",
"of",
"signal",
"transduction",
"animal",
"models",
"developmental",
"biology",
"model",
"organisms",
"molecular",
"development",
"signaling",
"pathways",
"signaling",
"in",
"cellular",
"processes",
"biology",
"molecular",
"biology",
... | 2012 | Igf1r Signaling Is Indispensable for Preimplantation Development and Is Activated via a Novel Function of E-cadherin |
Could some vaccines drive the evolution of more virulent pathogens ? Conventional wisdom is that natural selection will remove highly lethal pathogens if host death greatly reduces transmission . Vaccines that keep hosts alive but still allow transmission could thus allow very virulent strains to circulate in a population . Here we show experimentally that immunization of chickens against Marek's disease virus enhances the fitness of more virulent strains , making it possible for hyperpathogenic strains to transmit . Immunity elicited by direct vaccination or by maternal vaccination prolongs host survival but does not prevent infection , viral replication or transmission , thus extending the infectious periods of strains otherwise too lethal to persist . Our data show that anti-disease vaccines that do not prevent transmission can create conditions that promote the emergence of pathogen strains that cause more severe disease in unvaccinated hosts .
Infectious agents can rapidly evolve in response to health interventions [1] . Here , we ask whether pathogen adaptation to vaccinated hosts can result in the evolution of more virulent pathogens ( defined here to mean those that cause more or faster mortality in unvaccinated hosts ) . Vaccination could prompt the evolution of more virulent pathogens in the following way . It is usually assumed that the primary force preventing the evolutionary emergence of more virulent strains is that they kill their hosts and , therefore , truncate their own infectious periods . If so , keeping hosts alive with vaccines that reduce disease but do not prevent infection , replication , and transmission ( so-called “imperfect” vaccines ) could allow more virulent strains to circulate . Natural selection will even favour their circulation if virulent strains have a higher transmission in the absence of host death or are better able to overcome host immunity . Thus , life-saving vaccines have the potential to increase mean disease virulence of a pathogen population ( as assayed in unvaccinated hosts ) [2–4] . The plausibility of this idea ( hereafter called the “imperfect-vaccine hypothesis” ) has been confirmed with mathematical models [2 , 5–9] . Efficacy and mode of action are key . If the vaccine is sterilizing , so that transmission is stopped , no evolution can occur . But if it is non-sterilizing , so that naturally acquired pathogens can transmit from immunized individuals ( what we hereafter call a “leaky” vaccine ) , virulent strains will be able to circulate in situations in which natural selection would have once removed them [2] . Thus , anti-disease vaccines ( those reducing in-host replication or pathogenicity ) have the potential to generate evolution harmful to human and animal well-being; infection- or transmission-blocking vaccines do not [2–9] . Note that the possibility of vaccine-driven virulence evolution is conceptually distinct from vaccine-driven epitope evolution ( antigenic escape ) , in which variants of target antigens evolve because they enable pathogens that are otherwise less fit to evade vaccine-induced immunity . The evolution of escape variants has been frequently observed [4 , 10] . The imperfect-vaccine hypothesis attracted controversy [11–14] , not least because human vaccines have apparently not caused an increase in the virulence of their target pathogens . But most human vaccines are sterilizing ( transmission-blocking ) or not in widespread use or only recently introduced [4] . Moreover , unambiguous comparisons of strain virulence and the impact of vaccination on transmission require experimental infections in the natural host—clearly impossible for human diseases . The situation is different for veterinary infections . Here , we report experiments with Marek’s disease virus ( MDV ) , a highly contagious oncogenic herpesvirus that costs the global poultry industry more than $US2 billion annually [15] . We test a key prediction of the imperfect-vaccine hypothesis: that vaccination will elevate the fitness of highly virulent strains above that of less virulent strains . Chickens become infected with MDV by inhalation of dust contaminated with virus shed from the feather follicles of infected birds . In a contaminated poultry house , chicks are infected soon after hatching and remain infectious for life [16] . The virus can remain infectious in poultry dust for many months [17 , 18] . As originally described , Marek’s disease ( MD ) was a paralysis of older birds , but by the 1950s , “acute MD” characterised by lymphomas in multiple organs in younger birds was occurring . This became the dominant form of MD , with increasing virulence , characterised by more severe lymphomas and mortality at increasingly early ages and , under some circumstances , paralysis and death in the first weeks of life , well before lymphoma formation [15 , 19] . MDV has been evolving in poultry immunized with leaky anti-disease vaccines since the introduction of the first vaccines in 1970 [15 , 19–24] . All MD vaccines are live viruses administered to 18-day-old embryos or immediately after hatch , and vaccinated birds can become infected and shed wild-type virus [25–28] . Wild-type MD viruses are so-called serotype 1 viruses . First-generation vaccines include a serotype 3 herpesvirus of turkeys called HVT; second-generation vaccines are a combination of HVT and SB-1 , a serotype 2 isolate . Third-generation vaccines are based on an attenuated serotype-1 virus isolate CVI998 , the so-called Rispens vaccine [15 , 19–24] .
Our first three experiments involved Rhode Island Red ( RIR ) chickens , a breed that has not been subject to the intensive selective breeding and outcrossing that characterizes modern commercial chicken strains . Specific pathogen-free ( SPF ) parent birds were unvaccinated , and so offspring used in our first two experiments were free from maternally derived antibodies . In our first experiment , we infected 8-d-old chicks with five strains of MDV chosen to span the virulence spectrum defined by Witter and colleagues [21 , 29] . The viral strains varied from the less virulent HPRS-B14 , which killed 60% of unvaccinated birds over 2 mo , to the highly lethal Md5 and 675A , which killed all unvaccinated birds in 10 d ( Fig 1 , top panels ) . When age-matched birds were vaccinated 8 d earlier with HVT , the first MDV vaccine to go into commercial use , survival improved dramatically , with a few deaths occurring only late in the experiment , and then only in birds infected with the most virulent strains ( Fig 1 , top panels ) . We collected dust from the isolators containing infected birds and measured the concentration of virus genomes in the dust using real-time PCR . At contemporaneous time points , vaccinated birds shed fewer virus genome copies than unvaccinated birds infected with the same viral strain ( Fig 1 , middle panels ) . Those patterns reflected viral loads in the feather follicles ( S1 Fig ) . Critically , the infectious period of unvaccinated birds infected with our two most virulent strains was less than a week because hosts died so rapidly . During that week , barely any virus was shed ( Fig 1 , middle panels ) . In contrast , the infectious period of the least virulent strains continued for the entire experiment ( almost 2 mo ) . Thus , the least virulent strain shed several orders of magnitude more virus from unvaccinated birds than did the virulent strains ( Fig 1 , bottom panels ) . By preventing death , vaccination greatly increased the infectious period of the most virulent strains , increasing the total amount of virus shed by several orders of magnitude , and increasing it above that of the least virulent strain ( Fig 1 , bottom panels ) . Thus , the net effect of vaccination on both host survival rates and daily shedding rates was to vastly increase the amount of virus shed by virulent strains into the environment . To confirm that virus shed into the environment was a robust proxy for overall bird-to-bird transmission potential , we co-housed birds infected with our three most virulent strains with immunologically-naïve sentinel birds ( Experiment 2 ) . When unvaccinated birds were infected with the two most lethal strains ( Md5 and 675A ) , they were all dead within 10 days ( Fig 2A ) , before substantial viral shedding had begun ( S2 Fig ) . Consequently , no sentinel birds in those isolators became infected ( Fig 2B ) and none died ( Fig 2C ) . In contrast , when HVT-vaccinated birds were infected with either of those hyperpathogenic strains , they survived for 30 days or more ( Fig 2A ) , allowing substantial viral shedding ( S2 Fig ) . All co-housed sentinels consequently became infected ( Fig 2B ) and went on to die as a result of MDV infection ( Fig 2C ) . Thus , in accordance with the imperfect-vaccine hypothesis , vaccination enabled the onward transmission of viruses otherwise too lethal to transmit , putting unvaccinated individuals at great risk of severe disease and death . Interestingly , the viral strain 595 was slightly less virulent than the other two viruses ( taking a day longer to kill half of the unvaccinated birds , and 6 d longer to kill them all ) ( Fig 2A ) . This slightly reduced mortality rate prolonged the viral shedding from unvaccinated birds , so that about 100-fold more virus was shed into the environment by the 595-infected cohort than from the cohorts infected by the two more lethal strains ( S2 Fig ) . This was evidently sufficient for transmission , because all co-housed sentinels eventually became infected ( Fig 2B ) and went on to die ( Fig 2C ) . Thus , slight reductions in lethality can be sufficient to allow onward transmission . Nonetheless , even for strain 595 , vaccination led to more rapid infection of sentinels ( Fig 2B; median time to positivity 9 d earlier than in unvaccinated birds , p < 0 . 05 ) , thus increasing the rate at which secondary cases were generated , a critical determinant of both viral fitness and case incidence in a rising epidemic . The high mortality rates we observed in unvaccinated chickens infected with our most virulent strains are due to early mortality syndrome , which involves the rapid onset of paralysis , disorientation and an inability to feed and move , followed by death [30–33] . In today’s modern industry , parental birds are almost always vaccinated against MDV , which results in the transfer of maternal antibody to chicks . These antibodies appear to be protective against the early mortality syndrome [30–33] . This raises the prospect that vaccination of laying hens could also permit onward transmission of viral strains that would be too lethal to otherwise transmit from offspring birds . We tested this possibility with further experiments using our most ( 675A ) and least ( HPRS-B14 ) virulent virus strains , again in Rhode Island Red birds , but this time including chicks derived from hens vaccinated 4–5 wk prior to egg lay with a standard commercial Rispens vaccine ( Experiment 3 ) . Vaccination of hens enhanced the survival of offspring experimentally infected with HRPS-B14 ( Fig 3A , p < 0 . 05 ) . Maternally derived antibody had no detectable effect on the replication of that viral strain in the feather tips ( S3 Fig panel A , p > 0 . 05 ) and , while it somewhat suppressed the amount of infectious virus shed into the environment early in infections ( Figs 3B and S3B ) , it did not affect the rate at which sentinel birds became infected with HRPS-B14 ( Fig 3C , p > 0 . 05 ) and few sentinels died ( Fig 3D ) . Thus maternal protection had little impact on the transmission success of our least virulent strain . However , presence of maternal antibody greatly impacted the transmission success of the most virulent strain ( 675A ) . As expected , the offspring of vaccinated hens survived for longer following infection with 675A virus than did maternally derived antibody-negative chicks ( Fig 3A , p < 0 . 05 ) . As we found in our first two experiments , very little of the highly virulent strain was shed from birds with no immune protection before they died ( Figs 3B and S3 ) . Consequently , no sentinels became infected ( Fig 3C ) . But birds with maternal protection survived longer to shed more virus ( Figs 3A , 3B , and S3B ) , so that all sentinel birds became infected ( Fig 3C ) and died ( Fig 3D ) . Maternal vaccination was not as protective as direct vaccination of offspring ( cf . Fig 3A with Fig 2A and the top panels of Fig 1 ) . Nonetheless , vaccination of laying hens , like the vaccination of offspring , enabled the onward transmission of the hyperpathogenic strain from offspring ( Fig 3C ) . Again , these data are consistent with the imperfect-vaccine hypothesis . Our experiments above show that direct vaccination of birds or vaccination of parent hens makes possible the onward transmission of viral strains otherwise too lethal to transmit , and thus that unvaccinated individuals are put at increased risk of severe disease and death . However , in a modern commercial broiler setting , all birds in a flock would originate from vaccinated hens ( and so would be positive for maternally derived antibody ) , and also be vaccinated . We thus set out to determine whether our most virulent strain could transmit to vaccinated sentinels , a necessary condition for persistence of hyperpathogenic strains in the modern industry ( Experiment 4 ) . To mimic the current commercial broiler situation , we obtained modern commercial broiler birds derived from Rispens-vaccinated hens and , at 1 d of age , we HVT-vaccinated all the birds we would experimentally infect . Those birds were then infected with our most virulent viral strain ( 675A ) at 8 d of age . We cohoused those experimentally infected birds with sentinel birds , which were either HVT vaccinated or not . We performed this experiment twice . To accommodate changing regulatory requirements ( see Methods ) , we did the first replicate with birds housed in isolators until 35 d of age , after which they were moved to floor until they were 11 wk old ( Experiment 4a ) , and the second replicate with birds maintained in floor pens from 1 d of age until 7 wk of age ( Experiment 4b ) . All sentinels became infected , irrespective of vaccine status ( Fig 4A ) . Thus , vaccinated maternal antibody positive commercial birds shed wild-type virus that caused infections in both vaccinated and unvaccinated maternal antibody positive birds . Vaccination only slightly suppressed viral replication in the infections acquired by the sentinel birds ( Fig 4B ) . Importantly , all sentinels , vaccinated and unvaccinated , became virus positive in the feather follicles , meaning that they themselves started shedding . Vaccination protected sentinel birds from death ( Fig 4C ) , prolonging infectious periods by about 2 wk ( Fig 4D; standard error of the difference ±3 . 2 d , F1 , 36 = 19 . 9 , p < 0 . 0001 ) . Thus , not only does our most virulent strain transmit between modern commercial broilers when they are vaccinated , the duration of shedding in the next step in the transmission chain is also increased by vaccination .
MDV became increasingly virulent over the second half of the 20th century [19 , 21–24] . Until the 1950s , strains of MDV circulating on poultry farms caused a mildly paralytic disease , with lesions largely restricted to peripheral nervous tissue . Death was relatively rare . Today , hyperpathogenic strains are present worldwide . These strains induce lymphomas in a wide range of organs and mortality rates of up to 100% in unvaccinated birds . So far as we are aware , no one has been able to isolate non-lethal MDV strains from today’s commercial ( vaccinated ) poultry operations [19 , 23] . Quite what promoted this viral evolution is unclear . The observation that successively more efficacious vaccines have been overcome by successively more virulent viral strains has prompted many MDV specialists to suggest that vaccination might be a key driver [19–24 , 34–37] , though identifying the evolutionary pressures involved has proved challenging . There is no evidence in Marek’s disease that vaccine breakthrough by more virulent strains has anything to do with overcoming strain-specific immunity ( e . g . , epitope evolution ) ; genetic and immunological comparisons of strains varying in virulence suggest that candidate virulence determinants are associated with host–cell interactions and viral replication , not antigens [19] . The imperfect-vaccine hypothesis was suggested as an evolutionary mechanism by which immunization might drive MDV virulence evolution [2] , but there has been no experimental confirmation . Our data provide that: by enhancing host survival but not preventing viral shedding , MDV vaccination of hens or offspring greatly prolongs the infectious periods of hyperpathogenic strains , and hence the amount of virus they shed into the environment . Our data do not demonstrate that vaccination was responsible for the evolution of hyperpathogenic strains of MDV , and we may never know for sure why they evolved in the first place . Clearly , many potentially relevant ecological pressures on virulence have changed with the intensification of the poultry industry . For instance , as the industry has expanded , broilers have become a much larger part of the industry , and broiler lifespans have halved with advances in animal genetics and husbandry; all else being equal , this would favour more virulent strains [28] , so too might greater genetic homogeneity in flocks [38] or high-density rearing conditions [13] , or indeed increased frequencies of maternally derived antibody if natural MDV infections became more common as the industry intensified in the pre-vaccine era ( Fig 3 ) [39] . But whatever was responsible for the evolution of more virulent strains in the first place ( and there may be many causes ) , our data show that vaccination is sufficient to maintain hyperpathogenic strains in poultry flocks today . By keeping infected birds alive , vaccination substantially enhances the transmission success and hence spread of virus strains too lethal to persist in unvaccinated populations , which would therefore have been removed by natural selection in the pre-vaccine era . The relaxation of natural selection against hyperpathogenic strains revealed by our experiments arises because vaccination enhances host survival . In serial passage experiments with a rodent malaria , immunity induced by whole parasite immunization [40] or vaccination with a recombinant antigen [10] also promoted the evolution of virulence . However , by design , those experiments did not allow host death to impact pathogen fitness , and so the evolution towards increased virulence was driven in a different way . Evidently , immunity in that system is disproportionately efficacious against less virulent strains . Our MDV experiments were not designed to test for within-host selection , but there is some suggestion that vaccine-induced immunity better controlled the replication of the least virulent strain ( Figs 1 and S1 ) . In principle , these two evolutionary pressures ( within-host selection favouring virulent variants for their ability to evade immunity and vaccine-induced relaxation of between-host selection against virulence ) could together generate very potent selection for more virulent strains [4] . Within-host competition between strains could add further selection for higher virulence [41 , 42] . Vaccine failure in the face of virulent pathogens has been documented for at least two viruses other than MDV: feline calicivirus [43] and infectious bursal disease virus in poultry [44] . Both cases are also associated with long-term use of leaky anti-disease vaccines . Our data are also consistent with hypotheses purporting to explain virulence increases in two well-studied wildlife systems . First , strains of the poultry pathogen Mycoplasma gallisepticum in North American house finches have become increasingly virulent , probably due to the increasing incidence of partially immune birds after the bacterium emerged in finch populations in the 1990s [45] . Second , after well-documented declines in virulence following its release as a biocontrol agent in Australia , myxoma virus became increasingly virulent; that virulence evolution was most likely a consequence of increases in the genetic resistance and hence survival of wild rabbits in response to natural selection imposed by the virus [46] . In both cases , anti-disease protection induced by natural immunization ( finches ) or by genetic resistance ( rabbits ) prolonged the infectious periods of otherwise highly lethal strains . These cases and our data raise the prospect that a variety of disease mitigation technologies have the potential to drive virulence evolution , including disease-ameliorating drugs [7 , 47] or genetic enhancements of host resistance [48] . If these technologies prolong infectious periods of hyperpathogenic strains , as we have shown vaccination can , they too could create conditions favouring the emergence of highly lethal strains . This does not mean that such technologies should be avoided , particularly when alternative options are limited . Vaccination has massively reduced yield losses due to MD , despite the evolution [49] . However , when protecting all individuals is impossible , or evolution is ongoing , the use of additional transmission-blocking interventions such as improved hygiene might be essential . We suggest that the risk of outbreaks of hyperpathogenic strains be considered wherever disease interventions improve host survival without preventing pathogen transmission . Such situations might include vaccination against Newcastle disease [50] and avian influenza in poultry [51–53] and vaccination against Brucella in domesticated mammals [54] , as well as genetic enhancement of agricultural animals including fish and poultry . Whether leaky human vaccines could also create the conditions in which more virulent strains can thrive will depend , among other things , on the selective factors currently preventing the emergence of hyperpathogenic strains in human populations . Our data emphasize that a comprehensive understanding of the impact of vaccines on pathogens cannot end with Phase III clinical trials or post-implementation studies of antigenic or serotype frequencies [2 , 4 , 10 , 55] .
The HVT vaccine virus strain FC126 was second chick embryo fibroblast ( CEF ) passage stock from commercial HVT vaccine ( Poulvac , Fort Dodge Animal Health ) . Commercial CVI988/Rispens vaccine virus ( Nobilis Rismavac ) was from Intervet . The challenge virus strains ( seventh duck embryo fibroblast passage stocks ) were a gift from Dr . A . M . Fadly ( Avian Disease and Oncology Laboratory , United States ) . In the MDV literature , virulence ( pathotype ) is defined in terms of vaccine break-through [21 , 27 , 29] , with virus strains categorized into pathotypes denoted as mild , virulent , very virulent , or very virulent plus ( mMDV , vMDV , vvMDV , vv+MDV ) . In our experiments , we used up to five strains chosen to cover this spectrum . The strains were HPRS-B14 , 571 , 595 , Md5 , and 675A . HPRS-B14 has not been formally pathotyped , but would likely be categorised at the lower end of vMDV . The remaining four of these strains have been pathotyped as vMDV , vvMDV , vvMDV , and vv+MDV respectively [21] . Note , however , that for the purposes of the present paper , defining virulence in terms of vaccine resistance introduces semantic circularity . Consequently , in the main text and what follows here , we instead explicitly define ( measure ) virulence as lethality in immunologically naïve birds . For amplification of virus stocks , and to ensure there was no variation in virus passage history between experiments , 5-d-old Rhode Island Red ( RIR ) chickens were inoculated with 1 , 000 plaque forming units ( pfu ) of virus , via the intra-abdominal route . Lymphocytes isolated from spleens harvested at 14 d post infection ( dpi ) were cultured with primary CEF cells for 7 d , when cytopathic effect was clearly visible . The cells were passed two further times in CEF to produce virus stocks . The cell-associated virus stocks were titrated and stored in liquid nitrogen . CVI988/Rispens and HVT vaccines were administered via the subcutaneous route ( neck ) , and challenge virus via the intra-abdominal route . All studies and procedures involving animals were in strict accordance with the European and United Kingdom Home Office regulations and the Animals ( Scientific Procedures ) Act 1986 Amendment Regulations 2012 , under the authority of the Project Licenses PPL 30/2621 and PPL 30/3169 . Birds were individually identifiable with wing bands and had access to water and a vegetable-based feed ad libitum . Any bird deemed to have reached the humane endpoint was culled . In the main text , the humane endpoint was taken as the time at which “infection-induced death” occurred . Chickens which reached the humane endpoint from 5–10 dpi ( early mortality phase ) showed a rapid onset of paralysis , disorientation , reluctance to feed , reluctance to move . and reduced weight gain . In our experience , this endpoint precedes natural death by less than two hours . Chickens which reached the humane endpoint from 15 dpi onwards showed a gradual onset of reluctance to feed , lethargy , and reduced weight gain . In our experience , this endpoint precedes death by up to 24 h . These endpoints , and our estimates of their timing with respect to viral-induced death , were arrived at from small-scale pilot experiments that determined the necessity for close monitoring because of the rapid onset of virus-induced death . Any bird that was found dead was reported to the UK Home Office . The majority of culled chickens showed enlarged spleen with gross lymphoid lesions . The prevalence of visceral lesions was broadly in line with those described by Witter ( 1997 , his Table 4 ) for strains of corresponding pathotypes , despite differences in the breed and maternal antibody status of test chickens , and slight differences in the passage number of viral stocks . For Experiments 1–3 , chickens of the outbred Rhode Island Red ( RIR ) breed were hatched from the eggs of specified-pathogen-free flocks maintained at The Pirbright Institute . Chicks hatched from the eggs of unvaccinated hens were considered free from maternally derived antibody against MDV , and are hereafter referred to as MtAb-neg chicks . Chicks having maternally derived antibody against MDV ( MtAb-pos ) were hatched from eggs collected from RIR hens 4–5 wk after these hens were vaccinated with one commercial dose of Nobilis Rismavac CVI988/Rispens MDV vaccine ( Intervet ) . In Experiments 1–3 , chicks were housed in positive pressure , high efficiency particulate air ( HEPA ) -filtered avian isolators ( Controlled isolation Systems , US ) within rooms in the Experimental Animal House at The Pirbright Institute , Compton . Chickens were monitored up to four times daily , and any chicken considered to have reached the humane endpoint was culled by cervical dislocation . When experiments were terminated , any surviving birds were culled . Post mortem examination was performed on all culled chickens and the presence or absence of gross Marek’s disease lesions recorded . The isolators are designed to house 20 1-d-old chickens or five adult chickens . In groups in which mortality following infection was low , it was necessary to reduce crowding in an isolator at intervals , by culling some birds . In these cases , birds to be culled were randomly selected , and the number of infected and sentinel birds culled was arranged to maintain the appropriate infected:sentinel ratio . Any birds culled for the purposes of reducing crowding were not included in survival data calculations . For experiments 4a and b , commercial broiler breed chicks of the “Aviagen slow growing broiler line” were hatched from eggs supplied by Aviagen . Eggs were from CVI988/Rispens vaccinated hens , and therefore , all birds used were MtAb-pos as confirmed by ELISA ( see below ) . We used different housing protocols for Experiment 4 from those adopted in Experiments 1–3 because regulatory requirements changed over the course of our studies , with work with adult birds in isolators becoming strongly discouraged . This meant we moved to floor-housing birds for at least part of Experiment 4 , a condition that , anyway , more closely resembles housing conditions in the poultry industry . In Experiment 4a , birds were housed in isolators ( as described above ) until they were 35 d of age when groups were moved into floor pens within separate experimental rooms . Floor pens were constructed from metal barred caging panels that could have sections added to the layout to increase the pen area as birds became larger . Compressed straw pellets were used as bedding . In Experiment 4b , birds were housed within floor pens from 1 d of age in separate experimental rooms . To restrict the dissipation of dust and dander in first few weeks , the initial floor pen ( measuring 1 m × 1 m ) was partially contained using Perspex sheets attached to the wire caging to form a housing cube with open edges . As birds increased in size , additional non-Perspex covered cage sections were added to the initial cube to increase the pen area . Two hundred ( n = 200 ) MtAb-neg 1-d-old chicks were randomly allocated to ten groups , each group of 20 chicks being housed in a separate isolator , with two isolators ( A and B ) per room ( S1 Table ) . In each room , chicks in one isolator ( A ) were vaccinated with HVT at 1 d of age , while chicks in the second isolator ( B ) were not vaccinated . At 8 d post vaccination ( dpv ) , all chicks were challenged with one of five strains of MDV , each strain being used to infect a group of unvaccinated chicks , and a group of HVT-vaccinated chicks ( S1 Table ) . Doses of vaccine and challenge viruses were approximately 1000 pfu and 300–600 pfu per chicken , respectively . From each group , ten pre-selected chicks were feather-sampled [56 , 57] twice weekly until 55 dpi or until they reached the humane endpoint . Dust samples were also collected from each isolator twice weekly or until no chickens remained . Each time dust samples were taken , the pre-filter on the isolator air exhaust was removed and replaced with a new , clean filter . Within the isolator , the removed filters were shaken into a polythene bag to collect poultry “dust , ” which was transferred to tubes and stored at −20°C . One hundred and twenty ( n = 120 ) MtAb-neg 1-d-old chicks were randomly allocated to six groups , each group of 20 chicks being housed in a separate isolator , and there being two isolators ( A and B ) per room ( S2 Table ) . In each group , ten chickens were randomly selected to be the “shedder” birds ( i . e . , experimentally infected ) , while the remaining ten chickens were selected to be “in-contact sentinel birds” . In one isolator from each room ( A isolators ) , the shedder birds were vaccinated with HVT at one day of age . In B isolators , the shedder birds were not vaccinated . At 8 dpv all shedder birds were challenged with one of three strains of MDV . Sentinels were neither vaccinated nor challenged . Doses of vaccine and challenge viruses were approximately 2 , 750 pfu and 1 , 000–1 , 500 pfu per chicken , respectively . Sentinel birds are necessary to directly measure natural transmission rates , but once sentinels themselves become infectious , they make it difficult to determine how much virus is being shed by experimentally infected birds . For our studies of 675A , we therefore added two additional treatment groups ( Groups 4A and 4B , S2 Table ) by randomly allocating 20 ( n = 20 ) additional MtAb-neg 1-d-old chicks to two additional isolators , with ten birds in one isolator being vaccinated as above ( A isolator ) , and the ten in the other isolator not being vaccinated ( B isolator ) . All 20 of those additional birds were then experimentally infected at 8 dpv , as above . The existence of these two extra groups allowed us to estimate the viral shedding rates for birds experimentally infected with 675A without any issue of viral contamination from sentinels . We did not , however , have the resources to run analogous extra groups for Md5 or 595 , and so the dust data for those strains ( S2 Fig ) includes dust shed from sentinel birds , which may contain virus after about 20 d post-experimental infection , when the first sentinels began to become infectious ( Fig 2 , S2 Fig ) . Feather samples were collected from every shedder bird cohoused with sentinels twice weekly until 52 dpi or until they reached the humane endpoint . From 3 d onwards , at the same time points unless there were none alive , 150 μL blood samples were collected from every sentinel into 3% sodium citrate . At the same times , dust was collected and stored as described above . For the analysis of the 675A data , we used the survival data for all 20 experimentally infected birds of identical vaccine status ( i . e . , pooling the relevant data from the Group 1 and Group 4 isolators ) , the dust data ( S2 Fig panel B ) from the Group 4 isolators and the feather data ( S2 Fig panel A ) from the Group 1 isolators ( groups defined in S2 Table ) . Sixty ( n = 60 ) MtAb-neg 1-d-old chicks and 60 MtAb-pos 1-d-old chicks were randomly allocated to groups , each group being housed in a separate isolator , and there being two isolators ( A , containing 20 chicks , and B , containing ten chicks ) per room ( S3 Table ) . In A isolators , ten chickens were randomly selected to be shedder birds ( i . e . , experimentally infected ) , while the remaining ten chickens were selected to be in-contact sentinels . At 9 d of age , all shedder birds were challenged with one of two strains of MDV; sentinels were not challenged . In each B isolator , all ten chickens were challenged with one of two strains of MDV , and these isolators were used for collection of dust so that shedding of MDV from ten infected chickens could be accurately determined without “dilution” by dust from non-infected sentinels . Doses of challenge viruses were 100–500 pfu per chicken . Blood ( 150 μL ) was collected from five pre-selected chickens from each of the B isolators prior to challenge . Serum was stored at −20°C . Feather samples were collected from every shedder bird in the A isolators twice weekly until 52 dpi or until they reached the humane endpoint . From 3 dpi onwards , blood samples were collected from every sentinel bird at these same time-points ( or until such time that no chickens remained ) . At each of the above time-points ( or until such time that no chickens remained ) , dust was collected from B isolators and stored as described above . All tested chickens from the MtAb-neg group were negative for anti-MDV antibody ( assayed by ELISA; see below ) , while all tested chickens from the MtAb-pos group were positive for maternal antibody . In both Experiments 4a and 4b , 40 ( n = 40 ) commercial broiler breed chicks were hatched from eggs produced by CVI988/Rispens vaccinated hens and therefore all were MtAb positive . The basic experimental design for Experiments 4a and b was the same; four test groups of age-matched chicks; two groups acting as MtAb-pos , HVT vaccinated , experimentally infected shedder birds housed independently with one of two groups of MtAb-pos sentinel birds that differed in their vaccination status , i . e . , either HVT vaccinated or not ( see S4 Table ) . In both Experiments 4a and 4b , 40 1-d-old chicks were randomly placed into four groups across four isolators ( S4 Table ) . In each experiment , all birds except the ten within each unvaccinated sentinel group were vaccinated with approximately 1 , 500 pfu HVT FC126 via the subcutaneous route at one day of age . At 8 d of age , the two groups serving as shedders were challenged with approximately 725 pfu of vv+ MDV 675A via the intra-abdominal route . Blood ( 150 μL ) and feather samples were collected twice weekly from all birds until 21 dpi and thereafter weekly . Dust samples were collected weekly from the housing air extract filters , as above . Viral titres were assayed indirectly by PCR as follows . Peripheral blood lymphocytes ( PBL ) and feather tip samples were prepared as previously described [56 , 57] . Each dust sample was measured into triplicate 5 mg aliquots . Total DNA was prepared from each PBL , feather and dust sample using a DNeasy-96 kit ( Qiagen ) , according to the manufacturers’ instructions for extraction of DNA from cells ( PBL ) or from tissues ( feather tips and dust ) . Real-time quantitative duplex PCR ( q-PCR ) to amplify the MDV-1 meq gene and the chicken ovotransferrin ( ovo ) reference gene was used for absolute quantification of MDV genomes as previously described [56] . This assay does not detect HVT vaccine virus . All reactions using feather tips or dust samples as target DNA contained 10 μg bovine serum albumin to overcome the inhibitory effect of melanin pigment [56] . Standard curves , prepared using 10-fold serial dilutions of DNA from MDV1-infected CEF ( for meq reaction ) and non-infected CEF ( for ovo reaction ) and accurately calibrated against plasmid constructs of known target gene copy number , were used to quantify MDV genomes per 104 cells or per μg dust . Enzyme-linked immunosorbent assay ( ELISA ) was performed to measure maternally derived antibody against MDV in serum samples from hatched chicks [58] . Serum samples were tested in duplicate at 1:100 dilution on ELISA plates coated with MDV-infected or non-infected cell lysates . Serum from a non-infected chicken was used as a negative-control , and serum from an MDV1-infected chicken as a positive control . All data and the R code used to create all the figures are deposited in the Dryad repository: http://dx . doi . org/10 . 5061/dryad . 4tn48 [59] . For groups of chickens , mean values for virus genome copy number for PBL , feather tips or dust , were determined using the log10 transformed copy number for each individual sample . For feather tip data , 95% c . i . of the means were calculated using the t-distribution . For dust data , 95% c . i . of the mean data were approximated as ±2 standard errors . Plotted values of virus concentration in dust are for samples based on the cumulative dust shed since the previous plotted sample ( when filters were last changed ) . For each sentinel chicken , the time at which MDV was first detected in PBL by qPCR ( time to positivity ) was recorded . Time to positivity was taken as the first sampling time at which the q-PCR Meq Ct value was <40 for successive time-points . Statistical comparisons of survival or time to positivity were made using the Mantel-Cox test applied to Kaplan-Meier survival curves plotted using GraphPad Prism v5 . To assess transmission potential , a key component of viral fitness , we calculated the cumulative virus genome copy number ( VCN ) shed rate over the lifetime of an infection . Cumulative VCN shed is a good proxy for transmission potential because virus shed from feather follicles is the only source of infectious MDV , and because MDV-contaminated dust remains infectious for many months [16–18] . Cumulative VCN shed can be uniquely determined from three components; the dust shed from a bird over time , the concentration of VCN per unit shed dust over time , and the lifespan of an infection . We directly measured the latter two values in Experiments 1–3 , and the former value has been previously estimated [28] . Details of how we used these measures to calculate cumulative VCN are given in S1 Protocol . In Experiment 4 , we estimated viral genome concentration in feather follicles of sentinel birds . Because these birds were co-housed with experimentally infected birds , virus-negative feather shafts can become contaminated with dust from infected birds . We therefore set quantitative thresholds for virus positivity in feather pulp , as described in S2 Protocol . Duration of infectious period ( Fig 4D ) was taken as the time from when viral titres first exceeded this threshold until bird death . | There is a theoretical expectation that some types of vaccines could prompt the evolution of more virulent ( “hotter” ) pathogens . This idea follows from the notion that natural selection removes pathogen strains that are so “hot” that they kill their hosts and , therefore , themselves . Vaccines that let the hosts survive but do not prevent the spread of the pathogen relax this selection , allowing the evolution of hotter pathogens to occur . This type of vaccine is often called a leaky vaccine . When vaccines prevent transmission , as is the case for nearly all vaccines used in humans , this type of evolution towards increased virulence is blocked . But when vaccines leak , allowing at least some pathogen transmission , they could create the ecological conditions that would allow hot strains to emerge and persist . This theory proved highly controversial when it was first proposed over a decade ago , but here we report experiments with Marek’s disease virus in poultry that show that modern commercial leaky vaccines can have precisely this effect: they allow the onward transmission of strains otherwise too lethal to persist . Thus , the use of leaky vaccines can facilitate the evolution of pathogen strains that put unvaccinated hosts at greater risk of severe disease . The future challenge is to identify whether there are other types of vaccines used in animals and humans that might also generate these evolutionary risks . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Imperfect Vaccination Can Enhance the Transmission of Highly Virulent Pathogens |
Dengue is the most widespread vector-borne viral disease caused by dengue virus ( DENV ) for which there are no safe , effective drugs approved for clinical use . Here , by using sequential antigen panning of a yeast antibody library derived from healthy donors against the DENV envelop protein domain III ( DIII ) combined with depletion by an entry defective DIII mutant , we identified a cross-reactive human monoclonal antibody ( mAb ) , m366 . 6 , which bound with high affinity to DENV DIII from all four DENV serotypes . Immunogenetic analysis indicated that m366 . 6 is a germline-like mAb with very few somatic mutations from the closest VH and Vλ germline genes . Importantly , we demonstrated that it potently neutralized DENV both in vitro and in the mouse models of DENV infection without detectable antibody-dependent enhancement ( ADE ) effect . The epitope of m366 . 6 was mapped to the highly conserved regions on DIII , which may guide the design of effective dengue vaccine immunogens . Furthermore , as the first germline-like mAb derived from a naïve antibody library that could neutralize all four DENV serotypes , the m366 . 6 can be a tool for exploring mechanisms of DENV infection , and is a promising therapeutic candidate .
Dengue virus ( DENV ) causes the most prevalent mosquito-borne viral disease . Over 2 . 5 billion people are at risk for infection in over 100 countries , 50–100 million are infected with symptoms , and up to 50 , 000 die from dengue hemorrhagic fever ( DHF ) and dengue shock syndrome ( DSS ) each year [1 , 2] . No specific antiviral drug has been available against DENV infection; the only approved vaccine , Dengvaxia , has caused considerable controversy regarding its safety and potential benefits [3–6] . For decades , anti-DENV vaccine and biological drugs development has been hampered by the high sequence divergence ( 25–40% ) among the four DENV serotypes [7 , 8] . Such divergence leads to the fact that one antibody may not be sufficient to neutralize all DENV infection . Instead , the induced humoral immune response to one DENV infection can enhance the infection and disease processes brought by a subsequent infection with another DENV serotype [2–4] . These findings suggest that the development of new and broadly neutralization antibodies against all the serotypes of DENV could be promising candidate anti-DENV agents , and may also guide the design of effective and safe vaccine immunogens . The DENV envelope glycoprotein ( E protein ) , which mediates virus entry into cells , is the major neutralizing target of antibodies [9–13] . E protein is a type II fusion protein and consists of three domains: DI , DII , and DIII of which DIII has been proposed to contain a receptor binding domain [14–17] . Recent studies revealed that cross-reactive conserved epitopes exist on DII as well as DIII of the DENV E protein [14 , 16–18] . During the naturally-occurring primary DENV infection , a large fraction of the antibody repertoire consists of DII-specific antibodies which are , unfortunately , typically poor in neutralization and may increase the likelihood of severe disease upon subsequent infection through a mechanism known as antibody-dependent enhancement ( ADE ) [18–20] . In contrast , antibodies targeting DIII have proven to be the most potent neutralizing antibodies , but very few could be elicited in naturally infected individuals [18 , 19 , 21–35] . Despite this , previous studies indicated that anti-DENV DIII serotype-specific and cross-reactive antibodies could be elicited using DENV DIII as vaccine immunogen [36–43] and in infected humans [44–47] . It has also been demonstrated that the lysine at position 310 on DIII is the critical residue in the cross-reactive epitope [24] . Therefore , the conserved epitope on DIII represents an attractive target for the development of broadly neutralizing DENV antibodies . Here , we report the isolation of a potent DENV DIII-specific human monoclonal antibody ( mAb ) , designated as m366 . 6 , from a large naïve antibody library constructed by the blood of healthy adult donors . A competitive sorting strategy using a DIII mutant as competitor was applied to identify antibodies precisely targeting the conserved neutralizing epitope . To our knowledge , m366 . 6 is the first human mAb isolated from a naïve antibody library which could neutralize all the four serotypes DENV viruses . Importantly , both heavy and light chain genes of m366 . 6 are very close to their putative germline predecessors . Its fully human origin , the germline-like nature , combined with high-affinity and broad neutralizing activity toward all DENV serotypes , suggest that m366 . 6 is a promising candidate antiviral agent and may also provide a unique template for designing effective dengue vaccine immunogens .
We previously prepared some large naïve antibody libraries using peripheral blood B lymphocytes of non-immunized healthy donors and used them for panning/screening against viral and cancer targets [48–55] . In this study , we used a competitive library sorting strategy to isolate broadly neutralizing antibodies against DENV1-4 ( Fig 1A and 1B ) . The yeast-displayed naïve single chain antibody fragment ( scFv ) library was used to screen against the biotinylated DENV DIII , and , importantly , ten times concentration unbiotinylated DIII K310E mutant was used as the competitor . The yeast cells were selected to present the antibody-expressing cells that could bind well to the wild-type DIII instead of the DIII mutant , resulting in the isolation of antibodies that can target the cross-reactive neutralizing epitopes covering the residue Lysine310 [55] . Potent enrichment was achieved after four rounds of sorting , and a panel of antibodies were identified ( Fig 1B ) . Two antibodies , designated as m360 and m366 , bound potently to DENV DIIIs . Their scFv gene were fused with human IgG1 Fc for protein expression , and surface plasmon resonance ( SPR ) experiments were used to evaluate the antigens binding . The equilibrium dissociation constant ( KD ) of m360 for the DENV1-4 DIIIs were 5 . 8 nM , 5 . 1 nM , 0 . 1 nM and 8 . 3 nM , respectively . The mAb m366 displayed a broader binding profile compared with that of m360 , with the KD of 3 . 3 nM , 1 . 2 nM , 1 . 1 nM and 12 nM to DENV1-4 , respectively ( Table 1 , S1 and S2 Figs ) . To further improve the affinity of m360 and m366 with the four DENV serotypes , we constructed a mutant library using the error-prone PCR technologies . Following three cycles of mutagenesis and selection , two clones were identified from the enriched pool of yeast sorting , designated as m360 . 6 and m366 . 6 . Biacore analysis showed that the cross-reactive binding activities of m360 . 6 and m366 . 6 to all 4 DIIIs were preserved after the affinity maturation process . The KD of m360 . 6 for the DENV1-4 DIIIs were 0 . 3 nM , 42 pM , 2 . 3 pM and 33 nM , respectively ( S3 Fig ) . Although the binding to DENV1-3 DIIIs was improved , the m360 . 6 had only slightly increased binding affinity to DENV4 DIII compared to its parental mAb m360 . Notably , the m366 . 6 exhibited high affinity to all the DENV DIIIs . The KD of m366 . 6 for the DENV1-4 were 0 . 8 nM , 0 . 3 nM , 0 . 3 nM , and 1 . 9 nM respectively , which demonstrated that m366 . 6 could bind to all the four serotype DENV viruses with high avidity ( Table 1 , Fig 2 ) . We also assessed the binding specificity of m366 . 6 by ELISA , and the results showed that m366 . 6 had weak cross-reactivity with Zika virus ( ZIKV ) DIII and no binding with other irrelevant antigens ( S4 Fig ) . Next , we assessed the neutralization capacity of m360 . 6 and m366 . 6 against the four DENV serotypes using a DENV luciferase reporter viral particle ( RVP ) neutralization assay . We used DENV RVPs against the four dengue serotypes that are common strains in DENV research: DENV-1 ( WestPac 74 ) , DENV-2 ( S16803 ) , DENV-3 ( CH53489 ) , and DENV-4 ( TVP360 ) . The luminescent reporter expression was proportional to the number of RVPs added to BHK DC-SIGN cells , confirming the linear correlation between the extent of RVP infection and reporter gene expression . In consistent with the Biacore binding results , both m360 . 6 and m366 . 6 could neutralize all the four serotype DENV , and m366 . 6 displayed better neutralization than m360 . 6 , with the 50% neutralization titers ( IC50 ) of 22 , 2 . 4 , 0 . 85 , and 0 . 36 μg/ml against DENV1-4 respectively ( S5 Fig ) . To further evaluate the neutralization breadth of m366 . 6 IgG against the four DENV serotypes , a standard plaque reduction neutralization assay ( PRNT ) on BHK-21 cells was performed using DENV1-4 live viruses , including DENV-1 128 ( GenBank FJ176780 ) , DENV-1 GZ01/2017 ( S6 Fig , isolated from a DENV-1 infected patient in Guangzhou , China ) , DENV-2 43 ( GenBank AF204178 ) , DENV-3 80–2 ( GenBank AF317645 ) , and DENV-4 B5 ( GenBank AF289029 ) . An irrelevant human mAb G12 was used as the negative control [56] , and 2A10G6 , a broadly neutralizing mAb against all the four DENV serotypes , was used as the positive control [57 , 58] . As shown in Fig 3 , m366 . 6 IgG could neutralize all the four DENV serotypes . The 50% neutralization titers ( IC50 ) of m366 . 6 against DENV1-4 was 12 . 7 , 4 . 57 , 5 . 23 , and 23 . 31 μg/ml respectively ( Table 2 ) . We next used a well-established ADE assay to detect the in vitro ADE effect of m366 . 6 IgG . A mutated form of m366 . 6 IgG was also generated containing the leucine to alanine substitutions at positions 234 and 235 ( m366 . 6 IgG-LALA ) , which lacked binding to Fcγ receptors . The ADE effects of DENV-1 or DENV-2 by m366 . 6 IgG , m366 . 6 IgG-LALA , as well as 2A10G6 were measured . Interestingly , neither m366 . 6 IgG nor m366 . 6 IgG-LALA presented any ADE effect against different serotypes of DENV ( Fig 3F , S7 Fig ) . In contrast , potent ADE effects were observed for the DII-specific mAb 2A10G6 . These results showed that m366 . 6 IgG is a DENV DIII-specific mAb without detectable ADE effect . We further analyzed the sequences of mAbs using the IMGT tool to identify their closest VH and Vλ germline genes . The results indicated that m360 . 6 and m366 . 6 originated from different B-cell lineages ( Table 3 ) . The m360 . 6 VH gene was derived from the IGHV2-70 and the Vλ gene was from IGLV1-51 . In contrast , the m366 . 6 VH gene was derived from the IGHV3-21 and the Vλ gene was from IGLV3-21 . Interestingly , we found that the encoding genes of both m360 . 6 and m366 . 6 closely resembled their corresponding germline gene segments . Notably , m366 . 6 VH and Vλ gene shared 95 . 8% and 95 . 2% sequence identities with the IGHV3-21*01 and IGLV3-21*01 germlines respectively ( Fig 4A and 4B ) . These results indicated that the mAb m366 . 6 is a germline-like antibody , which , in general , could show better drug properties and lower immunogenicity compared to somatically hypermutated antibodies [59] . To further investigate the immunogenetic characteristics of m366 . 6-like antibodies , we analyzed in detail the IGHV3-21 recombination frequencies with specific IGHD and IGHJ genes families from naïve immunoglobulin M ( IgM ) repertoires of 33 health adult donors and neonatal IgM repertoires of 10 newborn babies , using next-generation sequencing data previously generated from our antibodyome studies [48] . By querying the m366 . 6 sequence from the IgM repertoires , 39 sequences were found to display m366 . 6-like V ( D ) J recombination from the genes IGHV-3-21 , IGHD1 , and IGHJ3 out of a total of 10 , 498 , 301 sequences from healthy adult IgM repertoires . In IgM repertoires of newborn babies , a similar recombination frequency was also observed , in which 111 sequences with m366 . 6-like V ( D ) J recombination were found from 5 , 617 , 227 sequences . Our analysis showed that IGHV3-21 is one of the most frequently used IGHV genes , and identified that many of those sequences sharing a significant degree of resemblance to m366 . 6 ( Fig 4C ) . In brief , analysis of these data showed the potential of eliciting robust immune responses with the m366 . 6-like germline antibodies by vaccination . To determine whether m366 . 6 can protect DENV infections in vivo , we firstly used a lethal DENV1-4 infection suckling mouse model . The mice were challenged with DENV1-4 at 200 PFU/mouse via intracranial injection . Four hours later , the mice were treated intracranial with a single dose ( 100 μg ) of m366 . 6 IgG , m366 . 6 IgG-LALA mutant and G12 ( unrelated antibody control ) . These animals were monitored for morbidity and mortality daily . As shown in Fig 5 , all the mice in control groups died from DENV infection , and most of them died within the first two weeks of viral challenge . Interestingly , there was no significant difference in therapeutic efficacy against DENV1-4 infection between m366 . 6 and the LALA-mutated m366 . 6 . The m366 . 6 IgG protected 100% DENV-1 , DENV-3 , DENV-4 and 83% DENV-2 infection whereas LALA-mutated m366 . 6 protected 83% DENV-1 , DENV-4 and 67% DENV-2 , DENV-4 respectively . Therefore , m366 . 6 has no detectable ADE as confirmed in both in vitro and in vivo experiments . We also used the AG129 ( types-I and -II IFN receptor deficient ) mice to test the therapeutic effect of m366 . 6 against DENV-2 ( S8 Fig ) . The results showed that all the mice in the control antibody treatment group died while the survival rate of mice can reach 67% in m366 . 6 treatment group , indicating that the antibody can also protect the lethal infection of DENV-2 in AG129 mice . Taken together , these results indicated that m366 . 6 can protect DENV1-4 infections in vivo . To map the epitope of the germline-like mAb m366 . 6 and identify in greater detail the structural basis of DENV neutralization , we employed multiple approaches ( Fig 6 ) . Sequence alignment of different DENV genotypes and mapping of the conserved amino acid residues of DENV DIII showed that four serotypes DENV DIIIs amino acid residues were different from one another between amino acids 300–393 ( Fig 6A ) . Subsequently , serotype 2 derived DIII consensus gene was randomly mutated to construct a yeast-displayed mutant library . Two rounds of sorting of those yeast cells showing expression on surface but lacking the binding to m366 . 6 was performed . A total of 35 binding escape mutants were aligned with the serotype 2 consensus protein sequence . Mutation frequency at each position was plotted against the residue position number . Similarly , 193 unique DIII sequences derived from naturally isolated serotype 2 dengue viruses from GeneBank were also aligned with the consensus sequence ( Fig 6B and 6C ) . The superimposed profiles of the two set of sequences showed that many of the escaped mutations located in the well-conserved area , indicating the broad cross-reactivity of m366 . 6 to naturally isolated dengue viruses . Besides , the epitope mapping shows that the m366 . 6 epitope is at close to or partially overlaps the dimerization interface between domains II and III . These results may explain why m366 . 6 is a potent cross-reactivity antibody to all the four DENV serotypes . Furthermore , computational docking of DENV DIII-m366 . 6 antibody complex was performed using ZDOCK method . We selected the three top scored docked complexes that contained the key residues identified from an experimental epitope mapping approach . One of the top scored docked models exhibited minimum clashes with appropriate protein interface parameters and was used to demonstrate the lactation the potential epitopes and their interactions with m366 . 6 antibody , which might shed light on the molecular mechanisms of broadly cross-reactive neutralization . Fig 6D showed the docking model of the DIII-m366 . 6 antibody complex in which these epitopes are highlighted in green surfaces . The docking model revealed a different orientation of antibody binding as compared to the DIII complex structure with Fab 1A1D-2 that was previously determined [35] . The epitope comprised of three structurally proximal regions , residues 305–311 in green , 325 , 327 and 361 in dark green , and 383–385 at the C-terminal in lime . One of the key residues , K310 , contacts the CDR-L1 of m366 . 6 which has a germline mutation . In Env-DIII-Fab-1A1D-2 complex structure , the residue K310 contacts the CDR-H1 . The hydrophobic residues , Ile and Trp , of CDR-H3 contact the center part of the epitope , and other loops H1 , H2 , L2 and L3 also involve in the binding . The surface area of the interface between DIII and m366 . 6 antibody in the model complex is 716 Å2 , a typical of antibody-antigen interactions . There are six hydrogen bonds likely to form and no salt bridges at the interface . In brief , the binding regions of the m366 . 6 may be close to or partially overlaps the dimerization interface between domains II and III , which might indicate the broad cross-reactivity of m366 . 6 to the four serotypes of DENV .
Dengue is a disease with a complex immune response orchestrated by host cells partially due to the presence of four serotypes of DENV . Importantly , after a primary DENV infection , one can be protected against or aggravate of a secondary infection with a different serotype , which bring many difficulties to develop an effective vaccine . Thus , it is very urgent to develop an effective and cross-reactive antiviral therapy against DENV infection . Monoclonal antibodies ( mAbs ) are of growing importance for protective and pathogenic immune responses to viruses . At present , there are many therapeutic antibodies to treat viral infections under development , such as antibodies for HIV-1 , SARS-CoV , MERS-CoV , Nipah and Hendra viruses , and H7N9 influenza virus [48 , 50 , 60–66] . Fortunately , screening antibodies from the large naïve libraries has enabled the rapid development of high-affinity human mAbs , especially for the rapid response to the outbreak of emerging viruses and diseases . We recently successfully identified two human germline-like mAbs against MERS-CoV and H7N9 influenza virus from the naïve library , named m336 and m826 , respectively [48 , 50] . They both can naturally exist with very low level of somatic hypermutation in the naïve library with which they have potent binding activity against the envelop proteins of MERS-CoV and H7N9 influenza virus . Most importantly , m336 and m826 all showed highly therapeutic effective in the animal models . Therefore , the naïve library screening can be quickly used to isolate germline-like antibodies that effectively bind to complex protein targets like those in DENV viruses . How to increase the neutralization breadth is a key issue in developing anti-DENV antibodies . Previous studies revealed two classes of broadly neutralizing antibodies to flaviviruses , including antibodies targeting the conserved epitopes in DII or DIII [16–18 , 20] . While the conserved fusion loop epitope ( FLE ) in DII is the immunodominant epitope in E protein , unfortunately , this epitope frequently induced poorly neutralizing and strongly infection-enhancing antibodies via ADE [18–20] . Therefore , DIII represents the ideal target for neutralizing antibodies . In this study , we applied a highly efficient yeast-display-based sorting strategy by using the highly diverse DENV DIIIs as antigen and the competitive sorting technique . By applying this method , we quickly and efficiently identified two human germline-like broad-spectrum anti-DENV mAbs ( m360 and m366 ) from the naïve scFv yeast library using the DIII antigen that make them as promising candidate therapeutics as well as the template for vaccine development . Another class of highly efficient broadly neutralizing antibodies that target the envelope dimer epitopes ( EDE ) from the secondary acute DENV infection plasmablasts has been identified by Dejnirattisai et al . [67] . These antibodies may especially get through with high somatic mutations from the secondary virus infection . Compared with the highly somatically mutated antibodies , germline-like antibodies typically have better safety and drug-related property [59] . Importantly , the Hendra and Nipah antibody m102 . 4 is a near germline antibody and exhibited a very good drugability , which was from a similar library that was also used to isolate our m366 and m366-like antibodies . m102 . 4 was successful as a candidate therapeutic mAb in animal models and was also completed the phase I clinical trial ( ACTRN12615000395538 ) without side effects [64] . To further improve the affinity of m366 with the four serotypes DENV DIIIs , we subjected m366 to affinity maturation process , and named it as m366 . 6 . Subsequently , we analyzed m366 . 6 sequence using the IMGT tool to identify its closest VH and Vλ germline genes . Interestingly , we found that m366 . 6 is still a germline-like antibody although it went through the mutation process in vitro , with over 95% identities of its VH and Vλ genes to the IGHV3-21*01 and IGLV3-21*01 germline respectively . In order to evaluate the neutralization effect of the m366 . 6 IgG , we used a standard plaque reduction neutralization with BHK21 cells to measure DENV infection and neutralization . The m366 . 6 IgG showed broadly neutralization towards the four serotypes DENV as well as a recent DENV isolate from clinical samples . More importantly , m366 . 6 did not present any ADE effects in different serotypes of DENV . The in vivo study results demonstrated the therapeutic potential of m366 . 6 against severe DENV1-4 infections . In brief , the m366 . 6 could neutralize the four serotypes DENV in vitro and protect the DENV infection mouse model in vivo without detectable ADE effects . We therefore expect that m366 . 6 has a likeness drugability of m102 . 4 and could be developed as a candidate therapeutic in the future . We have also localized the m366 . 6 epitope by using a combination of computational structural modeling , display-based antigen mutagenesis , and sequence-based analysis of mutants . The epitope appears to overlap with the epitope previously explored as targets for cross-reactive murine mAbs and close to or partially overlaps the dimerization interface between domains II and III . This further indicates that this epitope could be an important component of vaccine immunogens intended to elicit cross-reactive neutralizing antibodies . In progress are our experiments to crystallize the complex of m366 . 6 with DENV DIII that would allow precise determination of the m366 . 6 epitope . The major result of this study is the identification of a germline-like human mAb , m366 . 6 , from a naïve yeast antibody library which binds with high ( picomolar ) affinity to DIIIs from all serotypes and neutralizes the four DENV serotypes . There are two major implications from this finding: 1 ) m366 . 6 is a potential candidate therapeutic which could be further developed in preclinical and clinical settings . 2 ) the epitope of the germline-like mAb m366 . 6 could guide the design of effective candidate vaccine immunogens capable of eliciting m366 . 6 and/or m366 . 6-like antibodies .
BHK21 cells were cultivated in Dulbecco’s Modified Eagle Medium ( DMEM ) supplemented with 10% fetal bovine serum ( FBS ) ( Biowest ) . Mosquito cells C6/36 were cultured in RPMI-1640 medium supplemented with 10% FBS . All cells were maintained in a humidified atmosphere of 5% CO2 at 37°C incubator , except for C6/36 cells , which were cultivated at 28°C . DENV-1 128 ( GenBank FJ176780 ) , DENV-1 GZ01/2017 ( isolated from DENV-1 infected patient in Guangzhou ) , DENV-2 43 ( GenBank AF204178 ) , DENV-3 80–2 ( GenBank AF317645 ) , and DENV-4 B5 ( GenBank AF289029 ) were propagated in C6/36 cells by using RPMI 1640 medium and the titers were measured by standard plaque forming assay in BHK21 cells . DENV DIII genes from all 4 serotypes were synthesized by Genescript , Inc ( Nanjing , China ) , fused with IgG1 Fc and a C-terminal Avi-tag , and cloned into pSecTag expression vector . The DIII . 3 ( serotype 3 ) K310E mutant was generated through overlapping PCR . For the conversion of IgG1 from scFv , the heavy and light chains of scFv were amplified and recloned into the PTT-IgG1 vector . The plasmids were transfected into Expi293 cells ( Thermo Fisher ) for transient expression , and purified using protein G column ( GE Healthcare , Piscataway , NJ ) according to the manufacturer’s instructions . The purified protein was biotinylated by mixing with biotinylation reagents in PBS for 30 min on ice , according to the manufacturer’s instructions ( Pierce ) . A large yeast-displayed scFv library was used for antibody screening , and the screening protocols were essentially carried out as described previously [55] . Briefly , 10 μg of binotinylated DIII . 3-Fc and 1010 cells of the initial naïve library were mixed and washed by PBSA , and incubated with 100 μl streptavidin conjugated microbeads ( Miltenyi Biotec , Auburn , CA ) before loading onto the autoMACS system ( Miltenyi Biotec ) for sorting . After three rounds of sorting , the downsized library was further sorted against binotinylated DIII . 3-Fc ( 1 μg/ml ) but also using unbiotinylated K310E mutant ( 1 μg/ml ) as the competitor . The cells were stained by the addition of mouse anti-c-Myc antibody ( Roche ) , Alexa-488 conjugated goat-anti-mouse antibody ( Invitrogen ) , and PE-conjugated streptavidin ( Invitrogen ) for sorting on a FACSAria II cell sorter ( BD Biocsiences , San Jose , CA ) to isolate the positive binders . The plasmids of the positive clones were prepared by using Zyppy Plasmid Miniprep Kit according to the manufacturer’s instructions ( Zymo Research ) . To generate the m360 scFv and m366 scFv mutant libraries , random mutagenesis of the scFv genes were performed through error-prone PCR by using a GeneMorph II kit ( Stratagene ) following the manufacturer’s instructions with minor modifications . To further diversify the mutation profile , 3 uM of each of the two nucleotide analogues ( 8-oxo-deoxyguanosine triphosphate and 2'-deoxy-p-nucleoside-5'-triphosphate ) was mixed in the PCR reaction mixture . For the second and third cycle library constructions , an extra step of DNA shuffling PCR was inserted into the regular PCR cycles to combine the beneficial mutations obtained from previous maturation process . DNA shuffling PCR step was performed as following: 20 cycles of denature at 94 °C for 15 seconds followed by annealing/extension at 68°C for 1 second on the Bio-Rad MyCycler . Binding affinities of m360 scFv , m366 scFv , m360 . 6 scFv , and m366 . 6 scFv to the 4 DENV DIIIs were analyzed by surface plasmon resonance technology using a Biacore X100 instrument ( GE healthcare ) . The antibodies were covalently immobilized onto a sensor chip ( CM5 ) using carbodiimide coupling chemistry . A control reference surface was prepared for nonspecific binding and refractive index changes . For analysis of the kinetics of interactions , varying concentrations of antigens were injected at flow rate of 30 μl/min using running buffer containing 10mM HEPES , 150 mM NaCl , 3 mM EDTA , and 0 . 05% Surfactant P-20 ( pH 7 . 4 ) . The association and dissociation phase data were fitted simultaneously to a 1:1 Langumir global model by using the nonlinear data analysis program BIAevaluation 3 . 2 . All the experiments were done at 25°C . Neutralizing activity of mAbs was measured using a standard plaque reduction neutralization with BHK21 cells as previously described [57] . Briefly , 5-fold serial dilutions of mAbs were added to approximately 200 PFU of a variety of dengue virus strains and incubated for 1 h at 37°C . Then , the mixture was added to BHK21 cell monolayers in a 12-well plate in duplicate and incubated for 1 h at 37 °C . The mixture was removed , and 1 ml of 1 . 0% ( w/v ) LMP agarose ( Promega ) in DMEM plus 4% ( v/v ) FBS was layered onto the infected cells . After further incubation at 37 °C for 4 days , the wells were stained with 1% ( w/v ) crystal violet dissolved in 4% ( v/v ) formaldehyde to visualize the plaques . PRNT50 values were determined using non-linear regression analysis . PRNT50 data were calculated by doing a non-linear regression analysis using Sigmaplot ( Version 9 . 01 , Systat Software , Inc . , CA ) as previously described [57] . DENV RVPs from all four serotypes were pre-incubated with an equal volume of serially diluted antibodies ( 25 μg/ml to 0 . 0012 μg/ml pre-dilution or 12 . 5 μg/ml to 0 . 0006 μg/ml pre-dilution , as measured based on the dilution of antibody prior to combining with RVPs ) in DMEM infection media for 1 h at room temperature and transferred to wells of a 96-well plate . An equal volume of DENV RVPs were added to each well followed by slow agitation for 1 h at room temperature . BHK DC-SIGN cells were added to each well at a density of 30 , 000 cells per well followed by incubation at 37°C in 5% CO2 for 48 h . Cells were subsequently fixed in lysed and analyzed for luminescent reporter expression using the Wallac Victor . The percent infection for each concentration of mAb or serum was calculated , and the raw data was expressed as percent infection versus log10 of the mAb concentration or the reciprocal serum dilution . The data were fit to a sigmoidal dose-response curve using Prism ( GraphPad Software , La Jolla , CA ) to determine the titer of antibody that achieved a 50% reduction in infection . Maximum infection was determined in the absence of antibodies . The in vitro ADE assay was performed using K562 cells [57] . Briefly , serial 10-fold dilutions of antibodies under concentrations ranging from 100 to 0 . 01 μg/ml were mixed with DENV-1 or DENV-2 , and incubated for 1 h at 37 °C . Mixtures were then added to 2×105 K562 cells at multiplicity of infection of 0 . 1~0 . 25 for 2 h in 24-well plates . The cells were subsequently washed 3 times with serum free RPMI-1640 medium . After collecting cells by centrifugation , the cell pellets were re-suspended with RPMI-1640 medium containing 2% FBS and added to 24-well plates , then incubated for 4 days at 37 °C with 5% CO2 . The titer of viruses in the supernatant was then measured using a plaque assay . The ADE effect was calculated as different viral yields in the supernatant after infection in the presence of the added antibodies . The epitope mapping of m366 . 6 was performed using previously described protocols [55] . Briefly , random mutagenesis of the DENV DIII . 2 gene was performed using a GeneMorph II kit ( Stratagene ) . As described above , the yeast-displayed mutant library was mixed with biotinylated m366 . 6 scFv-Fc , washed , and stained by mouse anti-c-Myc antibody ( Roche ) , Alexa-488 conjugated goat-anti-mouse antibody ( Invitrogen ) , and PE-conjugated streptavidin ( Invitrogen ) . After two rounds of sorting on a FACSAria II cell sorter ( BD Biocsiences , San Jose , CA ) , the sorted cells were amplified and their plasmids were prepared and sequenced . Homology modeling of variable regions of heavy ( VH ) and light ( VL ) chains for m366 . 6 scFv antibody was carried out using the SWISS-MODEL workspace [68] by selecting the closest template structures ( PDB codes: 3QOS for heavy chain and 2DD8 for light chain ) , whose sequence similarities were 92% and 87% respectively . The VH-VL orientation of m366 . 6 scFv structure was assigned similar with one of the templates ( PDB code: 2DD8 ) that showed minimum steric clash for creating the final m336 . 6 scFv model . The crystal structure of DENV DIII serotype 2 ( PDB code: 2R29 ) was used for docking with the modeled scFv antibody m366 . 6 . Docking of scFv m336 . 6 to the dengue Env-III was performed by ZDOCK server ( http://zdock . bu . edu ) that uses a fast Fourier transform ( FFT ) -based rigid-body protein docking algorithm with scoring functions combining pairwise shape complementarity , desolvation and electrostatic energies . Based on the escape mutants that led to the loss of epitopes and available crystal structure of DENV DIII , we selected a list of residues as biological constrains , 307 , 309 , 310 , 311 , 327 , 361 and 383 , on the surface of Env-DIII as potential contacting residues for docking constraints . Similarly , one or two residues from each of CDR-H1 , H3 and L3 loops were chosen at the docking interface . CDR-H1 and H3 loops had dominant hydrophobic residues whereas CDR-L1 had a germline mutation , and they all had high antigen-contacting propensities [69] . Results from the top 2000 ZDOCK predictions were filtered using the user-defined residues and a 6 angstrom distance cutoff . Three predicted complexes were only kept as all residues selected come together at the interface and were further examined by PDBePISA ( Protein Interfaces , Surfaces and Assemblies ) . PyMOL was used for the analysis of docked model and graphical illustration [70] . The suckling mice were purchased from B&K Universal Group Limited ( Shanghai , China ) and housed under specific pathogen-free conditions at the animal facilities of the Shanghai Public Health Clinical Center , Fudan University ( Shanghai , China ) . Before infection , the mice were transferred to the Animal Biosafety Level 2 ( BSL-2 ) Laboratory ( Shanghai , China ) . One day mice were used for all experiments . All mice were intracerebrally injected with 200 PFU of DENV1-4 . At 4 h post infection , mice were passively transferred a single dose of 100 μg antibody m366 . 6 IgG , m366 . 6 IgG LALA mutant or G12 IgG as the negative control via intracerebrally injection . Survival rates and disease sings were monitored daily . The AG129 mice ( type I and type II interferon receptor-deficient ) were purchased from B&K Universal Group Limited ( Shanghai , China ) and housed under specific pathogen-free conditions at the animal facilities of the Shanghai Public Health Clinical Center , Fudan University ( Shanghai , China ) . Before infection , the mice were transferred to the BSL-2 Laboratory ( Shanghai , China ) . Groups of mixed-sex 4- to 6-week-old mice were used for all experiments . All mice were intraperitoneally injected with 2x106 PFU of DENV-2 in a volume of 200 μL . At 16 h post infection , mice were passively transferred a single dose of 500 μg antibody m366 . 6 IgG-LALA , or G12 antibody as the control via i . p . injection . Survival rates , weight loss , and disease sings were monitored daily . Specific-pathogen-free AG129 mice ( 4–6 weeks old ) and suckling mice were used for all experiments . All experimental protocols were reviewed and approved by the institutional committee of Fudan University ( Permit Number: 2018-A056-02 ) in accordance with the Guideline for Ethical Review of Animal Welfare ( GB/T 35892–2018 ) of the Chinese National Health and Medical Research Council ( NHMRC ) . | Dengue virus infects 50–100 million people each year . Infection is initiated by entry of the virus into cells mediated by the viral envelope glycoproteins . There are four closely related DENV serotypes , but they all are antigenically distinct , with each comprising several genotypes that exhibit differences in their infection characteristics in both the mosquito vector and in the human host . One of the confounding problems that has faced vaccine and biological drugs development for decades is the inability of antibodies to one serotype to protect against infection by another one . Instead , the induced humoral immune response to one dengue virus infection can enhance the infection and disease processes brought by a subsequent infection with another dengue serotype . In this study , by using a competitive sorting strategy to interrogate a human naïve antibody library , we identified a cross-reactive mAb , designated as m366 . 6 , against the four DENV serotypes . The mAb m366 . 6 possesses only few somatic mutations from the closest VH and Vλ germline genes and high affinity to DIII . Most importantly , the germline-like m366 . 6 demonstrated a broad spectrum of neutralization against the four DENV serotypes . Thus , m366 . 6 is a promising candidate therapeutics and its epitope may imply on the design of effective vaccine immunogens to elicit m366 . 6-like antibodies in vivo . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"dengue",
"virus",
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"animal",
"models",
"of",
"disease",
"immunology",
"microbiology",
"viruses",
"animal",
"models",
"vaccines",
"preventive",
"... | 2019 | A broadly neutralizing germline-like human monoclonal antibody against dengue virus envelope domain III |
TNF expression of macrophages is under stringent translational control that depends on the p38 MAPK/MK2 pathway and the AU–rich element ( ARE ) in the TNF mRNA . Here , we elucidate the molecular mechanism of phosphorylation-regulated translation of TNF . We demonstrate that translation of the TNF-precursor at the ER requires expression of the ARE–binding and -stabilizing factor human antigen R ( HuR ) together with either activity of the p38 MAPK/MK2 pathway or the absence of the ARE-binding and -destabilizing factor tristetraprolin ( TTP ) . We show that phosphorylation of TTP by MK2 decreases its affinity to the ARE , inhibits its ability to replace HuR , and permits HuR-mediated initiation of translation of TNF mRNA . Since translation of TTP's own mRNA is also regulated by this mechanism , an intrinsic feedback control of the inflammatory response is ensured . The phosphorylation-regulated TTP/HuR exchange at target mRNAs provides a reversible switch between unstable/non-translatable and stable/efficiently translated mRNAs .
TNF is a master cytokine of inflammatory signaling of macrophages . Its biosynthesis is tightly controlled to allow rapid secretion but also to avoid delay or leakiness in its down-regulation , which could result in exaggerated or persistent inflammation . The levels of regulation comprise transcription , processing , nuclear export and stability of the TNF mRNA , translation of pro-TNF , and shedding of TNF ( reviewed e . g . in [1] ) . Pro-TNF contains a leader sequence of 79 amino acids ( for mouse TNF ) and is synthesized as a type II membrane protein [2] , [3] . After initiation of ribosomal translation of TNF mRNA , followed by SRP-mediated arrest of ribosomal synthesis of the nascent protein chain and docking to the ER membrane , the C-terminal part of TNF containing the potential cleavage site is synthesized into the lumen of the ER . Subsequently , pro-TNF is transported in an LPS-stimulated manner from the trans Golgi network to the cell surface using tubular carriers that fuse with the recycling endosome [4] . At the cell surface , pro-TNF is cleaved and released by the TNF-converting enzyme TACE/ADAM17 [5] . The p38 MAPK/MK2/3 pathway [6] regulates TNF-biosynthesis mainly at the translational level . Inhibition of this pathway by small molecules , such as SB203580 or SB202190 , or deletion of its components , such as the downstream protein kinase MK2 , lead to a significant reduction of LPS-induced TNF production of macrophages although the LPS-stimulated increase in TNF mRNA concentration remains almost unaltered and the stability of mature TNF mRNA in these cells is only sightly reduced [7]–[9] . The post-transcriptional regulation of TNF biosynthesis by the p38 pathway depends on the AU-rich element ( ARE ) in the 3′ non-translated region of TNF mRNA [9] , [10] . So far the molecular mechanisms regulating ARE-dependent translation of pro-TNF via phosphorylation are not understood . However , various mRNA- and ARE-binding proteins have been identified as substrates of the p38 MAPK pathway , e . g . hnRNP A0 , tristetraprolin ( TTP ) , KSRP and poly ( A ) -binding protein 1 [11] , [12] , [13] , [14] , [15] . The mRNA-ARE and corresponding ARE-binding proteins ( ABPs ) , such as TTP , KSRP , HuR , TIA-1 and AUF1 , are mainly held responsible for the regulation of mRNA metabolism governed by exosome- , PARN- and CCR4/Not1-dependent degradation of mRNAs or by their storage in discrete cytoplasmic foci ( reviewed in [16] and [17]–[19] ) . For example , KSRP stimulates the rapid decay of ARE-containing mRNAs and its activity is inhibited via direct phosphorylation by p38 MAPK providing a mechanism of stress-dependent stabilization of ARE-containing mRNAs [13] . In contrast , HuR ( ELAV ) is a factor of constitutive nuclear and cytoplasmic stabilization of ARE-containing mRNAs [20] , [21] , which also binds to the ARE of TNF mRNA [22] , [23] . However , besides its mRNA stabilizing function , HuR also influences translation of specific mRNAs as measured by the association of these mRNAs to the polysomal fractions of cell lysates separated by density centrifugation . Gene-silencing of HuR blocks polysomal localization of cytochrome C- and nucleolin-mRNA indicating a positive effect of HuR on translation of these mRNAs [24] , [25] . In contrast , deletion of HuR increases the long-term shift of TNF mRNA to polysomal fractions after LPS treatment of macrophages [26] and its overexpression leads to a reduction of TNF-mRNA in polysomal fractions [27] indicating also an inhibitory effect of HuR on translation . The inhibitory effect of TTP in the regulation of TNF production first became obvious by significant cachexia in the TTP knockout mouse [28] , which was explained by increased TNF concentrations and a feedback effect of TTP on TNF production by its binding to the ARE and destabilization of the TNF mRNA [29] . Subsequently , TTP could be identified as a destabilizing factor for various ARE-containing mRNAs , including its own mRNA ( reviewed in [16] ) . In LPS-induced TNF biosynthesis TTP is genetically downstream to p38 MAPK and MK2 , since its deletion neutralizes the defect in LPS-induced TNF production seen in MK2 knockout mice [30] . MK2 directly phosphorylates TTP [11] . It is proposed that phospho-TTP/mRNA complexes are sequestered by 14-3-3 binding proteins [15] and that phospho-TTP is unable to recruit deadenylases [17] , [19] resulting in target mRNA stabilization . Via phosphorylation of SRF , MK2 also contributes to transcriptional activation of the TTP gene [31] . Interestingly , in MK2 knockout and , especially , in MK2/MK3 double knockout ( DKO ) macrophages a strong reduction of the TTP concentration is observed [32] suggesting a major role of the p38/MK2/3 pathway in the regulation of TTP expression . The predominant translational regulation of TNF by the p38 MAPK/MK2/3 pathway raises the question of the role played by certain direct substrates of these protein kinases and the mechanisms involved in this translational regulation . Here , we reconstituted MK2-dependent translational regulation of TNF in immortalized MK2/3-deficient mouse macrophages by re-introducing MK2 , its catalytic dead mutant or , as a control , GFP alone . We analyzed the requirements of MK2-dependent translation of native TNF mRNA for the presence of ABPs , such as TTP and HuR , and for the mRNA-binding MK2-substrate Ago2 . In this analysis , the combination of cytosol/ER-fractionation and subsequent polysome profiling provides additional mechanistic insights into the translational regulation of pro-TNF . The regulatory mechanisms of translation of TNF mRNA elucidated by this approach are also valid for TTP mRNA and contribute to the stringent feedback regulation of the inflammatory response .
We generated an immortalized macrophage cell line from MK2/MK3 double-deficient mice [32] by expression of v-raf and v-myc in bone marrow derived macrophages ( BMDM ) . For restoration of the “wild type” situation in these macrophages , we subsequently expressed MK2 by stable retroviral transduction using pMMP-MK2-IRES-GFP . As knockout control for this cell line , we used pMMP-IRES-GFP . To differentiate between the effects of catalytic activity of MK2 and the effects of MK2-dependent stabilization of p38 in the binary complex [33] , we also rescued the cell line with the catalytic dead mutant of MK2 ( pMMP-MK2K79R-IRES-GFP ) [31] . The generation of different cell lines by retroviral transduction after initial immortalization excludes the unwanted influence of different random events of immortalization in the cell lines to be compared . After retroviral transduction , cell lines were sorted and selected by preparative FACS for comparable expression of GFP . The level of expression of MK2 in the rescued cell lines was comparable to its level in immortalized wild type BMDM and in RAW264 . 7 cells ( Figure S1 ) . We compared basal and LPS-dependent expression and phosphorylation of the central components of the pathway ( p38 , MK2 ) and of relevant substrates of MK2/3 ( TTP , NOGO-B ) in the cell lines ( Figure 1A and Figure S2A ) . As controls , we also analyzed expression of HuR , TIA-1 , KSRP , GFP and GAPDH . While these controls showed comparable expression in both cell lines and were not induced by LPS , we detected strong induction of p38 activity by LPS in both cell lines . Although decreased stabilization of p38 protein by the lack of MK2/3 leads to reduced p38 concentrations in GFP-transduced MK2/3-deficient cells , the activity of p38 , detected by the antibody which only detects the dual phosphorylation in the activation loop , was similar in MK2-rescued and GFP-transduced cells . Obviously , increased p38 activation compensates for its lower expression not only in neurons [34] , but also in macrophages . As expected , MK2 is only detected and activated by LPS-treatment in MK2-rescued cells . Interestingly , there is a rapid induction of the MK2/3 substrate TTP by LPS-treatment , qualifying TTP as an immediate early gene . This LPS-induced expression of TTP is strongly reduced in GFP-transduced cells , a fact already known from MK2- and MK2/3-deficient primary cells [32] . In addition , a lack of phosphorylation of the ER membrane resident MK2 substrate NOGO-B , which is phosphorylated in its cytoplasmic domain [35] , is detected in GFP-transduced compared to MK2-rescued cells based on the loss of the slower migrating phosphorylated isoform of NOGO-B . Since the kinetics of p38/MK2 activation and TNF production in macrophages is fast ( MK2 activity peaks after 20 min , maximum of TNF production is reached after about 60 min ) , we determined TNF-mRNA and -protein concentration 1 h after LPS-treatment . The relative intracellular TNF mRNA concentration as represented by the TNF mRNA/actin mRNA ratio and the length of the polyA-tail of TNF-mRNA do not significantly differ between MK2-rescued and GFP-transduced cells ( Figure 1B and Figure S3 ) . In contrast , both pro-TNF protein and secreted TNF are significantly increased in MK2-rescued cells compared to the GFP-transduced control ( Figure 1C , 1D ) . Thus , the translational control of TNF biosynthesis by MK2 is clearly reflected in this cellular system . The catalytic activity of MK2 is necessary for this control , since rescue of the macrophages with the kinase-dead mutant MK2-K79R does not release the translational repression of TNF ( Figure S2 ) . To monitor ribosome occupancy of TNF mRNA in these cells as a measure of translational initiation and elongation , we applied density gradient centrifugation to distinguish between TNF mRNPs , TNF mRNA containing monosomes and polysomes . Furthermore , since pro-TNF is synthesized as a type II membrane protein by ER-directed translation , we decided to combine cytosol/ER-fractionation by saponin treatment , which was modified after [36] , with subsequent density centrifugation . A typical fractionation is shown in Figure 2A . While mRNPs , ribosomal subunits , scanning ribosomal subunit , initiated translation and stalled ribosomes with signal recognition particle ( SRP ) bound to the nascent peptide chain before docking to the ER membrane should distribute between fractions 1–4 , polyribosomes are expected in fraction 6 and above . Polyribosomes containing mRNAs of secreted or membrane proteins , such as pro-TNF , are expected in the polyribosomal fractions of the ER subfraction , while polyribosomes containing mRNAs of cytosolic proteins together with all monosomes including SRP-stalled monosomes with nascent proteins are expected in the cytosolic subfraction . The overall distribution of TNF mRNA after LPS-stimulation differs between MK2-rescued and GFP-transduced cells as seen for the cytoplasm/ER ratio , which is significantly lower in MK2-rescued cells ( Insert in Figure 2A and Figure S4 for absolute mRNA levels ) , while the distribution of ß-actin mRNA is not significantly different . Thus MK2 is required for preferential ER localization of TNF mRNA . We then performed polysome profile analyses of specific mRNAs . To ensure biological significance , we always performed at least two independent experiments comprising separate LPS-stimulation of cells , density gradient fractionation of cell lysates and qRT-PCR detection of specific mRNAs . Data were only considered significant in the instances where the biological repeats yielded the same qualitative results . The repeats of key experiments are displayed in Figure S5 . As seen in the polysome profiles for TNF mRNA ( Figure 2B ) , its cytosolic population occurs mostly in the free mRNP fraction . Only a low concentration of monosomal complexes of TNF mRNA is detected in the cytosolic fraction of MK2-transduced cells and , to a slightly lower degree , also in GFP-transduced cells . In the ER fraction , a clear peak of larger polysomal complexes of TNF mRNA ( fractions 9–10 ) is seen in the presence of MK2 and this peak is completely missing in the absence of MK2 in cells transduced with GFP only . In parallel , the free mRNP signal is decreased for TNF mRNA in the presence of MK2 . The existence of the larger polysomal complexes of TNF mRNA depends on catalytic activity of p38 and MK2 , since the peak corresponding to larger polysomal complexes is reduced after treatment of the cells with the p38 inhibitor SB202190 and completely disappears for cells expressing the catalytic dead MK2 mutant instead of wild type MK2 , respectively ( Figure 2B ) . As a control , efficient translation of actin mRNA in ER and cytosol fractions is detectable for both MK2-rescued and GFP-transduced cells ( Figure 2C ) . The finding that ß-actin mRNA coding for a cytosolic protein is also translated at the ER indicates the fact that not all proteins synthesized at the ER are secreted or integral membrane proteins [37] and that ER-associated mRNAs serve a global role also for translation of cytosolic proteins [38] . As further control , we monitored the mRNA distribution of the secreted chemokine KC ( Cxcl1 ) , which is also synthesized as pro-protein at the ER and which is known to be regulated by MK2 at the level of mRNA stability , but not at the level of translation [32] . KC mRNA distribution is independent of the presence of MK2 ( Figure 2D ) . Efficient translation of this mRNA proceeds mainly at the ER . However , KC mRNA peaks in earlier polysome fractions ( around 7–8 ) compared to TNF mRNA ( peak at fractions 9–10 ) , since the ORF of KC mRNA is 291 nucleotides only , compared to 708 nucleotides of TNF mRNA , allowing only a smaller number of ribosomes to elongate at the same mRNA molecule . In contrast to ß-actin mRNA , KC mRNA in the cytosolic fraction is mainly detected in monosomes . These monosomes probably represent translationally initiated monosomes with the SRP-arrested nascent protein chain . In cytosolic fractions the amount of ribosome-free KC mRNPs is comparable to the amount of KC mRNA in monosomes . In contrast , the amount of ribosome-free TNF mRNPs is significantly higher than the amount of monosomal TNF mRNA , indicating that initiation is the critical step for TNF mRNA translation . Taken together , this fractionation analysis clearly demonstrates a specific role for the catalytic activity of the protein kinase MK2 in the regulation of ER-directed translation of pro-TNF in macrophages . To demonstrate that the cell lines generated reflect the situation in primary BMDMs we extended the analysis of LPS-induced MK2-dependent translation of TNF mRNA to wild-type , MK2-deficient and MK2/3 double-deficient primary BMDM . There are no qualitative differences in the overall polysome profiles of wild type and MK2/MK3 double-deficient BMDM ( Figure 2E ) . However , after 1 h of LPS-stimulation TNF mRNA is detected in a clear polysomal peak for wild type cells only , while in MK2- and MK2/MK3-deficient cells this peak is missing ( Figure 2F ) . As control , distribution of ß-actin mRNA in the polysome profile does not show clear differences between wild type and knockout BMDM ( Figure 2G ) . We postulated that proteins involved in the p38/MK2-dependent regulation of translation should exist in specific fractions of the monosome/polysome profile or the cytosol/ER fractions of LPS-treated macrophages depending on the presence of MK2 . Therefore , we analyzed the relative concentration of various candidate proteins in the different fractions of lysates of LPS-stimulated MK2-rescued or GFP-transduced macrophages ( Figure 3 ) . In the density centrifugation of total cell lysates ( Figure 3A ) , MK2 and p38 - as freely diffusible small proteins - are mainly present in the ribosome-free fraction , which contains also mRNPs . The small ribosomal protein S6 , which was used to monitor ribosome distribution , is present in monosomal and polysomal fractions independent of the presence of MK2 , indicating that there is no general effect of MK2 on translation ( cf . also Figure 2C , 2D ) . Transcript regions free of bound ribosomes are cleaved by RNaseA treatment resulting in destruction of polyribosomes ( Figure 3B ) and in S6 being shifted to monosomes ( Figure 3A ) . The MK2 substrate NOGO-B is present in the ribosome-free mRNP , the monosomal and polysomal fractions , indicating that monosomes and polysomes may dock to the ER . A double band for NOGO-B , characteristic of the slower migrating phosphorylated and faster migrating non-phosphorylated isoform , is only seen in MK2-rescued cells in all fractions indicating that NOGO-B is a specific substrate for MK2 also in macrophages ( cf . Figure 1A ) and that phosphorylation by MK2 does not change its overall distribution in the gradient . Since remaining ER structures cannot be degraded by RNase A , NOGO-B distribution is not completely changed after RNase treatment and seems not to be a ribosome-associated protein . The distribution of various mRNA-binding proteins ( TIA-1 , KSRP , TTP , HuR , and Ago2 ) was analyzed . RNA binding of TTP , HuR and Ago2 is necessary for their distribution in the gradient , since RNase A pre-treatment leads to almost complete disappearance of these proteins from the gradient ( Figure 3A ) . As for NOGO-B , the lack of phosphorylation of TTP by MK2 in the GFP-transduced macrophages does not significantly change the overall distribution of TTP in the gradient . Remarkably , only HuR and Ago2 show a clear two peak-distribution corresponding to the peaks of monosome and polysome fractions as represented by the S6 distribution . This observation strengthens the notion that these proteins may control ribosomal translation . There is no significant MK2-dependent redistribution detected for the proteins analyzed . This is not unexpected , since TNF mRNA and other ARE-containing mRNAs are only a minor part of the total cellular mRNAs analyzed in this overall fractionation . We also compared the distribution of selected proteins between the cytosolic and ER fractions in dependence on the presence of MK2 . There are no qualitative differences in the overall polysome profiles of MK2-rescued and GFP-transduced cells or of the ER or cytosol fractions detected after 1 h LPS-stimulation as measured by absorbance at 254 nm ( Figure 3C ) . However , ß-actin is mainly present in the cytosolic fraction , while the major portion of NOGO-B is detected in the ER fraction ( Figure 3D ) . The ribosomal protein S6 is present in both fractions representing cytosolic and ER-directed translation . The nuclear histone protein H3 is detectable in the total extract but not in the cytosolic or ER fractions , indicating that the fractionation indeed excludes nuclei . The distribution of ß-actin , NOGO-B , S6 and H3 is independent of the presence of MK2 . p38 MAPK , MK2 , and another protein kinase downstream to p38 , Msk1 [39] , are detected exclusively in the cytosolic fraction . Only in MK2-rescued macrophages , NOGO-B displays the double band for the phosphorylated and non-phosphorylated isoforms in the ER-fraction , indicating that NOGO-B can serve as a substrate for MK2 although both proteins are detected in different fractions . Most interestingly , TTP is detectable in the cytosol and ER of GFP-transduced cells , but shows a clear exclusion from the ER fraction in MK2-rescued macrophages , although expressed to a higher total concentration in these cells . Such complete exclusion is not seen for the other mRNA-binding proteins tested ( Ago2 , KSRP , TIA-1 , and HuR ) . The MK2-dependent exclusion of TTP from the ER fraction containing ER-bound polysomes , also loaded with TNF mRNA ( cf . Figure 2B ) , indicates a possible inhibitory role of TTP in translational regulation , which can be neutralized by MK2 . In accordance with its synthesis as a type II-membrane protein , pro-TNF is almost exclusively detected in the ER franction and its level is significantly reduced in the GFP-transduced cells ( Figure 3D , cf . Figure 1C ) The MK2-dependent absence of TTP in the ER fraction , where TNF mRNA is actively translated , and the fact that TTP is a substrate for MK2 [11] at least indirectly suggest an MK2-dependent role of TTP in repression of TNF translation . This notion is also in agreement with the observation that genetic deletion of TTP increases TNF production and renders it MK2-independent [30] . To prove this hypothesis , we performed TTP knockdown ( Figure 4A ) and analyzed ER-directed pro-TNF translation . As seen in Figure 4B , TTP knockdown does not influence translation of ß-actin mRNA . Interestingly , TTP knockdown specifically stimulates a significant shift of TNF mRNA into ER-bound polysome fractions in LPS-treated GFP-transduced macrophages ( Figure 4C ) . Since in LPS-treated MK2-rescued macrophages TNF mRNA is already detected in the polysomal fractions , the amount of polysomal TNF mRNA is only slightly increased further by the knockdown of TTP ( Figure 4D ) . These results clearly identify TTP as a specific repressor of TNF translation , which acts downstream of MK2 . Since the distribution of HuR in density centrifugation parallels ribosomal distribution ( Figure 3A ) and since it is known that HuR also binds to the ARE of native TNF-mRNA [23] , we analyzed the effect of knockdown of HuR on the translation in MK2-rescued macrophages . siHuR knockdown is efficient and does not significantly influence translation of ß-actin mRNA ( Figure 4E , 4F ) . Remarkably , upon HuR knockdown TNF mRNA is excluded from polysomal ER-bound fractions ( Figure 4G ) , reduced in monosomal cytosolic fractions and increased in the mRNP fractions ( Figure 4H ) . This indicates that HuR is necessary for the translation and specifically involved in translational initiation of TNF mRNA . We then asked whether the de-repression of translation seen upon TTP knockdown also depends on the presence of HuR . We performed double siRNA knockdown of TTP and HuR in LPS-treated GFP-transduced macrophages ( Figure 4I ) . Knockdown is efficient for both proteins . As for the TTP knockdown alone , the TTP/HuR double knockdown does not influence translation of ß-actin mRNA ( Figure 4J ) . In contrast , the polysome localization of TNF mRNA induced by TTP knockdown ( cf . Figure 4C ) is not observed in the TTP/HuR double knockdown ( Figure 4K , 4L ) indicating that HuR is necessary for both types of de-repression , whether caused by the reduction or phosphorylation of TTP . Hence , in contrast to TIA-1 , which acts as constitutive repressor [40] , HuR is a constitutive activator of TNF translation in these cells . Similar to HuR , Ago2 also parallels ribosomal distribution ( Figure 3A ) . Ago2 has been described to be essential for TTP- and miR16-dependent TNF mRNA degradation [41] and for translational regulation of TNF mRNA upon serum starvation [42] , [43] . Furthermore , Ago2 is also a substrate for MK2 [44] . Therefore , we analyzed the role of Ago2 in MK2-dependent translational control of TNF mRNA by knockdown experiment . Although Ago2 itself is a central element of the mechanism of siRNA-mediated regulation of specific mRNAs , its knockdown is quite efficient in MK2-rescued and GFP-transfected cells ( Figure S6 ) . However , knockdown of Ago2 does not significantly alter TNF mRNA distribution in MK2-rescued or GFP-transduced macrophages ( Figure S6 , lower panels ) indicating that Ago2 is not regulating LPS-induced translation of TNF mRNA under these conditions . We asked whether the changes in the presence of TNF mRNA in the polysomal fractions seen for the down-regulation of the different components in MK2-rescued and GFP-transduced macrophages are also reflected by changes in pro-TNF biosynthesis of these cells . For this reason , pro-TNF levels before and 4 h after LPS stimulation were semi-quantitatively determined by Western blot analysis . For two independent experiments Western blot signals were quantified and normalized by the ß-actin signal ( Figure 4M , 4N ) . Before LPS stimulation almost no pro-TNF signal was detectable ( Figure 4M ) , indicating that the pro-TNF detected after LPS-treatment represents newly synthesized protein . After LPS-stimulation the amount of newly synthesized pro-TNF is significantly higher in MK2-rescued compared to the GFP-transduced cells as also seen in Figure 1C and Figure 3D . For MK2-rescued cells the knockdown of TTP , which caused a slight increase in the TNF mRNA polysomal fraction ( Figure 4A ) , also lead to a further small increase in pro-TNF synthesis ( Figure 4N ) . Knockdown of TTP in the GFP-transduced macrophages resulted in a strong increase in pro-TNF level corresponding to the strong shift of TNF mRNA to the polysomal fraction ( cf . Figure 4C ) . Consistently , knockdown of HuR significantly inhibited pro-TNF synthesis in MK2-rescued cells reflecting the decrease in the polysomal amount of TNF mRNA detected under this condition ( cf . Figure 4G ) . Most importantly , parallel knockdown of TTP and HuR resulted in a strongly decreased pro-TNF synthesis compared to the single TTP knockdown reflecting the differences in polysomal TNF mRNA seen in Figure 4L . Taken together , the strict correlation between the abundance of TNF mRNA in the polysomal fraction and the production of pro-TNF after LPS treatment confirms the notions that the TNF-mRNA detected in the polysomal fractions is translationally active and that translation of pro-TNF is regulated by TTP and HuR . Since TTP and HuR are both able to bind to the ARE of TNF-mRNA [23] , [29] and since HuR is necessary for the release of the translation block after phosphorylation or knockdown of TTP ( Figure 4L ) , we were interested in whether both proteins compete in ARE-binding and whether this competition is controlled by phosphorylation of TTP by MK2/3 . To analyze this scenario in vitro , we first expressed strep-tagged HuR and TTP in HEK293T cells and purified these proteins by binding to strep-tactin beads . We subjected different amounts of purified strep-HuR and -TTP to Western blotting and detected the proteins using enzyme-coupled strep-tactin . While strep-HuR is detected in a single band of approximately 40 kDa , strep-TTP is seen in a more diffuse group of bands around 50 kDa , which probably arises from multiple PTMs [45] introduced into TTP also in the HEK 293T cells ( Figure S7A ) . To prove ARE-binding activity of the purified proteins , we performed EMSA using a TNF mRNA ARE-derived RNA probe containing six overlapping AUUUA pentamers labeled with the infrared dye DY-681 , which can be detected by imaging ( Figure 5A ) . In EMSA , HuR displays the typical two band pattern [23] while TTP retards the ARE probe in one band , as already known [46] . Specific antibodies against HuR and TTP lead to supershifts respectively demonstrating the identity and specificity of the complexes ( Figure 5A ) . In an in vitro kinase assay , TTP but not HuR is further phosphorylated by MK2 in the presence of its activator p38 ( Figure 5B ) . The observation that strep-HuR is not phosphorylated by p38 in this experiment is in contrast to the published finding that GST-HuR is phosphorylated by p38 in vitro [47] and indicates that the same site is already phosphorylated by Chk2 during overexpression in the HEK293 cells as already described for HeLa cells [48] . We then incubated the purified proteins and mixtures thereof with the DY-681-labeled ARE mRNA , stabilized protein-RNA-complexes by UV cross-linking and visualized ARE-probe bound to protein after SDS-PAGE by infra-red imaging ( Figure 5C ) . Although cross-linking of the ARE-probe to HuR seems more efficient , binding of the probe to both proteins can be detected . In the mixture both proteins show competitive binding to the probe with similar binding in comparable concentrations and a significant displacement of the competitor at about threefold molar excess of the other protein ( Figure 5C ) . Competitive binding of TTP and HuR to the ARE was then analyzed in the presence of active MK2 ( Figure 5D ) . Remarkably , p38/MK2 activity strongly and robustly shifts the binding equilibrium towards HuR in this assay . Even in the presence of an about four-fold molar excess of TTP ( 100 ng TTP vs . 25 ng HuR ) , almost exclusive binding of HuR to the ARE is detected in the presence of p38 and MK2 . We also tested whether MK2 or p38 alone are able to shift the binding towards HuR ( Figure S7B ) . It became clear that only MK2 together with its activator p38 is able to initiate the TTP-HuR-exchange at the ARE-probe , indicating the necessity of catalytically active MK2 in the assay . We then analyzed the ARE binding affinity of HuR and TTP , respectively , phosphorylated by p38/MK2 activity by EMSA and cross-linking experiments using CIP-dephosphorylated protein as control . By titrating increasing concentrations of these proteins ( HuR: 2 . 5×10−7 M–2×10−6 M ) with the same concentration of the ARE probe ( 7 . 5×10−14 M ) approximate Kd values were determined as described in [49] . In line with being no substrate for MK2 ( Figure 5B ) , no significant change in binding affinity of HuR to ARE ( kd 5 . 0–7 . 5×10−7 ) could be observed in the presence of catalytically active MK2 ( Figure 5E ) . In contrast , catalytic activity of p38/MK2 lead to a clear reduction of the affinity of TTP to ARE represented by a shift in the approximate Kd from 5×10−7 for the non-phosphorylated protein to 6–8×10−6 in the presence of active MK2 ( Figure 5F ) . At a fixed TTP concentration of 4×10−6 M significant changes in ARE-binding depending on p38/MK2 phosphorylation could be demonstrated by EMSA ( Figure 5G ) . The reduction in affinity was dependent on phosphorylation of TTP by MK2 , since the MK2 phosphorylation site mutant TTP-AA bound with the same affinity to the ARE like the CIP-treated TTP ( not shown ) and did not show altered affinity in the presence of p38/MK2 ( Figure S7D ) . Furthermore , the binding equilibrium between HuR and the TTP single or double mutants was not changed in the presence of p38/MK2 ( Figure S7E ) . The fact that phosphorylation by MK2 weakens the affinity of TTP for ARE while HuR affinity remains nearly unchanged qualitatively explains that the competitive binding equilibrium between TTP and HuR to the ARE is shifted towards HuR in the presence of MK2 . To examine whether the MK2-dependent shift of the ARE-binding equilibrium between TTP and HuR is relevant in vivo in LPS-stimulated macrophages , we monitored the binding of endogenous HuR to TNF mRNA by immunoprecipitation of endogenous HuR ( RNA-IP ) . MK2-rescued and GFP-transduced macrophages were stimulated for 1 h with LPS and the IP was carried out using HuR- and , as negative control , GFP-antibodies . TNF mRNA in the IPs was quantified by qRT-PCR . Compared to the GFP-IP , specific accumulation of TNF mRNA in the HuR-IP is detected ( Figure 5H ) . Since it is known that HuR constitutively binds to an U-rich element in the 3′UTR of ß-actin mRNA [50] and since ß-actin mRNA is not induced by LPS , we quantified ß-actin mRNA in the HuR-IP to monitor the efficacy and further specificity of the IP ( Figure 5I ) . When normalizing the HuR-IP by ß-actin mRNA , a significantly increased TNF mRNA/ß-actin mRNA ratio is observed in LPS-treated MK2-rescued macrophages compared to the GFP-transduced cells ( Figure 5J ) . This finding suggests that an MK2-dependent shift of the binding equilibrium between HuR and TTP at the ARE of TNF mRNA might also occur in vivo . We could not obtain complementary data for TTP binding to TNF mRNA , since the efficacy of TTP-IP using the available antibodies is low . In addition and more importantly , the strong difference in the expression level of TTP between MK2-rescued and GFP-transfected cells ( Figure 1A , Figure 3D ) leads to large variations in IP-efficiency which make a comparison between these cell lines impossible . Since the mechanism elucidated regulates initiation of translation before docking to the ER , we were interested in whether synthesis of cytoplasmic proteins is regulated in the same manner . TTP itself is a cytoplasmic protein and it is known that TTP also binds to its own mRNA , which contains a 3′ UTR ARE of three clustered AUUUA pentamers [51] . Therefore , we analyzed translation of TTP mRNA in the cytosolic fraction for its dependence on MK2 catalytic activity and on the presence of the TTP antagonist HuR ( Figure 6 ) . The content of TTP mRNA in polysomal fractions is significantly increased in MK2-rescued macrophages when compared to GFP-transduced cells ( Figure 6A ) . The p38 inhibitor blocks this increase and the catalytically dead mutant of MK2 is not able to increase TTP translation ( Figure 6B ) . Both treatments increase the concentration of TTP mRNA in monosomes . These findings indicate that catalytic activity of MK2 also stimulates translation of TTP mRNA in the cytosol . Interestingly , knockdown of HuR strongly inhibits MK2-mediated translational stimulation of TTP mRNA and leads to the accumulation of TTP mRNA in mRNPs ( Figure 6C ) . This inhibition is accompanied by a decrease in TTP biosynthesis as detected by Western blot 4 h after LPS-stimulation ( Figure 6D ) . These findings strongly support the notion that translation of TTP mRNA is regulated by a mechanism similar to that described for TNF mRNA .
We provide evidence that translational control of TNF-mRNA in macrophages might be achieved by a phosphorylation-regulated competitive mRNA association of the two ABPs TTP and HuR . In this scenario TTP acts as a translational repressor while HuR is a translational activator . By LPS-induced MK2-mediated phosphorylation of TTP the competitive binding equilibrium between TTP and HuR is shifted towards HuR and leads to a stimulation of translation ( Figure 7 ) . This mechanism is not restricted to the translation of the type II membrane protein pro-TNF , but is also valid for the translation of the cytosolic protein TTP itself , indicating that this regulation by ABP exchange occurs in an early common step of both processes , probably during translational initiation . This notion is strengthened by the observation that knockdown of HuR leads to an accumulation of cytosolic TNF- and TTP-mRNPs ( Figure 4H and Figure 6 ) and to a reduction of TNF mRNA-containing monosomes ( Figure 4H ) . In regard to the possible molecular mechanisms of translational initiation ( reviewed in [52] ) , the binding of TTP to the ARE in the 3′ region of the mRNAs probably interferes with eIF4G-dependent mRNA circulation or with the binding of the 43S pre-initiation complex ( PIC ) to the mRNA ( Figure 7 ) . Since no significant accumulation of mRNAs in the 43S-fraction is observed , inhibition of 43S PIC scanning or of joining of the 60S ribosomal subunit seems unlikely . Furthermore , it is not clear whether TTP directly interacts with components of the general translation machinery or whether it recruits further proteins which interfere with translational initiation . From the analysis of TTP's mRNA destabilizing function it is known that TTP can act as a binding platform for various proteins involved in mRNA decay [53] . Furthermore , an mRNA-dependent interaction between TTP and poly ( A ) -binding proteins ( PABPs ) has been detected [19] , [53] , [54] and , interestingly , a direct interaction of TTP and PABP C1 was recently demonstrated in a Y2H screen [55] . Hence , one may speculate that TTP-PABP-interaction could interfere with PABP-eIF4G-interaction and could prevent PABP-eIF4G-eIF4E-mediated circularization of the mRNA as a prerequisite for translation ( Figure 7 ) . Taking into account that PABP-1 is also a direct substrate of MK2 [14] , its phosphorylation by MK2 could further contribute to the weakening of the TTP-PABP-interaction . The molecular mechanism of the phosphorylation-driven change in the competitive binding equilibrium of HuR and TTP at the ARE is probably based on the fact that the affinity of TTP is reduced after phosphorylation . However , one should also take into account that cytoplasmic concentrations of HuR and , especially , of the immediate early gene TTP [31] are highly flexible and contribute to the binding equilibrium . A possible scenario of LPS-stimulated regulation could be that newly synthesized TTP is prevented from ARE binding by early phosphorylation via MK2/3 allowing efficient translation of the target mRNAs by the help of HuR . Subsequent decrease in MK2-activity , which peaks after a mere 30 min period , and dephosphorylation of TTP by protein phosphatase 2A [56] could then lead to increased binding of TTP to the ARE , resulting in feedback regulation by translational arrest and destabilization of the target mRNA . Upon LPS-stimulation of macrophages , RNA-IP experiments indicate an early MK2-dependent shift of the binding equilibrium towards HuR supporting the idea that the regulatory mechanism proposed might be relevant in vivo . Recently , a translational repression by TTP has been demonstrated using reporter constructs carrying an ARE in the 3′ UTR in transfected 293T HEK cells [57] . In this system , TTP cooperates with the general translational repressor RCK/P54 , which belongs to the DEAD-box helicase family and displays ATP-dependent RNA-unwinding activity [58] . Hence , it cannot be excluded that RCK/p54 also contributes to TTP-dependent translational repression of TNF mRNA in macrophages . The ubiquitin E3 ligase cullin 4B is a recently identified TTP interacting protein which slightly changes polysome loading of TNF mRNA [59] . However , it is also possible that cullin 4B is involved in ubiquitination and subsequent degradation or functional modification of TTP , as already known for its MEKK1-induced TRAF-2-mediated K63 ubiquitinylation [60] . It has also been described that TTP-facilitated binding of miRNAs , such as miR16 or miR369-3 , are involved in regulation of the decay of ARE-containing mRNAs [41] and stimulation of translation of TNF mRNA upon serum starvation [42] , respectively . A recent screen for miRNAs , which target the 3′ UTR of TNF mRNA , revealed miR-125b and miR-939 as further candidates . However , repression of these miRNAs did not influence TNF mRNA expression , making it unlikely that these miRNAs regulate translation of TNF mRNA [61] . Furthermore , knockdown of Ago2 , a target of MK2 and an essential component of the miRNA system , does not influence LPS-induced TNF translation in macrophages ( Figure S3 ) . This makes it rather unikely that miRNAs regulate MK2-dependent , LPS-induced translation of TNF mRNA in macrophages . It is known that TTP destabilizes ARE-containing cytokine mRNAs of IL-1ß , IL-2 , IL-3 , IL-6 , IL-10 , TNF , GM-CSF and that , in many cases , this destabilization is relieved by the p38 pathway ( reviewed in [16] ) . Even for the global LPS-stimulated regulation of mRNA decay by TTP it has been convincingly shown that the activity of the p38 pathway inversely correlates with TTP's mRNA degrading activity which arises about 3–9 h after LPS-treatment when p38/MK2/3 activity is back to basal levels [62] . The p38/MK2/3-dependent replacement of TTP by HuR at ARE-containing mRNAs would provide a simple and elegant explanation for this regulation: The binding of TTP to a specific mRNA is the prerequisite for its constitutive degradation . When TTP is displaced from the ARE by HuR ( or other ABPs ) , the specific mRNA is no longer targeted to the constitutive degradation pathway and stabilized . It is interesting to note that the stability of KC mRNA is also regulated by TTP [63] , even in a MK2-dependent manner [32] ( Figure S8A , S8B ) , while KC mRNA is not translationally regulated by MK2 ( Figure 2D ) and we do not observe disappearance of the majority of KC mRNA from the polysome fraction upon HuR knockdown ( Figure S8C ) . KC mRNA contains three isolated AUUUA motifs and two doubles of AUUUAUUUA , which are sufficient for TTP binding and regulation of stability [64] . Interestingly , the isolated AUUUA motifs of KC mRNA are not sufficient for HuR binding . Furthermore , although at least its AUUUAUUUA stretch fits rather well to the HuR consensus UUUUUUU , there is no HuR binding reported for KC mRNA in the transcriptome wide screens [65] , [66] . Hence , KC is an example of separation of TTP function from HuR action , since HuR is not able to compete with TTP binding to KC mRNA and is not necessary for its translation . TTP has been postulated as the essential component of the feedback regulation of the LPS-stimulated TNF-response [29] . Increased TTP phosphorylation by MK2 , which neutralizes TTP repressor function [15] , [30] , is paralleled by transcriptional activation of the TTP gene , which belongs to the group of immediate early genes ( Figure 1A and [67] ) , as a result of phosphorylation of the transcription factor SRF by MK2 [31] . Interestingly , the feedback regulation by TTP also comprises TTP's binding to an ARE in the 3′UTR of its own mRNA [51] . Here , we have demonstrated that translation of TTP mRNA is also stimulated by MK2 , probably by the same mechanism of ARE-replacement of phospho-TTP by HuR at the ARE of TTP mRNA , since knockdown of HuR inhibits translation and protein expression of TTP ( Figure 6 ) . Hence , MK2 not only stimulates transcription but also translation of TTP and rapidly enables the re-synthesis of non-phospho-TTP , which again limits TNF expression and TTP expression itself . The parallel translational stimulation of TNF and TTP by the same MK2-dependent mechanism for the first time explains the paradox observation that both , TNF biosynthesis and TTP expression are strongly reduced in MK2-deficient macrophages [32] , whereas complete deletion of TTP leads to increased TNF biosynthesis [29] . In addition , since both TNF and TTP are further reduced in MK2/3 double-deficient macrophages when compared to MK2-deficient cells , it is highly likely that MK3 shares these functions of MK2 . The fact that ABPs interact with their cognate mRNAs and regulate their own expression has been comprehensively described [68] . Furthermore , a complex network of post-transcriptional cross-regulation of expression between ABPs such as HuR , TIA-1 , KSRP , AUF1 is known to exist [68] . However , in the experimental system of immortalized macrophages applied in this study , we could not detect significant changes in expression of other ABPs ( TIA-1 , KSRP ) as a result of knockdown of TTP and HuR . Furthermore , the binary in vitro-binding system using purified HuR and TTP proteins , which is sufficient to observe the MK2-dependent exchange between TTP by HuR , suggests that the regulatory mechanism postulated could function without the involvement of other ABPs apart from TTP and HuR . On the other hand , specific binding of isoforms of the mRNA-binding protein AUF1 to TTP has been described [69] . Interestingly , this interaction increases RNA-binding affinity of TTP in an in vitro assay about five-fold . Hence , it cannot be excluded that further ABPs modulate the proposed regulation in vivo . HuR is predominantly held responsible for constitutive stabilization of ARE-containing mRNAs [20] and demonstrably binds to the ARE of TNF mRNA [22] , [23] . Besides the TNF mRNA stabilizing function , HuR has also been described to influence translation . In macrophages , HuR acts as a homeostatic coordinator of expression of ARE-containing mRNAs [26] , which , when deleted , also increases inflammatory cytokine production of macrophages . This effect could be explained by the fact that HuR is essential for efficient expression of TTP ( Figure 6 ) necessary for down-regulation of the inflammatory response . On the other hand , LPS-stimulated TNF production is significantly inhibited in the NZW mouse strain containing two different 3-base insertions in the TNF-ARE , which inhibit binding of HuR to the ARE [70] . This finding indicates a positive role of the specific binding of HuR to the TNF-ARE for TNF expression and is in agreement with our observations that expression of HuR is necessary for efficient translation of TNF . It is interesting to note that stress-induced release of miR-122-mediated translational repression of CAT-1 mRNA also requires HuR [71] indicating a more general role of HuR in counteracting translational repression , not only by ABPs but also by miRNAs . HuR is mainly localized in the nucleus but continuously shuttles between cytoplasm and nucleus [72] . There are various post-translational modifications of HuR described , which result in changes in its subcellular distribution [73] . Interestingly , p38 phosphorylates HuR at T118 in a stress-dependent manner resulting in significant cytoplasmic accumulation of HuR [47] . Expression of constitutively active MK2 also leads to cytoplasmic accumulation of endogenous HuR [74] . This p38/MK2/3-stimulated increase in the cytosolic HuR concentration could be a prerequisite for translational stimulation of ARE-containing mRNAs by HuR ( cf . Figure 7 ) . Hence , the p38/MK2 pathway may regulate specific translational initiation by catalytic activity of both , p38 and MK2 , in parallel at the levels of translocation and regulation of activity of ABPs .
Primary MK2/3 DKO BMDMs were immortalized as previously described [75] . Retroviral transduction of MK2/3 immortalized BMDMs with vectors encoding MK2 wild type , kinase dead MK2 ( K76R ) and the empty vector ( GFP ) was carried out as described [31] , [75] . Immortalized and retroviral transduced MK2/3 DKO BMDMs were grown in DMEM containing 10% FCS , 2 mM L-glutamine , 100 U penicillin G/ml , 100 mg streptomycin/ml and 0 , 1 mM non essential amino acids mixture ( Life Technologies/Invitrogen ) under humidified conditions with 5% CO2 at 37°C . HEK293T cells were grown under the same conditions and were transfected with the calciumphosphate method . The p38 inhibitor SB202190 ( Sigma ) and LPS ( Escherichia coli 0127:B8 , Sigma ) were used at concentrations of 5 µM and 1 µg/ml , respectively . Primary BMDM were derived as previously described by using 10 ng/ml M-CSF ( Wyeth/Pfizer ) [30] . siRNA-mediated knockdown in BMDMs was performed following the instructions of the protocol for knockdown in RAW264 . 7 cells using HiPerFect transfection reagent ( Qiagen ) . 8×104 cells were seeded in 100 µl growth medium in a 24-well plate . Appoximately 187 . 5 ng siRNA was mixed with 3 µl HiPerFect and 100 µl Opti-MEM ( Life Technologies/Invitrogen ) and incubated for 5 minutes at room temperature . The mixture was added dropwise to the wells and after 6 hrs of incubation 400 µl complete growth medium was added . The next day the medium was changed to complete medium . The highest efficiency for the different knockdown experiments was achieved after 48 hrs of siRNA treatment . For siRNA transfections of 10 cm cell culture plates the protocol was upscaled according to the instructions of the HiPerFect handbook ( Qiagen ) . The following mouse specific siRNAs ( target sequences ) were used ( Qiagen ) : siTTP: 5′-CCTGAGAATCCTGGTGCTCAA-3′ ( Mm_Zfp36_6 ) , siHuR: 5′-CAGAAACATTTGAGCATTGTA-3′ ( Mm_Elavl1_4 ) , siAgo2: 5′-CACTATGAATTGGACATCAAA-3′ . For control knockdown Allstars negative Control siRNA ( Qiagen 1027281 ) was used . Western blotting was performed as described [31] . Blots were developed with an ECL detection kit ( Santa Cruz Biotechnology ) and digital chemoluminescence images were taken by a Luminescent Image Analyzer LAS-3000 ( Fujifilm ) . Primary antibodies used were: anti-eEF2 ( 2332 ) , anti-Histone3 ( 9715 ) , anti-MK2 ( 3042 ) , anti-pMK2pT222 ( 9A7 ) ( 3316 ) , anti-p38 ( 9212 ) , anti-pp38pT180/pY182 ( 9211 ) , anti-pPKD ( 4381B ) and anti-S6 ( 5G10 ) 2217 ( all from Cell Signaling ) , anti-Ago2 ( M01 ) ( Abnova ) , anti-GAPDH ( 6C5 ) Mab374 ( Millipore/Chemicon ) , anti-ß-Actin ( C4 ) sc-47778 , anti-GFP ( B-2 ) sc-9996 , anti-HuR ( 19F12 ) sc-56709 , anti-Msk1 ( H-19 ) sc-9392 , anti-TIA-1 ( C-20 ) sc-1751 , anti-TNF ( L19 ) sc-1351 ( all from Santa Cruz Biotechnologies ) . Antibodies against KSRP [76] , TTP ( SAK21B ) [77] , TTP-pS178 [56] and NOGO-B [35] were described previously and were kindly provided by Drs . A . R . Clark ( London ) , G . Stoecklin ( Heidelberg ) and P . Cohen ( Dundee ) , respectively . Streptactin-HRP conjugate ( IBA BioTAGnology ) and secondary HRP-conjugated antibodies ( Santa Cruz Biotechnologies ) were used . The mouse TNF-alpha Ready-SET-Go ! kit from eBiosciences ( 88–7324 ) was used for TNF ELISA . Quantification of TNF and ß-actin Western blot signals was performed with the Multi Gauge 3 . 2 software ( Fujifilm ) . For each individual experiment two different exposure times were analyzed and then normalized to the ß-actin signals . Two independent Western blot experiments with different sample loading were carried out . For inter-experimental comparison the signals were normalized by the intensity of non-stimulated MK2-rescued cells transfected with control siRNA . BMDMs of a 6 cm plate of 80% confluence were dissolved in 150 µl extraction buffer ( 20 mM Tris pH 8 . 0 , 140 mM KCl , 5 mM MgCl2 , 0 . 5 mM DTT , 0 . 1 mg/ml cycloheximide and 0 . 5 mg/ml Heparin ) . Then 1% ( w/v ) Saponin ( ICN chemicals ) dissolved in DMSO was added to a final concentration of 0 . 1% ( v/v ) followed by 20 minutes incubation on ice with partial vortexing . Cells were then centrifuged 5 minutes at 500×g . The supernatant ( cytsosol ) was kept separately on ice . The pellet ( microsomes and nuclei ) was washed with extraction buffer and was centrifuged again ( 500×g ) . The pellet was dissolved in extraction buffer containing 0 . 5% v/v NP-40 ( Fluka ) . Both cytosol and dissolved microsomal fraction were then centrifuged 10 minutes at 7500×g . The resulting supernatants represent the cytosolic and the microsomal ER fraction free of nuclear components . 1×107 BMDMs were used for a single gradient experiment . The cells were washed twice with 1×PBS containing 0 . 1 mg/ml cycloheximide and differentially lysed as described above . For RNaseA treatment controls , 1 . 5 mg/ml RNaseA ( Carl Roth ) was added to the lysates for 15 min on ice . The cytosolic and ER ribosome extracts ( 400 µl each ) were loaded on linear sucrose gradients ( 12 ml ) ranging from 50% ( bottom ) to 10% ( top ) sucrose containing 140 mM KCl , 20 mM Tris pH 8 , 5 mM MgCl2 , 0 . 1 mg/ml cycloheximide , 0 . 5 mg/ml Heparin and 0 . 5 mM dithiothreitol ( DTT ) . After loading , the extracts were separated by ultracentrifugation in a SW40 . 1 Ti Rotor ( Beckmann-Coulter ) for 2 hrs at 35000 rpm and 4°C . Subsequently , 12 gradient fractions ( each 1 ml ) were collected using a UA-6 UV/VIS device ( Teledyne/ISCO Inc . ) that was connected to an optical unit allowing OD documentation . RNA isolation from the gradient fractions was performed by adding 1/10 volume of 3 M Na-acetate ( Sigma ) and 1 volume of isopropanol and overnight precipitation at −20°C . RNA was pelleted by centrifugation at 13000×g for 20 min at 4°C . For the analysis of co-sedimenting proteins , trichloro acetic acid ( TCA ) was added to each fraction ( final 10% v/v ) . Proteins were precipitated overnight at 4°C and subsequently centrifuged for 20 min at 13000×g at 4°C . Pellets were washed twice with ice cold acetone , dissolved in 100 µl 2×SDS-loading buffer and heated for 5 minutes at 95°C . RNAs were isolated by resolving the cells or pellets obtained from polysome gradient fractions in lysis buffer RA1 of the RNA NucleoSpin II kit and subsequent processing ( Macherey+Nagel ) . For cDNA synthesis , the cDNA first strand cDNA synthesis kit ( Thermo/Fermentas ) was used . cDNAs were diluted 1∶20 for detection in qRT-PCR reactions . For detection of TTP and KC cDNA predesigned and FAM-labelled TaqMan primer mixtures from Applied Biosystems were used ( Mm00457144_m1 Zfp36 , Mm00433859_m1 Cxcl1 ) . TNF cDNA was amplified using a labelled probe ( 5′-FAM – CAC GTC GTA GCA AAC CAC CAA GTG GA – BHQ1-3′ ) together with flanking primers ( forward: 5′-CAT CTT CTC AAA ATT CGA GTG ACA A-3′ and reverse: 5′-TGG GAG TAG ACA AGG TAC AAC CC-3′ ) . ß-actin cDNA was amplified with the VIC-labelled predesigned probe from Applied Biosystems ( 4352341E ) allowing two channel detection of one cDNA . Amplifications were carried out in a 1× SensiFAST Probe No-ROX buffer system ( Bioline ) using a Rotor-Gene-Q device ( Qiagen ) . The threshold cycle ( CT ) of each individual PCR product was calculated by the software of the instrument . 15 cm diameter plates of HEK293T cells were transfected with the expression constructs ( pcDNA3-His-Strep-TTP , -TTP-S52A , -TTP-S178A , -TTP-AA ( S52 , 178A ) and pEXPR-IBA105-HuR ) and lysed 24 hrs post transfection . Strep-tagged HuR and TTP protein was purified by affinity chromatography using streptactin beads ( IBA TAGnologies ) as described previously [76] . Proteins were eluted with desthiobiotin ( IBA TAGnologies ) and the protein concentration was determined by Coomassie staining of the bands in SDS-PAGE compared to BSA standards using the Multi Gauge quantification software 3 . 2 ( Fujifilm ) . Dephosphorylation of purified proteins was achieved by incubation with calf intestinal phosphatase ( CIP ) for 15 minutes at 37°C . Strep-tag purified proteins were dissolved in EMSA-shift buffer ( 20 mM HEPES pH 7 . 6 , 3 mM MgCl2 , 40 mM KCl , 5% Glycerol , 2 mM DTT , 4 µg tRNA ) to give a total volume of 20 µl and were incubated with 75 fmol of an 5′-DY681-labbeled AU-rich RNA-probe ( 5′-AUU UAU UUA UUU AUU UAU UUA UUU A-3′ ) for 25 minutes at 4°C . For supershift experiments 0 . 2 µg of specific antibody was added to the mixture after 10 min of preincubation at 4°C . The reaction mix was then loaded onto a 4% native shift gel after a pre-run of 30 minutes at 80 V and 4°C and separated at 80 V for 90 minutes in 0 . 25× TBE buffer at 4°C . The detection of RNA-protein complexes was performed by visualization of DY681 on a LiCOR Odyssey infrared-scanner . As competition assay , purified HuR protein ( 25–300 ng ) was first incubated together with 75 fmol of the 5′-DY681-labelled AU-rich RNA-probe ( see above ) in a 20 µl reaction mix in 96-well plates . Where indicated , 300 ng of the purified recombinant protein kinases His6-MK2 ( in 1 . 0 µl ) and GST-p38 ( in 0 . 3 µl ) ( Menon et al . 2010 ) and 0 . 5 µl 10 mM ATP were added . After 10 min at 30°C , purified TTP protein was added and incubated for another 15 min minutes at 30°C . Then , the RNA-protein crosslink was performed by UV auto cross-linking using a Stratalinker ( Stratagene ) . The cross-linking products were separated by SDS-PAGE and detected using the LiCOR Odyssey infrared scanner . For detection of phosphorylation , strep-tag purified proteins were mixed together with E . coli-expressed and purified His6-MK2 , GST-p38 , GST protein and radiolabelled gamma-33P-ATP as described previously ( Menon et al . 2010 ) . Purified recombinant Hsp25 served as a positive control for MK2 kinase activity . Samples were resolved by SDS-PAGE and analyzed by phospho-imaging on a FLA-5000 ( Fujifilm ) system . MK2-rescued and GFP-transduced macrophage lines were stimulated with LPS for 1 h , UV treated for 30 seconds ( 120 mJ/m2 ) and subsequently lysed in buffer containing 30 mM HEPES ( pH 7 . 4 ) , 150 mM NaCl and 0 . 5% v/v NP-40 with protease inhibitors . For RNA-IP , 1 µg of anti-HuR ( mouse IgG1 ) and , as negative control , 0 . 5 µg of anti-GFP ( mouse IgG2a ) were incubated overnight at 4°C with the same amounts of cross-linked lysates in a volume of 0 . 5 ml . Afterwards 30 µl of Protein G Sepharose ( GE Healthcare ) suspension blocked with 50 µg/ml t-RNA for 1 h were added . After further incubating 2 h at 4°C , the beads were washed extensively in lysis buffer containing 0 . 25% v/v NP-40 . The associating RNAs were eluted by vigorous vortexing of the beads in 350 µl RA1 lysis buffer ( RNA NucleoSpin II kit ( Macherey+Nagel ) ) . Samples of the total lysates ( input ) and the re-dissolved precipitates were then analyzed by qRT-PCR . | For immediate response and better control of gene expression , eukaryotic cells have developed means to specifically regulate the stability and translation of pre-formed mRNA transcripts . This post-transcriptional regulation of gene expression is realized by a variety of mRNA-binding proteins , which target specific mRNA sequence elements in a signal-dependent manner . Here we describe a molecular switch mechanism where the exchange of two mRNA-binding proteins is regulated by stress and inflammatory signals . This switch operates between stabilization and efficient translation of the target mRNA , when the activator protein of translational initiation binds instead of the phosphorylated destabilizing protein , and translational arrest and degradation of the target , when the non-phosphorylated destabilizing protein replaces the activator . This mechanism is specific to the mRNA of the inflammatory cytokine tumor necrosis factor ( TNF ) -α and the mRNA of its regulator protein TTP and , hence , enables fast inflammatory response and its stringent feedback control . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Experimental",
"Procedures"
] | [
"signal",
"transduction",
"inflammation",
"molecular",
"cell",
"biology",
"immunity",
"protein",
"translation",
"gene",
"expression",
"molecular",
"biology",
"immunology",
"biology",
"mapk",
"signaling",
"cascades",
"rna",
"stability",
"signaling",
"cascades"
] | 2012 | The p38/MK2-Driven Exchange between Tristetraprolin and HuR Regulates AU–Rich Element–Dependent Translation |
Sporulation in the bacterium Bacillus subtilis is a developmental program in which a progenitor cell differentiates into two different cell types , the smaller of which eventually becomes a dormant cell called a spore . The process begins with an asymmetric cell division event , followed by the activation of a transcription factor , σF , specifically in the smaller cell . Here , we show that the structural protein DivIVA localizes to the polar septum during sporulation and is required for asymmetric division and the compartment-specific activation of σF . Both events are known to require a protein called SpoIIE , which also localizes to the polar septum . We show that DivIVA copurifies with SpoIIE and that DivIVA may anchor SpoIIE briefly to the assembling polar septum before SpoIIE is subsequently released into the forespore membrane and recaptured at the polar septum . Finally , using super-resolution microscopy , we demonstrate that DivIVA and SpoIIE ultimately display a biased localization on the side of the polar septum that faces the smaller compartment in which σF is activated .
Asymmetric cell division and differential gene expression are hallmarks that underlie the differentiation of a progenitor cell into two genetically identical , but morphologically dissimilar daughter cells [1]–[5] . The rod shaped Gram-positive bacterium Bacillus subtilis , which normally divides by binary fission to produce two identical daughter cells , undergoes such a differentiation program , termed sporulation , when it senses the imminent onset of starvation conditions ( reviewed in [6]–[8] ) . During sporulation , B . subtilis first divides asymmetrically by elaborating a so-called “polar septum” that produces two unequal-sized daughter cells: a larger “mother cell” and a smaller “forespore” ( Fig . 1A ) that each receive one copy of the genetic material . After asymmetric division , the daughter cells remain attached and a compartment-specific transcription factor called σF is exclusively activated in the forespore . This activation step is critical because it sets off a cascade of transcription factor activation events , each in an alternating compartment , resulting in the expression of a unique set of genes in each daughter cell , which ultimately drives the rest of the sporulation program [9] , [10] . Subsequently , the forespore is engulfed by the mother cell and eventually the forespore achieves a partially dehydrated state of dormancy in which its metabolic activity is largely arrested and is released into the environment when the mother cell ultimately lyses- the released cell is termed a “spore” ( or , formally , an “endospore” ) [11] . Several factors that are required for the switch from medial to asymmetric division have been identified , but the mechanisms underlying this switch remain largely unknown . Similarly , the biochemical basis for the activation of σF has been well elucidated , but the cell biological basis for how this activation is achieved exclusively in the forespore is less well known . At the onset of sporulation , FtsZ , the bacterial tubulin homolog that provides the force for membrane invagination during cytokinesis , initially assembles at mid-cell into a ring-like structure called the “Z-ring” [12]–[14] . At this time , an integral membrane protein called SpoIIE is also produced in the pre-divisional cell and co-localizes with FtsZ via a direct interaction [15]–[17] . Instead of constricting at mid-cell , though , the Z-ring next unravels and extends outward towards each pole via a helix-like intermediate and finally reassembles as two separate Z-rings near the two poles of the bacterium; SpoIIE similarly redeploys to the two polar positions with FtsZ [18] . This redeployment of the Z-ring requires SpoIIE and increased expression of ftsZ from a second sporulation-specific promoter [18]–[23] . Next , one of the two polar Z-rings constricts [24] , [25] , thereby elaborating the polar septum on one end of the bacterium . Although FtsZ constricts at this site and eventually dissipates into the cytosol , SpoIIE somehow remains associated with the polar septum [15] , [16] , [26]–[28] . A recent report demonstrated that SpoIIE is released into the forespore membrane soon after septum formation is complete and that it is recaptured at the polar septum [29] . Interestingly , the total level of SpoIIE before release and after recapture was similar suggesting that SpoIIE is exclusively released into the forespore membrane after septum formation , but the mechanism by which FtsZ could preferentially release SpoIIE into the forespore membrane is not known . After formation of the polar septum SpoIIE performs a second function in which it activates σF [23] , [30] . Prior to asymmetric division , σF is synthesized in the pre-divisional cell , but is held inactive by an anti-sigma factor called SpoIIAB [31] , [32] . After asymmetric septation , SpoIIE , whose C-terminus harbors a phosphatase domain [33] , [34] , dephosphorlyates an anti-anti-sigma factor ( SpoIIAA ) , which then binds and sequesters SpoIIAB , thereby relieving σF inhibition [23] , [35]–[37]- somehow , this activity is manifested only in the forespore compartment . Some evidence has suggested that this compartment exclusivity is ultimately due to a preferential localization of SpoIIE on the forespore side of the polar septum [38] , [39] , but how and when this asymmetric localization initially arises has not been clear . In this study , we examined the subcellular localization of DivIVA , a peripheral membrane protein made of coiled-coil domains that spontaneously assembles into a higher order structure , at the onset of asymmetric division . During vegetative growth of B . subtilis , DivIVA localizes to nascent cell division sites at mid-cell at the very onset of membrane constriction [40] . It has been proposed that negative membrane curvature , such as that which arises on either side of a division septum where it meets the lateral edge of the cell , provides a geometric cue that drives the localization of DivIVA to assemble into ring-like structures on both sides of a division septum [40]–[43] . During normal growth , DivIVA rings serve as platforms that recruit the MinCD complex [40] , [44]–[46] , which inhibits FtsZ assembly , to either side of the nascent cell division septum [47] . As a result , aberrant FtsZ assembly immediately adjacent to a newly forming septum ( and thus , the formation of “minicells” devoid of DNA ) is inhibited and membrane constriction occurs once , and only once , at mid-cell [40] , [47] . At the onset of sporulation , DivIVA performs a second function: DivIVA rings collapse into patches at the two hemispherical cell poles [40] where it anchors the origins of replication of the two replicated chromosomes ( via a DNA-binding protein called RacA ) [41] , [48]–[50] , thereby assuring that both the forespore and mother cell receive one copy of the chromosome . Here , we report that DivIVA localizes to the polar septum and plays an additional role at the onset of sporulation . Deletion of the divIVA gene or depletion of DivIVA protein after its chromosome anchoring function resulted in a severe asymmetric septation defect due to an inability of cells to redeploy FtsZ and SpoIIE from medial to polar positions . As a result , cells arrested at this stage of sporulation , unlike other division mutations reported to cause an asymmetric division defect , prematurely activated σF in a compartment-unspecific manner . We discovered that DivIVA and SpoIIE exist in a complex with one another in sporulating cells and that , when co-produced in vegetative cells , SpoIIE did not persist at division septa in the absence of DivIVA , consistent with a model in which DivIVA is required to briefly anchor SpoIIE at the polar septum during sporulation once FtsZ begins to constrict and subsequently leaves the septum , before SpoIIE is released into the forespore membrane . Finally , employing super-resolution microscopy , we observed that DivIVA initially localized to both sides of the polar septum at the very onset of membrane invagination , but that it preferentially persisted at the forespore side once septation was completed . In contrast , SpoIIE preferentially localized to the forespore side of the polar septum at the onset of membrane constriction , and maintained this biased localization pattern until the completion of polar septum formation . Eventually , SpoIIE was released into the forespore membrane . We propose that DivIVA performs a previously unappreciated role in asymmetric division and compartment-specific activation of σF during sporulation .
DivIVA was previously reported to localize to medially-placed division septa that form during vegetative growth in Bacillus subtilis , but not to the asymmetrically-placed “polar” septum that forms at the onset of sporulation [50] . This implied that DivIVA is somehow able to detect a feature that is exclusive to vegetative division septa , and is absent at asymmetric septa . DivIVA , however , reportedly localizes to division septa by detecting a geometric localization cue , not a chemical cue such as a pre-localized protein: namely , it is thought to preferentially embed onto negatively curved ( concave ) patches of membrane , such as those that are present where a division septum meets the lateral edge of a rod-shaped cell [41] , [42] . Since negative membrane curvature is a feature that is shared by medially-placed division septa that are formed during vegetative growth and asymmetrically-placed polar septa during sporulation , we hypothesized that , if present during sporulation , DivIVA should indeed also localize to the polar septum in sporulating cells . To verify if DivIVA is present at the time of asymmetric septation , we first observed the total levels of DivIVA protein at various time points in synchronized cultures of sporulating B . subtilis by immunoblotting with antibodies specific to DivIVA . Entry into the sporulation pathway was followed by epifluorescence microscopy and enumerating the number of cells that had elaborated a polar septum . Although not all cells in the culture typically initiate sporulation , by 2 . 5 hours after initiation of the program approximately half of the cells had elaborated a polar septum , yet DivIVA protein levels were largely unchanged , and remained constant for at least the first four hours after induction of sporulation , suggesting that DivIVA persists well into sporulation ( Fig . 1B ) . To examine the subcellular localization of DivIVA in sporulating cells , we first constructed a DivIVA-GFP fusion with a flexible linker that had been previously used to construct a nearly fully functional DivIVA-CFP fusion [44] , produced under the control of its native promoter at the ectopic , non-essential amyE locus . As reported for the CFP fusion , cells harboring DivIVA-GFP as the only copy of DivIVA were of similar size as wild type cells and produced very few minicells , suggesting that the fusion protein is largely functional ( Fig . S1 ) . The localization of DivIVA-GFP to the polar septum was observed in 93% of sporulating cells that we examined ( n = 105; Fig . 1C; the remaining 7% of cells counted displayed either no or weak DivIVA-GFP signal ) . Reconstruction of deconvolved Z-stacks of a cell that was beginning to elaborate a polar septum ( cell #3 , Fig . 1C ) revealed that DivIVA formed a ring-like structure ( Fig . 1D , top row ) , similar to DivIVA ultrastructure found at vegetative septa as reported previously [40] , [43] . Representation of the reconstructed fluorescence signal of DivIVA-GFP and DAPI ( indicating the chromosome ) as a surface revealed two separate populations of DivIVA: one at the extreme poles that fulfills the chromosome anchoring role of DivIVA patches , and a second , ring-like localization of DivIVA at the nascent polar septum through which the chromosome was threaded ( Fig . 1D , bottom row ) . Time lapse epifluorescence microscopy revealed that soon after formation of the polar septum initiated , DivIVA-GFP localized at that site and remained associated with the completed septum ( Fig . S2B ) . Next , we examined if the DivIVA ring remains at the site of septation even after membrane constriction ( similar to what has been observed at vegetative septa ) or if it collapses with the rest of the division machinery as cytokinesis proceeds . We observed the localization of DivIVA-CFP and FtsZ ( the major component of the divisome that drives membrane constriction [51] ) , fused to YFP , in sporulating cells that produced both proteins and had just initiated asymmetric septation . In these cells , FtsZ-YFP localized as a band whose width was less than the width of the entire cell ( Fig . 1E , top row ) , consistent with a pattern of constriction of the division machinery . In the same cells , though , DivIVA-CFP remained as two foci at the site of division that did not overlap with the fluorescence signal from FtsZ-YFP , consistent with the formation of a static ring that did not constrict . Reconstruction of deconvolved Z-stacks revealed that FtsZ-YFP formed a collapsing disk-like structure during constriction of the nascent polar septum , whereas DivIVA-CFP remained as a ring-like structure with an outer diameter that was larger than that of FtsZ-YFP ( 1E , bottom row ) . We conclude that , similar to the situation at vegetative septa , DivIVA localizes to the polar septum at the onset of sporulation and remains , at least initially , at that site even after elaboration of the septum . To study the role that DivIVA may play at the polar septum , we sought to monitor sporulation in cells harboring a deletion of the divIVA gene using fluorescence microscopy . However , ΔdivIVA cells are severely elongated relative to wild type cells during vegetative growth and display a severe sporulation defect , presumably because the morphology of the cells is so different . Moreover , DivIVA is required at the onset of sporulation to anchor the origins of the replicated chromosomes to the two poles via an anchoring protein called RacA , thereby rendering the straightforward cytological analysis of a simple ΔdivIVA mutant during sporulation difficult . We therefore engineered a strain in which DivIVA could be proteolytically degraded after its role during vegetative growth and chromosome anchoring had finished , but before the polar septum was elaborated . To this end , divIVA at its native locus was replaced with divIVA-FLAG fused to an altered ssrA peptide tag named ssrAEc . Additionally , the sspB gene from E . coli , which encodes an adaptor that specifically delivers SsrAEc-tagged proteins to the ClpXP proteolytic machinery , was produced under the control of an inducible promoter from an ectopic site on the chromosome , so that DivIVA-FLAG-SsrAEc could be specifically degraded upon addition of inducer [52] . The dimensions of cells harboring divIVA-FLAG-ssrAEc were similar to that of wild type ( Fig . S1 ) , indicating that this allele of divIVA was functional . Finally , to identify individual cells that had properly entered the sporulation pathway , GFP was produced in this strain under the control of a sporulation-specific promoter ( PspoIIG ) . Two hours after the induction of sporulation , 80% of otherwise wild type cells ( n = 140 ) had both produced GFP ( indicating that they had initiated sporulation ) and had elaborated a polar septum ( Fig . 2A ) ; similar fractions of cells harboring divIVA-ssrAEc as the only allele of divIVA ( 73% , n = 142; Fig . 2B ) or sspB only , in the absence or presence of the inducer IPTG , ( 81% , n = 124; and 82% , n = 189 , respectively; Fig . 2C–D ) both entered the sporulation pathway and elaborated polar septa , indicating that neither the SsrAEc tag on DivIVA , nor the presence of the SspB adaptor affected entry into sporulation or asymmetric division . When expression of sspB was induced 45 minutes after the synchronized induction of sporulation , DivIVA-SsrAEc was largely undetectable in 15 minutes ( Fig . S3; compare t60 between +IPTG and −IPTG ) . Surprisingly , when examined by fluorescence microscopy , only 7% of the cells that had initiated sporulation ( evidenced by production of GFP; n = 147 ) elaborated a polar septum . Interestingly , these cells also displayed a condensed and elongated chromosome architecture as evidenced by DAPI staining that appeared untethered at the poles , similar to that observed in a ΔracA strain ( [48]; Fig . 2F: compare the cell marked with white arrow and gray arrow ) , suggesting that the classically defined “Stage I” of sporulation ( in which chromosomes replicate once and condense ) had been achieved , but that “Stage II” , in which the polar septum is formed , was blocked . In the absence of inducer , only 42% of these cells ( n = 172; Fig . 2E ) elaborated a polar septum ( we suspect due to leaky expression of sspB that led to overall reduced levels of DivIVA-SsrAEc as seen in Fig . S3 ) , suggesting that SspB-mediated degradation of DivIVA-FLAG-SsrAEc blocked asymmetric division . To eliminate the possibility that this observed defect in polar septation was an unrelated consequence of the strategy involving the timed degradation of DivIVA-FLAG-SsrAEc , we repeated the experiment in strains harboring a deletion of divIVA , but whose elongation phenotype was suppressed by the additional deletion of minCD [48] , [50] , [53] . Again , to monitor cells that had entered the sporulation pathway , we introduced a GFP reporter produced under the control of a sporulation promoter . Whereas 90% ( n = 120 ) of otherwise wild type cells and 56% ( n = 114 ) of ΔminCD cells producing GFP formed asymmetric septa ( Fig . 2G–H ) , only 5% ( n = 126 ) of ΔdivIVA cells and 14% ( n = 140 ) of ΔdivIVA ΔminCD cells producing GFP elaborated asymmetric septa ( Fig . 2I–J ) , indicating that the absence of DivIVA resulted in an asymmetric division defect . To test if this defect was due to the inability of RacA to interact with DivIVA , we examined asymmetric septation in cells harboring a deletion in racA . The absence of RacA alone had a modest asymmetric septation defect ( Fig . 2K; 75% septation , n = 100 ) . Deletion of both racA and minCD reduced the frequency of asymmetric septation to 32% ( n = 125; Fig . 2L ) , possibly due to an additive effect of removing both MinCD and RacA function , but still the defect was not as severe as the ΔdivIVA defect . Taken together , we conclude that DivIVA plays a previously unappreciated role in the asymmetric placement of the polar septum at the onset of sporulation and that in the absence of DivIVA , cells are arrested at the classically defined “Stage I” of sporulation in which chromosome condensation occurs , but polar septation is prevented . At the onset of sporulation , FtsZ , the bacterial tubulin homolog that drives membrane constriction during cytokinesis , initially assembles as a ring at mid-cell , but then redeploys towards the two poles and forms two polar-localized rings , at which time only one ring constricts to form the polar septum . This redeployment requires an increase in expression of the ftsZ gene [18] , which is mediated by the activation of a second sporulation-specific promoter ( “P2” ) that is dependent on the Spo0A transcription factor , the master regulator of entry into sporulation [21] . To test if DivIVA affects expression levels of ftsZ , we monitored ftsZ transcription by placing the lacZ gene , which encodes β-galactosidase , under the control of the P2 promoter of ftsZ and measured at different time points the β-galactosidase activity in synchronized sporulating cells that harbored this construct . β-galactosidase activity of the P2 promoter reached its peak between 1 h–1 . 5 h after the induction of sporulation in otherwise wild type cells ( Fig . 3A ) . In cells harboring a deletion of either minCD alone or divIVA and minCD , the profile of β-galactosidase was nearly identical to that of wild type cells , and β-galactosidase activity in ΔdivIVA cells even continued to rise after t1 . 5 , suggesting that the asymmetric septation defect in ΔdivIVA ΔminCD cells is not due to the absence of the transcription burst of ftsZ during sporulation . Next , we checked the steady state levels of FtsZ protein by immunoblotting cell extracts prepared from various strains of B . subtilis . At the time that sporulation was induced steady state FtsZ protein levels relative to σA in the absence of MinCD , DivIVA , or MinCD and DivIVA were similar to that of wild type , and remained similar even 90 min after the induction of sporulation ( Fig . 3B , Fig . S4A ) . Although the burst in transcriptional activity did not evidently result in a sustained increase in steady state levels of FtsZ protein ( measured at a population level at the time points that we tested ) , the data nonetheless indicate that the steady state levels of FtsZ protein were not reduced in the absence of DivIVA , and that the failure of FtsZ rings to redeploy in the absence of DivIVA is not due to a reduction in FtsZ level . To test if the asymmetric septation defect in the absence of DivIVA was due to the failure of FtsZ to redeploy to polar sites or due to the inability of FtsZ rings to constrict after redeployment , we monitored the subcellular localization of the division machinery in sporulating cells . To avoid complications arising from the co-production of natively produced FtsZ and ectopically produced FtsZ-GFP , we examined the localization of an FtsZ-associated protein that promotes FtsZ assembly ( ZapA ) fused to GFP that is frequently used as a proxy for FtsZ localization [54] , [55] . ZapA-GFP robustly re-deploys to the polar septum approximately 60 min after the induction of sporulation ( Fig . S4B ) . To identify cells that had initiated sporulation , we also introduced a cassette that expressed mCherry under the control of an early sporulation-specific promoter ( PspoIIA ) . Additionally , to ensure that ΔdivIVA mutants may not simply be slightly delayed in elaborating polar septa , we observed all cells two hours after the induction of sporulation ( at a time when polar septation was usually completed in wild type cells and engulfment was initiating ) and enumerated the total number of cells harboring either a completed septum or a polar-localized ZapA-GFP . In otherwise wild type cells , two hours after the induction of sporulation , polar septa were elaborated , or were about to be elaborated , as evidenced by ZapA-GFP localization at a polar position ( Fig . 3C , arrow; Fig . S4B ) , in 93% of cells ( n = 100 ) that had produced mCherry . In ΔdivIVA cells , only 5% of sporulating cells produced a polar septum ( Fig . 3D ) , but in ΔminCD cells , approximately one third of cells either elaborated a polar septum or displayed ZapA-GFP at a polar site ( Fig . 3E ) . In ΔdivIVA ΔminCD cells , though , only 5% of cells that initiated sporulation elaborated polar septa ( Fig . 3F ) at this time point . In almost half of these sporulating cells , ZapA-GFP remained at mid-cell ( 18% ) or immediately adjacent to mid-cell ( 31% ) and had not redeployed to a polar site ( Fig . 3F ) . In the remaining 46% , no ZapA-GFP structure was detected , suggesting that the protein was diffusely localized in the cytosol . We conclude that the asymmetric septation defect in the absence of DivIVA is due to the failure of FtsZ to redeploy and assemble into Z-rings at polar sites . Since DivIVA typically forms a platform to recruit other proteins to particular subcellular sites ( MinJ to division septa , or RacA to the cell poles [44] , [45] , [48] , [56] ) , we speculated that DivIVA could anchor a sporulation protein at the polar septum . Given the asymmetric septation defect caused by the absence of DivIVA , we wondered if DivIVA could influence the activity of SpoIIE , a transmembrane protein that is involved first in shifting the division septum from the medial to the polar site at the onset of sporulation; and then in activating the first compartment-specific sigma factor , σF , specifically in the forespore [23] , [30] . A functional SpoIIE-GFP fusion initially localizes as a ring at mid-cell , likely via a direct interaction with FtsZ , and then redeploys to polar sites with FtsZ ( [18]; Fig . 4A; Fig . S1D ) . However , unlike FtsZ , which constricts during septation and eventually dissipates in the cytosol after membrane constriction is finished , SpoIIE remains at the polar septum until septum formation is complete and is then redistributed throughout the forespore membrane to perform its second function in σF activation [29] . However , the mechanism by which it initially persists at this site is not known [15] , [26] , [27] , [29] . In 66% of observed sporulating cells ( n = 161 ) , SpoIIE-GFP persisted roughly uniformly along the entire length of the polar septum . In the remaining 34% , we observed two separate populations of SpoIIE at the polar septum: one near the center of the polar septum and another that remained near the lateral edge of the cell ( Fig . 4A , cell 1 ) . To see if SpoIIE co-constricts with FtsZ at the polar septum , we examined the localization of both proteins in sporulating cells producing FtsZ-mCherry and SpoIIE-GFP . At the onset of asymmetric division , both FtsZ-mCherry and SpoIIE-GFP co-localized as bands whose widths were approximately equal to the width of the cell ( Fig . 4C , top row , arrowhead ) . However , as FtsZ constricted to form the polar septum , FtsZ-mCherry localized as a band whose width was less than the width of the cell , whereas in 91% of these observed cells ( n = 100 ) SpoIIE-GFP remained as a band whose width was approximately equal to the width of the cell ( Fig . 4C , arrow ) . Reconstruction of deconvolved Z-stacks revealed that , before FtsZ constriction initiated , both FtsZ-mCherry and SpoIIE-GFP assembled into ring-like structures ( Fig . 4C , bottom row , arrowhead ) , but upon constriction of FtsZ at the polar septum , FtsZ-mCherry formed a disk-like structure , whereas SpoIIE-GFP remained as a ring-like structure with an outer diameter that was larger than that of FtsZ-mCherry ( Fig . 4C , arrow ) . This localization pattern of SpoIIE was similar to that observed for DivIVA-CFP at the polar septum ( Fig . 1E ) and consistent with a model in which FtsZ constricts , while DivIVA and SpoIIE remain associated ( SpoIIE albeit briefly ) at the lateral edge of the polar septum . To test if SpoIIE and DivIVA interact in vivo , we constructed a cell that produced , under the control of their native promoters , a functional DivIVA with a C-terminally appended FLAG tag ( Fig . S1 ) in addition to the untagged version of DivIVA; as well as SpoIIE-GFP ( also functional in vivo; ( Fig . S1D; [27] ) . We then purified DivIVA-FLAG , using anti-FLAG antibodies , from detergent-solubilized cell extracts and examined various fractions collected during purification by immunoblotting . Although DivIVA localizes to the membrane , its association with the membrane is likely tenuous [57]; accordingly it appeared largely in the soluble fraction of cell extracts even without the inclusion of detergent ( Fig . S5 ) . However , the polytopic membrane protein SpoIIE was initially insoluble , but was effectively solubilized by addition of detergent ( Fig . S5 ) . The results in Fig . 4D indicate that SpoIIE-GFP co-purified with DivIVA-FLAG , as did the native , untagged version of DivIVA . As a negative control , the unrelated protein σA was not retained on the column . To ensure that an unrelated membrane-associated protein did not co-purify with DivIVA-FLAG , we repeated the experiment using a strain that overproduced the membrane-associated protein SpoVM-GFP , and observed that SpoVM-GFP also did not co-purify with DivIVA-FLAG . Furthermore , when the purification was performed with cell extract which did not produce DivIVA-FLAG , neither SpoIIE-GFP nor DivIVA was retained on the column , suggesting that retention of SpoIIE-GFP was specifically dependent on the presence of the FLAG-tagged DivIVA . Finally , we performed the reciprocal pulldown experiment in which we purified functional SpoIIE-FLAG ( Fig . S1D ) and observed that DivIVA , but not σA , co-purified with SpoIIE-FLAG . As a negative control , when the SpoIIE-GFP was purified with anti-FLAG antibodies , SpoIIE-GFP , DivIVA , and σA were not retained on the column , suggesting that the specific co-purification of DivIVA was mediated by SpoIIE-FLAG . We therefore conclude that SpoIIE and DivIVA interact with each other in vivo and that this interaction likely takes place at the polar septum . To test if DivIVA localization at the polar septum is dependent on SpoIIE , we examined the localization of DivIVA-GFP in sporulating cells in the absence of SpoIIE . Even though SpoIIE is involved in shifting the septum formation site to the polar position during sporulation , deletion of spoIIE does not completely abolish asymmetric division and about 30% of the cells elaborate a polar septum [18] . In cells harboring a deletion of spoIIE , DivIVA-GFP localization was unaffected and it continued to localize at the polar septum ( Fig . 4B ) . Taken together , we conclude that DivIVA is in complex with SpoIIE , and that the localization of DivIVA to the polar septum does not depend on SpoIIE . Next , we examined the subcellular localization of SpoIIE-GFP , produced under control of its native promoter , in the presence and absence of DivIVA . In otherwise wild type cells , SpoIIE was found at the polar septum or at a potential site of asymmetric septation in 91% of cells 1 . 5 h after the induction of sporulation ( Fig . 4E ) . ΔdivIVA cells , though , due to the defect in polar septation , displayed SpoIIE-GFP in only 12% of cells ( Fig . 4F ) . Once polar septation was restored by deletion of minCD , nearly half of the cells displayed SpoIIE-GFP at the polar septum ( Fig . 4G ) . However , in ΔdivIVA ΔminCD cells , again due to the defect in polar septation , SpoIIE-GFP was found at the polar septum in only 6% of cells ( Fig . 4H ) . Rather , like the localization of ZapA ( and by extension , FtsZ ) , in a majority of cells SpoIIE-GFP was observed either at mid-cell ( 25% ) or at a site immediately adjacent to mid-cell ( 69% ) . To ensure that the SpoIIE localization defect in the absence of DivIVA was not due to a defect in transcription levels of the spoIIE gene , we placed the lacZ gene under the control of the spoIIE promoter and measured β-galactosidase activity in sporulating cells at different time points . The results in Fig . 4M indicated spoIIE transcription was largely unaffected in cells harboring a deletion in divIVA , minCD , or both , as compared to wild type . Thus , in the absence of DivIVA , SpoIIE failed to redeploy from mid-cell to its customary polar positions . Does the interaction between SpoIIE and DivIVA play a role in transiently sequestering SpoIIE at the polar septum ? Whereas the dependence of DivIVA-GFP localization on SpoIIE was readily measured by deleting the spoIIE gene ( Fig . 4B ) , the converse experiment was not straightforward to perform , since deletion of divIVA resulted in the failure to elaborate polar septa ( Fig . 4H ) . We therefore produced SpoIIE-GFP under the control of an inducible promoter in vegetative cells , in the absence of other sporulation factors , and examined its localization either in the presence or absence of DivIVA . When produced in vegetatively growing wild type cells , SpoIIE localized to division septa and persisted at 82% ( n = 146 ) of mature septa after cytokinesis had completed ( [26]; Fig . 4I ) . It should be noted , however , that we observed the eventual release of SpoIIE from mature septa ( unlike DivIVA ) in wild type cells after approximately 2–3 cell generations , suggesting that an interaction between SpoIIE and DivIVA , even in vegetative cells , may be transient . In cells harboring a divIVA deletion , SpoIIE readily localized to future division sites ( presumably dependent on FtsZ ) , but persisted at only 12% ( n = 122 ) of mature septa ( Fig . 4J ) . To ensure that this reduction in localization was not due to the infrequent septum formation , we suppressed the cell elongation phenotype of the ΔdivIVA strain by introducing a deletion in minCD [48] , [50] , [53] . In ΔminCD ΔdivIVA cells , cells were approximately of wild type length and septa were elaborated much more frequently . In these cells , SpoIIE-GFP still failed to persist at mature septa ( 19% , n = 100; Fig . 4L ) . As a control , the deletion of minCD alone did not abolish the persistence of SpoIIE at division septa ( 43% , n = 100; Fig . 4K ) . Taken together , we conclude that although DivIVA is not required for the initial recruitment of SpoIIE to the future site of cell division ( consistent with a model in which FtsZ initially recruits SpoIIE [16] , [17] ) , the transient persistence of SpoIIE at mature septa after cytokinesis has finished depends on DivIVA . To verify if the absence of DivIVA affects the second function of SpoIIE , which is to activate σF specifically in the forespore , we used a strain in which gfp was under the control of a σF-controlled promoter ( PspoIIQ ) and monitored the forespore-specific production of GFP in the presence or absence of DivIVA . In an otherwise wild type strain , 97% ( n = 131 ) of cells that produced GFP produced it exclusively in the forespore ( Fig . 5A ) . In cells harboring a deletion of divIVA , of the cells that produced GFP , 95% ( n = 108 ) of them displayed uncompartmentalized activation of σF ( Fig . 5B ) . Since ΔdivIVA cells are morphologically so dissimilar to sporulating wild type cells , we also examined σF activation in ΔminCD ΔdivIVA cells . In this strain , only 6% of GFP-producing cells displayed forespore-specific activation of σF ( n = 106 ) , whereas deletion of minCD alone resulted in proper forespore-specific σF activation in 71% ( n = 143 ) of the cells ( Fig . 5C–D ) , suggesting that DivIVA is required for compartment-specific activation of σF . This unspecific activation of σF in the absence of DivIVA was unlike the phenotype seen in the absence of other division factors , such as FtsZ , FtsA , DivIC , and FtsL , in which asymmetric division was impaired , but σF activation was prevented as well [58]–[61] . Analysis of a σF responsive promoter ( PspoIIQ ) fused to lacZ indicated that the total amount of σF activity at a population level was not significantly affected in the absence of DivIVA ( Fig . S6A ) . Thus , the absence of DivIVA instead primarily affected the compartment specificity of σF activation . A previous study had reported that a spoIIEV697A mutant allele caused premature activation of σF , which in turn resulted in an asymmetric septation defect in that strain [62] . Does the uncompartmentalized activation of σF in the absence of DivIVA , then , result in the asymmetric septation defect that we observed in these strains ? If so , then deletion of sigF should correct the asymmetric septation defect in the absence of DivIVA . We therefore introduced a sigF deletion in strains that also harbored a divIVA deletion and monitored asymmetric septation by fluorescence microscopy . Since sigF deletion results in a disporic phenotype in which two polar septa are elaborated , we grouped cells containing either one or two polar septa together in one category as being able to form at least one polar septum . At 1 . 5 hours after the induction of sporulation , wild type cells elaborated a polar septum in 67% of cells ( n = 100; Fig . S6B ) ; in the absence of σF , 51% of the cells were still able to elaborate at least one polar septum at this time point ( n = 100; Fig . S6C ) . In a ΔminCD ΔsigF strain 46% of the cells ( n = 100; Fig . S6D ) were able to form polar septa , a similar fraction as the ΔsigF strain ( minicells were distinguished from forespores by observing the presence of DNA in forespores using DNA stain ) . In contrast , in the absence of DivIVA and σF there were no observable polar septa , and in a ΔminCD ΔdivIVA ΔsigF strain only 7% of the cells displayed asymmetric septation ( Fig . S6E–F; n = 100 ) , suggesting that deletion of ΔsigF was not able to suppress the polar septation defect caused by the absence of DivIVA . We therefore conclude that DivIVA is required for the forespore-specific activation of σF , and that the defect in polar septation observed in the absence of DivIVA is not due to the promiscuous activation of σF in a compartment-unspecific manner . The biochemical basis of how SpoIIE activates σF has been extensively studied [23] , [35]–[37] , but the cell biological mechanism underlying how it exerts this function in a forespore-specific manner has remained largely unclear . Recently , Guberman et al . developed an algorithm which refines diffraction-limited images by interpolating the space between adjoining pixels and concluded that SpoIIE preferentially localizes to the forespore side of mature polar septa . To more directly visualize the localization of SpoIIE and DivIVA both at mature polar septa , and those septa that are just beginning to form , we employed a super-resolution fluorescence microscopy technique called Structured Illumination ( SIM ) , which can potentially increase the resolution of fluorescence microscopy two-fold [63] . Indeed , this increase in resolution has previously allowed us to distinguish the localization of DivIVA-GFP on either side of a ∼80 nm wide division septum in vegetative cells [40] . We first observed the localization pattern of DivIVA-GFP and SpoIIE-GFP using a commercially available SIM setup ( DeltaVision OMX ) . As shown in Fig . 6A , membrane invagination at the onset of asymmetric division could be visualized by the increase in membrane staining near a pole along the lateral edges of the cell . DivIVA-GFP , produced under the control of its native promoter , localized initially on either side of this site of membrane invagination , similar to its reported behavior at nascent vegetative septa [40] . However , shortly after the completion of polar septation , DivIVA-GFP was found preferentially on the forespore side of the septum- a preference that persisted even after the onset of engulfment when the polar septum began to curve ( Fig . 6B–C ) . This localization pattern was unlike the localization displayed by DivIVA-GFP during vegetative division , where DivIVA-GFP localized on both sides of a division septum and persisted on both sides long after septation was completed . The mean fluorescence intensity of polar DivIVA patches in cells harboring an adjacent polar septum ( 8286±3321 , n = 50 ) was similar to that of cells harboring no adjacent septa ( 9046±3072 , n = 50 ) , consistent with a scenario in which no significant net exchange of DivIVA molecules appears to take place from the pole to the polar septum ( compare patches in cell 1 and cell 2 in Fig . 1C ) . We next observed the localization of SpoIIE-GFP under the control of its endogenous promoter in sporulating cells using the same SIM setup . At nascent polar septa , we observed that in many cells SpoIIE-GFP preferentially localized on the face of the polar septum that abutted the forespore ( Fig . 6D ) . After septum formation was completed ( Fig . 6E ) , SpoIIE localized to the forespore side of the septum , and finally , as reported previously , SpoIIE was released from the polar septum and localized uniformly in the membrane surrounding the forespore ( Fig . 6F; [29] ) . We were initially unable to properly quantify a large enough dataset of cells displaying the forespore-preferential localization of SpoIIE due to a laser-induced phototoxicity effect , using several different commercially available SIM setups , which deformed many of the polar septa that we observed ( examples are shown in Fig . S7A ) . Curiously , we had not observed such severe phototoxic effects when examining vegetative septa . We therefore used a different implementation of this super-resolution technique called MSIM [64] that greatly reduced this phototoxic effect . To observe initial SpoIIE-GFP localization in a larger number of cells , we used a previously described engulfment-deficient strain ( ΔspoIIDM ) in which the polar septum remains flat [24] . Using MSIM , we again observed that DivIVA-GFP at nascent asymmetric septa localized initially on both forespore and mother cell sides ( Fig . 6G ) . After polar septation was complete , in 81% of sporulating cells ( n = 22 ) , DivIVA-GFP persisted only on the forespore side of the mature septum ( Fig . 6H ) , whereas in only 14% of cells , DivIVA-GFP localized on both sides of the polar septum . In the case of SpoIIE-GFP , MSIM revealed that SpoIIE preferentially localized on the forespore side in both nascent and mature asymmetric septa ( Fig . 6I–J; Fig . S9 ) - the shift in the GFP channel towards the forespore side was revealed more readily by the linescan graph of normalized fluorescence intensity of both membrane stain and SpoIIE-GFP at the septum . In total , 58% displayed clear forespore side specific SpoIIE-GFP localization ( n = 38 ) , while the remaining 42% displayed ambiguous localization in which the GFP channel largely overlapped that of the membrane stain . Of note , we did not detect any cells in which localization of SpoIIE-GFP was preferentially on the mother cell side of the polar septum . One complication of this analysis , though , was that we were unable to determine if the forespore-proximal localization of SpoIIE-GFP in the cells occurred immediately upon septation or if we were imaging the cells after SpoIIE-GFP was released into the forespore membrane and subsequently recaptured . In order to eliminate from our analysis those cells that had recaptured SpoIIE-GFP , we examined SpoIIE-GFP localization in the absence of SpoIIQ , the protein responsible for the recapture of SpoIIE to the polar septum [29] . Whereas SpoIIE-GFP localized exclusively to the flat polar septum in 81% of engulfment-defective ( ΔspoIIDM ) sporulating cells ( n = 100 ) , in the absence of SpoIIQ , SpoIIE-GFP localized to the polar septum in only 52% of the sporulating cells ( n = 118 ) . In the remaining cells , SpoIIE-GFP was uniformly distributed around the forespore membrane , consistent with the inability of these cells to recapture SpoIIE at the polar septum ( [29] ( Fig . S8 ) ) . To verify on which face of the polar septum SpoIIE localized to initially before release into the forespore membrane , we employed yet another implementation of SIM , termed instant structured illumination microscopy ( ISIM ) . This technique eliminates the need for the digital post-processing required in other SIM implementations , directly providing a ∼1 . 4-fold increase in resolution in the raw images , followed by subsequent deconvolution which then provides the full 2-fold resolution improvement relative to conventional widefield fluorescence imaging , allowing us to collect super-resolution images more rapidly [65] . Similar to the MSIM results , when using ISIM , DivIVA-GFP initially localized to both sides of the polar septum in 67% ( n = 36 ) of engulfment-defective ( ΔspoIIDM ) sporulating cells ( Fig . S10 ) , while the remaining 33% displayed forespore-proximal localization . However , 76% of mature polar septa displayed DivIVA-GFP on the forespore-proximal side , while in the the remaining 24% , DivIVA-GFP was found on both sides ( Fig . S10 ) . Next , we examined SpoIIE-GFP localization using ISIM . SpoIIE-GFP appeared to localize preferentially on the forespore-side of the asymmetric septum ( Fig . 6K–L ) in engulfment-deficient ( ΔspoIIDM ) cells . In cells that additionally harbored a deletion of spoIIQ , in those cells which did not release SpoIIE-GFP into the forespore membrane , we observed that SpoIIE-GFP localized preferentially to the forespore-side in 64% ( n = 45 ) of nascent polar septa and 67% ( n = 69 ) of mature polar septa ( Fig . 6M–O; Fig . S9 ) , consistent with a model in which SpoIIE preferentially localizes to the forespore-proximal face of the polar septum before being released into the forespore membrane . To further ensure that the forespore-biased localization of DivIVA-GFP and SpoIIE-GFP were not due to erroneous image registration by the image processing software , or due to variations in the microscope stage when the positions of emission filters changed , we conducted the experiments with fluorescent beads that fluoresce at both red and green wavelengths and aligned the beads' fluorescence in both channels as an internal control . In the example in Fig . S7B , the arrowhead indicates the location of a bead adjacent to a sporulating cell containing a nascent asymmetric septum . As shown in the overlay , there was no shift between red and green channel for the bead , whereas SpoIIE-GFP in the sporulating cell lying adjacent to the bead was preferentially localized to the forespore-proximal side of the polar septum ( Fig . S7B ) . We conclude that DivIVA initially localizes to either side of nascent polar septa and localizes preferentially to the forespore side once septation is complete . SpoIIE , however , preferentially localizes to the forespore-proximal side of the polar septum from the very onset of membrane invagination and remains associated with the forespore-proximal face of the polar septum until it is released into the membranes surrounding the forespore .
DivIVA of B . subtilis is an extensively studied protein made of coiled-coil domains that resembles eukaryotic tropomyosins [53] , [66] , and has two well known functions . First , during vegetative growth , DivIVA arrives at mid-cell shortly after cytokinesis initiates and recruits the components of the Min system to mid-cell to prevent aberrant septation on either side of the site of membrane constriction [40] , [44] , [45] , [47] . It has been proposed that an increase in negative membrane curvature , which arises on both sides of the division septum as the membrane constricts , drives this localization of DivIVA to mid-cell [41] , [42] . Second , during sporulation , DivIVA localizes to the extreme cell poles and anchors the two origins of replication of the replicated chromosomes to each pole so that the forespore and mother cell each ultimately receive one copy of the chromosome [48]–[50] . In this report , we demonstrate a third role for DivIVA in which it localizes to the polar septum during sporulation and resides in complex with a multifunctional transmembrane protein called SpoIIE . SpoIIE is initially recruited to mid-cell at the onset of sporulation by FtsZ , after which it is required for the efficient redeployment of the Z-ring from mid-cell to polar positions to initiate asymmetric division . After asymmetric division commences , SpoIIE is initially recruited to the polar septum by FtsZ , but a significant population of SpoIIE remains associated with the polar septum even after FtsZ constricts and leaves that location . Shortly thereafter , SpoIIE is released exclusively into the forespore membrane to perform its second function in activating the first forespore-specific transcription factor σF [23] , [30] . Subsequently , SpoIIE is recaptured at the forespore face of the polar septum [29] where it may participate in remodeling the peptidoglycan at the polar septum [22] . In the absence of DivIVA , we found that the two main functions of SpoIIE were disrupted , in that asymmetric septation did not occur , and that σF was prematurely activated in a compartment-unspecific manner . Interestingly , the deletion of several other cell division genes have only been reported to cause an asymmetric cell division defect , but not the premature activation of σF [58]–[61] . We therefore propose that , in addition to SpoIIE , DivIVA is required for the redeployment of FtsZ to the polar division sites and for the compartment-specific activation of σF in the forespore . Our studies began with the observation that DivIVA localizes to the polar septum during sporulation . In an effort to determine the function of DivIVA at this location , we deleted divIVA and suppressed the elongation defect caused by this deletion by introducing a deletion in minCD as well . As previously observed , deletion of minCD alone had a mild defect in asymmetric division [67]–[70] , but the additional deletion of divIVA resulted in the nearly complete abrogation of asymmetric division . That this defect was not primarily dependent on the absence of MinCD was established in an experiment in which we achieved the controlled degradation of DivIVA alone at a time point after it had completed its chromosome anchoring function , but before cells had initiated asymmetric division . Removal of DivIVA alone in this manner similarly abrogated asymmetric division at the onset of sporulation . Ultimately , we observed that the defect in asymmetric septation in the absence of DivIVA was due to the inability of these cells to redeploy FtsZ and SpoIIE from medial to polar positions . Interestingly , although polar septation was not directly measured at the time , Cha and Stewart had hypothesized that DivIVA may play an active role in the asymmetric positioning of the polar septation in the very first report of the divIVA locus in B . subtilis [68] . Additionally , a recent report suggested that a severe sporulation defect in a strain in which Spo0A rapidly accumulated at the onset of sporulation was perhaps due to the premature suppression of divIVA expression , consistent with our hypothesis that DivIVA plays an additional , indispensible role during sporulation [71] . Previously , Ben-Yehuda and Losick had demonstrated that , in vegetatively growing B . subtilis , the slight overproduction of FtsZ and the production of SpoIIE were sufficient to generate polar septa [18] . We propose that , in this context , DivIVA , which is abundant during vegetative growth , is also required for the redeployment of FtsZ to generate polar septa during vegetative growth . It is currently unclear to us how DivIVA mechanistically mediates asymmetric division . At the onset of sporulation , the replicated chromosomes are extensively remodeled so that it may form the “axial filament” structure that extends from one pole to the other in what has traditionally been named “Stage I” of sporulation . Interestingly , in the absence of DivIVA , cells appeared to be arrested at this stage which , to our knowledge , is only the second description of a gene whose deletion results in a so-called “Stage I” arrest , not simply a delay [72] . Proper chromosome segregation has also been implicated in formation of the polar septum , in a manner that depends on the DNA-binding protein Spo0J [73] . Interestingly , DivIVA was shown to interact with Spo0J in a sporulation-specific manner , and independent of the Min system [74] , [75] . Recently , the absence of RefZ , a DNA-binding protein that binds to origin- and terminus-proximal regions of the chromosome was reported to delay asymmetric septation , and it was proposed that RefZ may either facilitate repositioning of FtsZ at a polar position or promote disassembly of FtsZ at mid-cell sites [76] . It is therefore conceivable that the asymmetric septation impairment in the absence of DivIVA is a result of defective chromosome remodeling or segregation . It is also possible that the asymmetric septation defect in the absence of DivIVA may be through a metabolic pathway that affects septum formation . Deletion of citC , which encodes isocitrate dehydrogenase , also resulted in inhibition of asymmetric septation [72] . Loss-of-function mutations in the spoVG gene suppressed this defect , and overproduction of SpoVG in otherwise wild type cells resulted in delayed σF activation [77] . Curiously , SpoVG homologs were recently shown to be site-specific DNA-binding proteins [78] , thereby possibly providing a link between asymmetric septation and chromosome remodeling or segregation . In addition to its less understood role in redeploying the Z-ring from mid-cell to polar sites , SpoIIE is required for the activation of σF , the biochemical basis of which has been well studied [23] , [30]–[32] , [34] . However , the cell biological basis for the forespore-specific activation of σF has been less well understood for two reasons . First , although FtsZ is thought to recruit SpoIIE to the polar septum , it had been unclear how SpoIIE remains associated with the polar septum after FtsZ constricts at that site and ultimately dissipates from that location , or how FtsZ may selectively release SpoIIE into the forespore membrane upon completion of constriction . We have observed a previously unappreciated step in which SpoIIE forms a ring-like structure at the polar septum during FtsZ constriction , before the release of SpoIIE into the forespore membrane . Based on our observation that DivIVA and SpoIIE co-localize at the polar septum and the discovery that DivIVA and SpoIIE copurify with each other , we propose that DivIVA is required for this initial sequestering of SpoIIE at the polar septum ( Fig . 7 ) . Consistent with this model , SpoIIE , when produced in vegetative cells , stably associated with vegetative division septa only in the presence of DivIVA . In sporulating cells , in the absence of DivIVA , polar septa were not efficiently formed and σF was prematurely activated in a compartment-unspecific manner . Second , the mechanism by which SpoIIE exerts its activity specifically in the forespore was also unclear . SpoIIE is released into the forespore membrane and it is promptly recaptured , nearly quantitatively , at the polar septum , suggesting that SpoIIE is exclusively released into the forespore , and not into the mother cell membrane [29] . Another report suggested , based on the interpolation of fluorescence signal in diffraction-limited images , that at mature polar septa SpoIIE was preferentially detected on the forespore side of the polar septum [39] -this localization of SpoIIE was likely after its recapture at the polar septum . In this report , we examined cells that displayed nascent polar septa , as well as cells that had elaborated mature polar septa , using three super-resolution techniques called SIM , MSIM , and ISIM , and observed that the biased localization of SpoIIE on the forespore side could be observed at the very onset of membrane invagination . In contrast , we observed DivIVA initially on either side of nascent polar septa , similar to its localization pattern in vegetative division septa . However , unlike its behavior in vegetative cells where it was ultimately equally distributed on both sides , we observed that at mature polar septa , DivIVA , like SpoIIE , also preferentially localized to the forespore side . The mechanism for this shift in distribution is currently unclear . It may , for example , involve the selective degradation of DivIVA on the mother cell face of the polar septum or may involve a transfer of pole-localized DivIVA molecules to the forespore-proximal side of the polar septum which we were unable to measure using our current experimental setup . Such a transfer has indeed been recently reported for vegetative septa , which may occur in 5–20 mins [79] . Curiously , the presence of Min proteins , which are typically recruited by DivIVA , on the forespore-proximal face of the polar septum has been implicated in maintaining the polarity of DNA translocation [80] . The basis for the early bias in localization of SpoIIE is currently not known , but this pattern highlights the establishment of asymmetry , at a compartment-specific level , in the developing sporangium long before polar septation is completed . We propose that the interaction of SpoIIE with DivIVA allows for its retention at the forespore side of the newly forming polar septum until septation is complete . After completion of polar septum formation , similar to the release of MinCD after completion of vegetative septa [47] , SpoIIE may be liberated to redistribute along the membrane surrounding the forespore . It is tempting to speculate that the initial interaction of SpoIIE with DivIVA may inhibit the phosphatase activity of SpoIIE until it is released into the forespore membrane , which may also explain the premature activation of σF in the predivisional cell in the absence of DivIVA . Curiously , the N-terminus of MinC shares structural similarity with that of SpoIIAA , the protein that is dephosphorylated by SpoIIE [81] . Thus , once released into the forespore membrane , the phosphatase activity of SpoIIE may be uninhibited and σF may be activated . Therefore , asymmetrically placed DivIVA may act as a molecular beacon that signals the completion of septum formation for the subsequent activation of σF . An outstanding question , though , will be to determine how the asymmetry of SpoIIE is established at such an early time point before the mother cell and forespore compartments are even separated .
All strains used in this study are congenic derivatives of B . subtilis PY79 [82] . Genotypes of strains used are provided in Table S1 . B . subtilis competent cells were prepared as described previously [83] . β-galactosidase activity of cell samples collected at time points indicated was measured with modifications as described previously [84] , [85] . To express divIVA-linker-cfp from the amy locus , PdivIVA-divIVA without the stop codon was PCR amplified from PY79 chromosomal DNA using primers that abutted HindIII and NheI sites ( “oP10” 5′-AAAAAGCTTTCGTGTTTTCTGAGACA and “oP11” 5′-AAAGCTAGCTTCCTTTTCCTCAAATAC ) . linker-cfp was PCR amplified from DS4152 chromosomal DNA [44] using primers abutting NheI and BamHI sites ( “oP54” 5′-AAAGCTAGCGGT TCCGCTGGCTCCGCTGCTGGT TCTGGCCTC and “oP55” 5′-AAAGGATCCTTACTTATAAAGTTCGTCCATGCCAAGTGTAATGCC ) . Both fragments ligated into integration vector pDG1662 to create pPE17 . To express divIVA-linker-gfp from the amy locus , pPE17 was digested with HindIII and XhoI to liberate PdivIVA-divIVA and part of the linker sequence . Next , gfp mut2 was PCR amplified from pKC2 [86] using primers that abutted the remaining linker sequences , and 5′ XhoI and 3′ BamHI sites ( “oP62” 5′-AAACTCGAGGGTTCCGGAATGAGTAAAGGAGAAGAACTTTTC and “oP47” 5′-AAAGGATCCTTATTTGTATAGTTCATCCATGCC ) . Both fragments were ligated into pDG1662 to create pKR227 . To replace divIVA at the native locus with divIVA-FLAG-ssrAEc , the final 400 nucleotides of divIVA ( omitting the stop codon ) were PCR amplified using primers that abutted BamHI and XbaI sites ( “DivIVA-C4005′Bam” 5′-AAAGGATCCAGTCAGAAAAGATTACGAAATTG and “DivIVA-C4003′ Xba” 5′ AAATCTAGATTCCTTTTCCTCAAATACAGCG ) , and ligated into Campbell integration vector pKG1268 [52] to create pKR226 . To express divIVA-FLAG from amyE , PdivIVA-divIVA was PCR amplified from PY79 chromosomal DNA using primers that abutted 5′ HindIII and 3′ BamHI sites , as well as a 3′ FLAG tag sequence ( “DivIVAprom5′Hind” 5′-CCCAAGCTTTCGTGTTTTCTGAGACAGCAG and “DivIVA3′FLAGBam” 5′-CGCGGATCCTTACTTGTCGTCATCGTCTTTGTAGTCTTCCTTTTCCTCAAATACAG ) , and ligated into pDG1662 to create pKR200 . To express spoIIE-gfp from amyE under the control of an IPTG-inducible promoter , spoIIE-gfp was PCR amplified from SB201 [87] using primers that abutted SalI and SphI sites ( “oP44” 5′- AAAGTCGACACATAAGGAGGAACTACTATGGAAAAAGCAGAAAGAAGAGTGAACGGG and “oP24” 5′- GCCGCATGCTTATTTGTATAGTTCATCCATGCC ) , and ligated into integration vector pDR111 to create pPE19 . To express IPTG-inducible spoIIE-FLAG from amyE , plasmid pPE48 was created by appending 3×-flag tag sequence ( GATTATAAGGATCATGATGGTGATTATAAGGATCATGATATCGACTACAAAGACGATGACGACAAG ) followed by a stop codon to the 3′ end of the spoIIE coding sequence in pPE19 via QuikChange mutagenesis ( Agilent ) . Strain KR610 ( ΔspoIIE::tet ) was created by the long flanking homology method [88] using primers ( “spoIIEKO-1” 5′-GCAAGTAGCCTTGTTGACAC , “spoIIEKO-2” 5′-CAATTCGCCCTATAGTGAGTCGTTCCTCTCATCTCCCACCTG , “spoIIEKO-3” 5′-CCAGCTTTTGTTCCCTTTAGTGAGCGCTTCCGTATAAATCAAATTTC , and “spoIIEKO-4” 5′-TTTCAAGACATTCACTTCAGAAG ) . For immunoblot analysis , cells were grown as described below , harvested , and cell extracts for immunoblot analysis were prepared by lysozyme treatment as described previously [89] . Extracts were separated by SDS-PAGE and immunoblotted using antisera raised against purified DivIVA-GFP , σA , GFP ( Covance , Inc . ) , or E . coli FtsZ ( courtesy of Sue Wickner; [90] ) as indicated . Where indicated , band intensities were quantified using ImageQuant software ( GE ) . B . subtilis overnight cultures grown at 22°C in casein hydrolysate ( CH ) medium were diluted 1∶20 into fresh CH medium and grown until OD600 reached ∼0 . 5 at 37°C . For induction of sporulation , cells were spun down and resuspended in Sterlini and Mandelstam ( SM ) medium as described previously [91] , for the time indicated . Sporulation efficiency was measured by growing cells in Difco sporulation medium ( DSM; KD Medical ) for at least 24 h at 37°C . The number of heat-resistant colony forming units ( cfu ) was obtained after incubation at 80°C for 10 min . For strains harboring genes that disrupted the thr locus , L-threonine ( 40 µg/ml final concentration ) was added to the culture as a nutritional supplement . Where specified , IPTG ( 1 mM final concentration ) was added at the indicated time to induce genes; xylose ( 0 . 5% final concentration ) , where indicated , was added at the time of resuspension . Cells were visualized as described previously [92] . Briefly , culture pellets ( from 1 mL culture ) that were grown as described above was washed with PBS and resuspened in 50–100 µL PBS containing 1 µg/ml ( final concentration ) of the fluorescent dye FM4-64 or 46 µg/ml ( final concentration ) TMA-DPH to visualize membranes and , where indicated , 2 µg/ml ( final concentration ) of DAPI to visualize DNA . Cells ( 5 µl ) were then placed on a poly-L-lysine coated glass bottom culture dish ( Mattek Corp . ; poly-L-lysine did not appreciably affect localization of DivIVA-GFP at the polar septum ( Fig . S2A ) ) . A pad made of 1% agarose in distilled water ( or SM media containing IPTG and FM4-64 for time lapse microscopy ) was cut to size and placed above the cells . Cells were viewed at room temperature ( or 32°C for timelapse ) with a DeltaVision Core microscope system ( Applied Precision ) equipped with a Photometrics CoolSnap HQ2 camera and an environmental chamber . Seventeen planes were acquired every 200 nm at room temperature and the data were deconvolved using SoftWorx software . Imaris software was used for three dimensional surface rendering of fluorescence data . For 3D-SIM , cells were prepared as described above and imaged using Delta Vision OMX Blaze ( Applied Precision ) . For MSIM , cells ( with fluorescent beads when indicated ) were labeled with FM4-64 as described above and placed on top of a glass slide and a freshly prepared poly-L-lysine coated coverslip ( #1 . 5 thickness , VWR ) was placed on top of the cell suspension; coverslips and slides were cleaned as described previously [93] . Fluorescence images were acquired using a custom MSIM system equipped with a 60× 1 . 45NA oil objective ( Olympus ) and appropriate filters ( Chroma , zt405/488/561 ( dichroic ) LP02-488RE-25 , NF03-561E-25 ( emission filters ) . Total exposure times for each 2D slice were either 1 s or 2 s depending on the signal intensity . 2D slices were acquired every 200 nm along the Z-axis for construction of 3D volumes , comprising a total axial range of 10–12 µm for each sample . The longer wavelength channel ( red ) was collected prior to the shorter wavelength channel ( green ) for all samples . Multicolor 100 nm diameter Tetraspeck microspheres ( Life Technologies , T-7279 ) were immobilized on the coverslip surface to enable precise alignment of each image channel . Alignment of the two channels was completed by translating the red channel relative to the green channel to maximize overlap of the reference microspheres using ImageJ . Images were deconvolved with a 3D Richardson-Lucy algorithm implemented in the python programming language ( 40 iterations , using a Gaussian PSF with x , y , z FWHM values 3 . 8×3 . 8×2 . 4 pixels , code available at code . google . com/p/msim ) [64] . ISIM imaging was performed using a previously described system [65] . Cells were prepared as described above . Fluorescence from labeled membranes was excited at 561 nm and GFP signal from fusion proteins was excited at 488 nm . A 525 nm bandpass filter ( Semrock , FF01-525/50-25 ) was used for GFP collection and a 561 nm notch filter ( Semrock , NF03-561E-25 ) was used for collecting membrane fluorescence . Exposure times for imaging the membrane probe were 40 ms or 80 ms ( depending on sample brightness ) and 400 ms for imaging GFP fusion constructs . 3D stacks were collected with a z step size of 200 nm . Resulting images were deconvolved using Richardson-Lucy deconvolution , and image channels were aligned using the multicolor beads for reference as described above . Line scans were completed using the line scan function in ImageJ , setting the line width to 5-pixels . Line scan measurements were exported to Microsoft Excel for rendering and display . FLAG-tagged proteins were immunoprecipitated from strains coproducing FLAG- and GFP-tagged proteins using the FLAG Immunoprecipitation kit ( Sigma ) . Samples were processed largely as described previously [84] . Briefly , a 20 ml culture of cells was induced to sporulate for 1 . 5 h , harvested , resuspended in 500 µl protoplast buffer ( 0 . 5 M sucrose , 20 mM MgCl2 , 10 mM potassium phosphate [pH 6 . 8] , 0 . 1 mg/ml lysozyme ) , and incubated at 37°C for 15 min to remove cell wall [94] . Protoplasts were harvested and cell extracts were prepared by resuspension in 1 ml of lysis buffer ( 50 mM Tris [pH 7 . 4] , 150 mM NaCl , 1 mM EDTA , 1% Triton X-100 ) . 100 µl of the extract was retained for analysis as the “load” fraction . The remaining cell extract was added to 20 µl of anti-FLAG affinity resin and incubated overnight at 4°C with light shaking . The antibody resin was centrifuged and supernatant was collected as the “unbound” fraction . The resin was then washed extensively with lysis buffer . Proteins associated with the resin were eluted through competitive elution with FLAG peptide-containing lysis buffer . Fractions were analyzed by immunoblotting using specific antisera as described above . | A central feature of developmental programs is the establishment of asymmetry and the production of genetically identical daughter cells that display different cell fates . Sporulation in the bacterium Bacillus subtilis is a simple developmental program in which the cell divides asymmetrically to produce two daughter cells , after which the transcription factor σF is activated specifically in the smaller cell . Here we investigated DivIVA , which localizes to highly negatively curved membranes , and discovered that it localizes at the asymmetric division site . In the absence of DivIVA , cells failed to asymmetrically divide and prematurely activated σF in the predivisional cell , largely unreported phenotypes for any deletion mutant in a sporulation gene . We found that DivIVA copurifies with SpoIIE , a protein that is required for asymmetric division and σF activation , and that both proteins preferentially localize on the side of the septum facing the smaller daughter cell . DivIVA is therefore a previously overlooked structural factor that is required at the onset of sporulation to mediate both asymmetric division and compartment-specific transcription . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"biology",
"biology",
"and",
"life",
"sciences",
"developmental",
"biology",
"microbiology"
] | 2014 | Asymmetric Division and Differential Gene Expression during a Bacterial Developmental Program Requires DivIVA |
We most often consider muscle as a motor generating force in the direction of shortening , but less often consider its roles as a spring or a brake . Here we develop a fully three-dimensional spatially explicit model of muscle to isolate the locations of forces and energies that are difficult to separate experimentally . We show the strain energy in the thick and thin filaments is less than one third the strain energy in attached cross-bridges . This result suggests the cross-bridges act as springs , storing energy within muscle in addition to generating the force which powers muscle . Comparing model estimates of energy consumed to elastic energy stored , we show that the ratio of these two properties changes with sarcomere length . The model predicts storage of a greater fraction of energy at short sarcomere lengths , suggesting a mechanism by which muscle function shifts as force production declines , from motor to spring . Additionally , we investigate the force that muscle produces in the radial or transverse direction , orthogonal to the direction of shortening . We confirm prior experimental estimates that place radial forces on the same order of magnitude as axial forces , although we find that radial forces and axial forces vary differently with changes in sarcomere length .
Strain energy storage in muscle systems is most often associated with stretched tendons or other elastic supporting materials [1] , [2] . In many instances , strain energy storage in skeletal and tendon structures has been shown to be a crucial component of the locomotor systems of animals , especially flying animals [3] . While muscle' role as a force generator has dominated research on animal locomotion , emerging studies posit diverse functional roles for muscles , including those of a brake , actuator , spring , or even a damper ( for a review see Dickinson et al . , 2000 ) [4] . Somewhat less attention has focused on the extent to which muscle itself plays a role in strain energy storage . That work which has been done has focused on the possibility of storing energy in the thick filaments , rigor cross-bridges , or the in the extensible accessory protein titin [5]–[7] . This assumption that active cross-bridges play a minor role is understandable: they generate force in activated muscle and are thought to be constantly cycling between freely diffusing and attached states and so would be expected to develop little deformation . However , recent work suggests that in certain situations the cross-bridges may be locked onto muscle' thin filaments , frozen into a lattice that can act to store energy [8] . Energy storage may be possible in the subset of bound cross-bridges in antagonistic muscles that absorb inertial energy of a periodically moved appendage . This energy storage permits locomotion that would otherwise be energetically unfeasible [3] . Additionally , energy storage in muscle has been proposed in non-cyclical movements such as the tentacular strike of the squid , stomatopods'raptorial appendage strike , or the tongue extension of toads [9]–[12] . In these cases of one-off sudden movements , even a set of cycling cross-bridges may store strain energy for release on the initiation of rapid movement through pre-movement activation and subsequent pre-movement strain of the cross-bridges occurring just before the onset of an explosive motion . Our spatially explicit half-sarcomere model lets us parse how strain energy is partitioned between the filaments and the cross-bridges in maximally activated isometric sarcomeres . We show that the cross-bridges may store the majority of the elastic strain energy . Cross-bridges are more often thought of as force generators than energy storage sites . The force generated by individual myosin heads arises from deformations as they form cross-bridges between the thick and thin filaments and undergo a rotation about a lever arm [13] . Interestingly , generating force by a rotation about a hinge implies that the vector of the generated force will have a component perpendicular to the direction of contraction [14] , [15] . This force component is in the radial direction , orthogonal to the axial force that is generated in parallel to the thick and thin filaments ( Figure 1 ) . Radial force was observed during contraction in intact muscle fiber experiments dating back to the 1950s [16] . Subsequent studies of radial forces placed them on the same order of magnitude as axial force [17]–[22] . These more recent experiments addressed radial force production through a proxy such as changes in fiber diameter or alterations of the muscle' radial compliance . The use of a lever-arm cross-bridge ( with extensional and torsional springs ) in the current spatially explicit model permits direct simulation of radial force production ( Figure 1B&D ) [15] . This cross-bridge model expands upon prior models ( Figure 1A&C ) [23]–[27] . Radial force may have functional implications . The internally generated radial force is a partial determinant of fiber radial compliance [21] , [22] . Alterations in radial fiber compliance are also a hallmark of dystrophic disorders [28] , [29] . Mis-regulation of the transmission of radial force produced during contraction may be a cause of the ultrastructural disorganization observed in histological studies of dystrophic muscle [30] . In addition to the more commonly analyzed axial forces , the model presented here addresses both radial force generation and the strain energy in the filaments and cross-bridges of the contractile lattice . These phenomena are linked , and are results of deformation of cross-bridges in the axial and radial directions . The interdependence of these properties is uniquely addressable using spatially explicit models of muscle contraction with lever-arm myosin geometries ( Figure 1 ) . We have developed such models based on protein structural information [15] . These models permit a fine parsing of energy locations which shows that cross-bridges store substantial elastic strain energy . Correlation of this cross-bridge energy with axial and radial forces suggests that radial cross-bridge strain could supply much of the energy stored in the contractile lattice .
In the fully activated conditions of our simulations , both the axial and radial forces quickly rise to an asymptotic maximum ( Figure 2 ) . This rise to a maximum value takes less than 50 ms . The exponential time constant of the rise to peak force is not significantly different between the axial and radial forces ( p = 0 . 31 ) . After steady force levels are reached , stochastic fluctuations in the number and states of bound cross-bridges show as noise in the force traces . This clean rise gives clear asymptotic maximum forces . The radial force , at all sarcomere lengths , is of the same order of magnitude as the axial force ( Figure 3 ) . At most sarcomere lengths below , where there is significant myofilament overlap , the radial force is larger than the axial force . At very short sarcomere lengths the radial force is as much as 2 . 4 times the axial force , although this is in a region where overall axial force levels are relatively small . These results agree with prior experimental studies which found radial forces of the same magnitude as axial forces [18]–[22] , [33] . The majority of strain energy stored in the complete contractile lattice of filaments and cross-bridges is partitioned in the cross-bridges ( Figure 4A&B ) . The relative distribution of energy between the cross-bridges and the filaments varies with sarcomere length ( Figure 4B ) . Even when the filaments reach their peak energy relative to the cross-bridges , at the sarcomere lengths where maximum force is produced , the cross-bridges still have more than three times the energy of the filaments . At very long and very short sarcomere lengths , where little axial force is produced , the energy stored in the cross-bridges dominates the system . The elastic energy storage may be more finely parsed: into the components located in each of the two springs constituting every cross-bridge and the components in each of the two filament types ( Figure 4C ) . These results show that the energies of the thick and thin filaments vary similarly across all sarcomere lengths . In contrast , the energies of the torsional and extensional spring which comprise the cross-bridge are quite different . The energy of the torsional springs is far less than that of the extensional springs at all but the smallest sarcomere lengths . Thus the majority of the elastic strain felt by the cross-bridge may arise from stretching , rather than rotation . Energy stored in the cross-bridges follows the radial force produced by the system ( Figures 3&4A ) . Radial force has a higher correlation with the cross-bridges'energy than does axial force ( linear fits show respective values of 0 . 97 and 0 . 69 ) . All of these relationships are significant ( ) . This suggests that radial strain in the system may disproportionately determine the energy stored in the cross-bridges or , put another way , radial deformation may be acting as a “idden”energy sink . As the sarcomere shortens below , the ratio of energy stored in the sarcomere to energy consumed through ATP use is elevated ( Figure 5 ) . This fraction of energy stored , or energy retention efficiency , is constant at sarcomere lengths longer than . The hydrolysis of ATP to ADP alters the free energy of a modeled cross-bridge by 8 . 8 RT ( ) [15] , [26] . A fraction of this energy is stored as continued deformation of the cross-bridge in its new state and a fraction drives deformation of the filaments experiencing cross-bridge forces . This energy is entirely released on the detachment of the cross-bridge and may partially appear as deformations induced in other bound cross-bridges . At sarcomere lengths longer than the ratio of the energy stored by the sarcomere to the power consumed by the sarcomere ( as measured by the rate of ATP consumption ) is constant ( Figure 5 ) . At shorter sarcomere lengths this ratio climbs: more of the input energy is stored in the sarcomere .
The elements of the sacomere's contractile lattice , cross-bridges as well as thick and filaments , are storing a substantial amount of energy . At peak energy levels where the 16% of bound cross-bridges in our model store of energy , each bound cross-bridge is storing approximately 20% of the work it is capable of producing across a power stroke [34] . The peak cross-bridge energy levels we see in of our half-sarcomere model is 0 . 084 J/kg of stored energy when the lattice is assumed to have the density of water [31] . This is more than 10% of the flight-permitting energy stored in Hemipteran flight muscle [35] . This strain energy is primarily stored in the cross-bridges—rather than in the thick and thin filaments—despite the turnover and energy dissipation inherent in the our model of cross-bridge kinetics [15] . Energy storage in the cross-bridges requires low cross-bridge turnover , as the deformation of an individual cross-bridge , and thus the energy in an individual cross-bridge , is released upon detachment . A “locked lattice” of tightly bound cross-bridges is likely to be present in a maximally activated isometric contraction , as simulated here , and where external factors such as temperature differentials reduce cross-bridge turnover [8] . Energy stored in the muscle's filaments and cross-bridges is then available for later release and utilization . Radial force' role in muscle remains unclear . Radial force may simply be a byproduct of the motor and filament geometry which has evolved to generate force or it may produce a useful effect . The high correlation between radial force and strain energy stored in the cross-bridges may indicate that radial force and distortion act as an energy storage mechanism which permits the cross-bridges to store more strain based energy than the thick and thin filaments . It is possible that radially associated energy could then be redirected to produce axial force , much as happens when energy is stored in the deformation of elastic solids . Such a mechanism would provide a means to store the energy powering after-stretch transient shortening , the shortening of recently stretched muscle against a load equal to its maximal isometric force [36] . Transient shortening has been suggested to be a result of energy elastically stored in cross-bridges at levels comparable to those seen in Figure 4 [34] . However , radial strain based energy storage will not necessarily register as force at the filament ends . As such , it may be difficult to address in experiments , although radial stiffness observations suggest a means by which such tension and energy storage could be quantified [22] . The variable energy retention efficiencies shown in Figure 5 represent a potential mechanism by which sarcomere parameters can determine a muscle' functional role , e . g . motor , brake , or spring . A muscle that stores little of its consumed energy and converts most into force acts as a motor , while a muscle that stores more of its consumed energy for later release is acting , at least prior to use of stored energy , as a spring . There are analogous selections between roles in lengthening and shortening muscle , as contrasted to the isometric conditions simulated here [4] . The non-constant storage of input energy means that changing the degree of filament overlap or lattice spacing affects the amount of energy stored in muscle and thus the amount of energy which may be released to power contraction . Thus it is possible that operating at different sarcomere lengths could change the efficiency with which muscle retains energy or dissipates it , driving the muscle towards functioning as a spring or a break . Energy storage in muscle is still a relatively unexplored field . The highly structured and three-dimensional nature of muscle makes it likely that , historically , we have overlooked forms of energy storage and efficiency regulation . This work points towards cross-bridges as a site where energy can be stored , either in preparation for use in rapid movements or to reduce the energy requirements of cyclical movements . Further investigation of how energy is partitioned between sub-sarcomeric structures and of how radial force is generated will continue to expand our understanding of these new mechanisms . Particularly , this work may help us to understand cases where energy is stored as deformations in an axis orthogonal to that of the direction of muscle shortening , such as in a proposed mechanism by which the heart stores elastic strain introduced into transverse fibers during filling [37] .
Four thick and eight thin filaments are arranged in an evenly spaced hexagonal lattice with toroidal boundary conditions . As shown in Figure S1 , this filament arrangement simulates an infinite lattice [26] . The filament numbers and arrangement of boundary conditions provides the smallest system in which no single thick filament connects to two sides of a single thin filament , and vice versa . The distance between the faces of these filaments , here referred to as the lattice spacing , is uniform and used to provide the distance across which myosin must diffuse in order to bind [15] . Lattice spacing changes with sarcomere length to maintain a constant lattice volume . Thus lattice spacing separating the faces of adjacent thick and thin filaments ( ) is given from sarcomere length ( ) by where is a proportionality constant . This proportionality constant ( ) is chosen to set the lattice spacing to 14 nm at a sarcomere length of , values consistent with a wide range of muscle types [31] . Along each thick filament are 60 myosin crowns , with three myosin heads per crown . The myosin heads on a given crown are azimuthally rotated by 120 degrees from their neighbors . The crowns are grouped into a three crown , 43 nm repeating pattern [26] . Progressing axially through the pattern , each crown is azimuthally rotated relative to the prior crown by . This rotation pattern is measured and described in Al-Khayat et al . , 2008 [38] . As a result of this crown rotation every myosin head faces an opposite a thin filament with which it may interact . Each thin filament is made up of two actin strands . Each strand hosts 45 actin binding-sites giving a whole filament 90 actin binding-sites [25] , [26] , [39] . Each binding site faces and interacts with one of three adjacent thick filaments . Consecutive binding sites on each strand are rotated by clockwise . The first binding site of one strand is offset from the first binding site of the other strand by half the axial distance between adjacent binding sites and a rotation of counter-clockwise . Further filament properties are listed in Table S1 . Our cross-bridges are comprised of one torsional spring and one extensional spring [15] . The axial and radial location of each myosin tip determines the angle and extension of the cross-bridge springs and thus the force the cross-bridge generates . The torsional spring simulates the power stroke via a change in rest angle . Inefficiency in converting , through ATP hydrolysis , chemical to mechanical energy during state transitions is accounted for as distortions of the cross-bridge . This inefficiency is manifest as heat . Additionally as we suggest below , mechanical strain energy which drives motion may also be returned as recoil of cross-bridges or filament backbones . The binding of an individual myosin head is determined by the distance to the nearest available binding site and energy landscape created by the properties of the head' constituent springs . The process is one of perturbation , distance calculation , and stochastic attachment . A myosin head is perturbed with a random Boltzmann distributed energy , providing a new myosin tip location [40] . Distance from the myosin tip to the nearest available binding site is calculated . Binding probability , which falls off exponentially as the distance to the binding site increases , is checked against a random number . Further transitions between loosely attached , force generating , and unattached states are determined as described in Williams et al . , 2010 [15] . The thick and thin filaments are coupled together by the cross-bridges . Each bound cross-bridge both generates and transmits force . This coupling yields a three dimensional network of springs . We solve for the root location of our spring-network at each time-step . The root is the set of locations of actin binding-sites and myosin crowns that provides no net axial force at any internal point in the spring-network . A modified form of the Powell hybrid method allows the actin and myosin locations to iteratively settle into their solution values [41] . At each time-step , actin and myosin locations are allowed to settle in the axial dimension while being held rigidly in the radial dimension . The total axial force ( ) of the system thus comes as the sum of unbalanced axial forces at the ends of each thick filament , where for a given thick filament , , is the filament stiffness , is the axial location of the end node location , is the axial location of the adjacent node , and is the resting separation between the two . The total radial force ( ) is the sum of the radial force experienced by all sides of the thick filament and thus ultimately the sum of the radial force of each cross-bridge ( ) , where , , and are the stiffness , length , and rest length of cross-bridge 's extensional spring while , , and are the stiffness , angle , and rest angle of cross-bridge 's torsional spring . The current model does not permit radial movement as it does not include a resistive radial force , i . e . forces in the radial direction that act against the radial deformation of the filament . Future models may treat the filaments as radially-deformable axially-tensioned beams subject to filament persistence length , electrostatic effects , and viscous stresses and thus be able to permit radial movement . Radial bending or deformation of the thick and thin filaments could potentially reduce the level of radial force within the lattice by increasing lattice spacing disorder . Reduced radial force has the potential to affect the partitioning of energy between the cross-bridges and filaments , shifting energy stored in cross-bridge deformation to a newly-created radial deformation component of the thick and thin filaments' energies . Radial bending of the filaments and subsequent changes in the distribution of axial and radial forces will be resisted by the highly constrained nature of the sarcomere lattice as well as the inherent stiffnesses of the filaments themselves . The energy in a cross-bridge or filament is the sum of the energy in every spring in that cross-bridge or filament . Thus the energy of a single cross-bridge ( ) is calculated , as in prior work [15] . The energy of a thick filament ( ) with crown locations is calculated asand the energy of a thin filament is calculated similarly . These energies are logged throughout the run of an instance of the model and their stable state is found at the conclusion of the simulation . A simulated contraction follows the course described in the diagram shown in Figure S2 . Briefly , each 1 ms time-step consists of allowing every myosin head to calculate the probability of changing from its current state into another state , check this probability against a random number , and transition or not based on the outcome . After the state of each myosin head has been established , the location of every interior point in the model is allowed to settle so that there is no net axial force on them . The axial force , radial force , and other properties of the system at that time-step are then recorded and a new time-step is begun if the contraction has not yet reached its end . The model was allowed to complete 10 contractions ( starting from unbound cross-bridges ) for every set of input parameters , each continuing for 400 ms ( 400 time-steps at 1 ms resolution ) . The asymptotically developed forces and energies were calculated as the mean of the force produced over the last 50 ms . These simulations took place on a dynamically created cluster of spot-priced machine instances in Amazon's EC2 service ( Figure S3 ) . Control of this cluster was with a first-in-first-out command queue hosted by Amazon's SQS . | Locomotion requires energy . Very fast locomotion requires a larger amount of energy than muscle can produce in such a short time period , thus muscle must use energy that it previously produced and stored as elastic deformation . Cyclical or repeated movements can be directly powered by muscle , but energy may be conserved in such cases through elastic energy storage . Traditionally we've looked primarily at tendons , insect exoskeletons , and bones as locations where this energy is stored . However , a small but growing body of literature has recently suggested the backbone filament proteins in muscle act as elastic storage locations . We suggest that the myosin motors themselves are capable of storing more energy than the filaments , energy that may be released to power very fast movements or reduce the cost of cyclical movements . We further suggest that this energy is stored in the radial deformations of myosin motors , in a direction that is perpendicular to the axis of muscle shortening . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"musculoskeletal",
"system",
"biophysic",
"al",
"simulations",
"physiology",
"muscle",
"functions",
"biology",
"anatomy",
"and",
"physiology",
"computational",
"biology",
"biophysics",
"biomechanics",
"muscle"
] | 2012 | Elastic Energy Storage and Radial Forces in the Myofilament Lattice Depend on Sarcomere Length |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.